The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Federal Reserve Staff Study- Volume I
New Monetary Control Procedures
Board of Governors of the Federal Reserve System
February 1981
Federal Reserve Staff Study
New Monetary Control Procedures
VOLUME I
Stephen Axilrod
Overview of Fmdmgs and Evaluation
Richard Davis
Monetary Aggregates and the Use of"Intermediate Targets" m Monetary Pohcy
Jared Enzler
Economic Disturbances and Monetary Pohcy Responses
Jared Enzler and Lewis Johnson
Cycles Resultmg from Money Stock Targetmg
Margaret Greene
The New Approach to Monetary Pohcy-A View From the Foreign Exchange Tradmg
Desk
Dana Johnson and Others
Interest Rate Vanabihty Under the New Operatmg Procedures and the Imtial Response m
Fmanc1al Markets
Peter Keir
Impact of Discount Pohcy Procedures on the Effectiveness of Reserve Targetmg
Fred Levm and Paul Meek
Implementmg the New Procedures The View From the Tradmg Desk
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Federal Reserve Staff Study
of the New Monetary Control Procedure·
Overview of Findings and Evaluation
by
Stephen H. Axilrod
February 1981
(Appeared originally as appendix to Monetary Policy Report to
Congress by the Board of Governors of the Federal Reserve System
dated February 25, 1981)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-Al-
February 1981
APPENDIX
Staff Study of the New Monetary Control Procedure:
Overview of Findings and Evaluation
This paper reviews experience with the new monetary control
procedure established in October 1979 and evaluates implications for
current and alternative control techniques.
The new procedure involved
employing reserve aggregates--on a day-to-day basis, nonborrowed reserves-as operating tools for achieving control of the money supply.
Less
emphasis was thereby placed on confining short-term fluctuations in the
federal funds rate--the overnight market rate reflecting the demand for
and supply of bank reserves.
The change in procedure, it should be
pointed out, represented a technical innovation rather than a change in
the broader objectives of monetary policy or in the monetary targets
themselves.
Target ranges for various measures of the money supply,
together with the actual behavior of money in the course of 1980, are
shown in the charts on the next three pages. The paper is divided into three sections.
Section I presents
-
an overview of findings about effects of the new monetary control
procedure on economic and financial behavior based on evidence gathered
in staff papers. 1 /
Because the new control procedure was designed to
strengthen the System's ability to control the money supply, section II
(page AlS) provides certain additional background analysis relevant to
assessment of the role of money as an intermediate target for monetary
policy.
Section III (page A21) then contains an evaluation of the current
operating procedure, and alternatives.
1/
A list of staff papers prepared appears on page A33.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Growth Ranges and Actual Monetary Growth
M-1A
B1lhons of dollars
400
- - Range adopted by FOMC for
1979 04 to 1980 04
~❖-9;,;:r
Range adJusted for unexpected shifts
-« ""~ into ATS and related accounts*
390
380
Rate of Growth
370
197Q 04 to 1980 04
5 O Percent
0
N
1979
D
J
F
M
A
M
J
J
A
S
0
N
D
1980
* The shaded hnes reflect ad1ustments that should be made for technical
reasons to the original range for M-1 A to allow for unanticipated shifts of ex1st1ng
deposits from demand deposits to interest-bearing transactions accounts, such
as ATS (automatic transfer savings) and related accounts At the beginning of
1 980 1t appeared that such shifts would have Just a hm1ted effect on growth of
M-1 A, and the longer-run growth range for M-1 A was set only ½ percentage
point below the growth range for M-1 B Passage of the Monetary Control Act
subsequently altered the financial environment by making permanent the
authority of banks to offer ATS accounts and by perm1tt1ng all 1nst1tut1ons to offer
NOW and s1m1lar accounts beginning 1n 1981 As the year progressed, banks
, offered ATS accounts more actively and more funds than expected were being
diverted 'to these accounts from demand deposits Such shifts are estimated to
have depressed M-1 A growth over the year 1 980 by ¾ to 1 percentage point
more than had been onginally anticipated The shaded range allows for these unant1c1pated shifts, and therefore in an economic sense more accurately
represents the intentions underlying the ong1nal target
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Growth Ranges and Actual Monetary Growth
M-1B
81llrons of dollars
420
- - Range adopted by FOMC for
1979 04 to 1980 04
l@%: Range adjusted for unexpected shifts
,\'-'- mto ATS and related accounts*
410
400
Rate of Growth
390
1971=( Q4 to 1980 Q4
7 3 Percent
0
N
1979
D
J
F
M
A
M
J
J
A
S
0
N
D
1980
* The shaded lines reflect adjustments that should be made for technical
reasons to the orrgmal range for M-1 B to allow for unant1c1pated shifts mto
mterest-beanng transactions accounts from savings deposits and other
instruments not included m M-1 B At the begInnmg of 1980 1t appeared that
such shifts would have Just a limited effect on growth of M-1 B, and the longer-run
growth range for M-1 B was set only ½ percentage point above the growth
range for M-1 A Passage of the Monetary Control Act subsequently altered
the financial environment by making permanent the authority of banks to offer
ATS accounts and by permitting all mstrtutIons to offer NOW and s1m1lar
accounts begmnmg m 1981 As the year progressed, banks offered ATS
accounts more actively and more funds than expected were being diverted to
the accounts Such shifts are estimated to have increased M-1 B growth over the
year 1980 by ½ to ¾ of a percentage point more than had been ant1c1pated
The shaded range allows for these unant1cIpated shifts, and therefore In an
economic sense more accurately represents the IntentIons underlying
the ong1nal target
Growth Ranges and Actual Monetary and Bank Credit Growth
M-2
BIiiions of dollars
1700
-
Range adopted by FOMC for
1979 04 to 1980 04
Rate of Growth
1979 04 to 1980 04
9 8 Percent
1600
1550
J
FM
AM
1979
J
J
AS
0
N
D
1980
M-3
BIiiions of dollars
2000
Rate of Growth
1979 04 to1980 04
9 9 Percent
1900
1800
0
N
D
J
FM
AM
1979
J
JASON
D
1980
Commercial Bank Credit
Billions of dollars
1280
Rate of Growth
9%
1979 04 to 1980 04
7 9 Percent
1220
1160
0
N
1979
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
D
J
F
M
A
M
J
J
1980
A
S
O
N
D
-A2I.
Overview of Findings with Regard to Experience
since Adoption of New Procedure
Questions investigated in reviewing experience with the new
control procedure included, among others, its impact on precision of
money control, volatility of interest rates, the course of economic
activity, and exchange market conditions.
There were, of course, other
influences on financial markets and the broader economy that were surely
of far more importance than the particular technical innovations under
consideration here.
Indeed, a major problem has been to distinguish the
impacts of the new procedure per'~ from larger influences operating on
the economy.
This difficulty is particularly acute given the relatively
short period of time since the new procedure was implemented--a period
of time that may have been too short for market participants to have
fully adjusted to the new environment and a period of time in which
markets were buffeted by changing inflationary expectations, fiscal
uncertainties, credit controls, and oil price shocks.
A.
Relation between reserves and money
1.
Over the operating periods between FOMC meetings, actual nonbor-
rowed reserves fell below the Trading Desk's operating target by about
0.1 of 1 percent on average; the average absolute miss was about 0.4 of
1 percent.
These deviations reflected in part errors in projection of
uncontrollable factors affecting reserves (such as float).
In addition,
the Desk at times accommodated to variations relative to expectations in
banks' demand for borrowing in the course of a bank statement week (for
example, an unexpected willingness by banks to obtain reserves by borrowing heavily over a weekend).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Total reserves came out somewhat above
-A3intermeeting period paths, by ahout 0.2 of a percent on average; the
absolute miss averaged about 0.8 of a percent.
The individual intermeet-
ing period misses reflected deviation of money stock from short-run
targets, variations in excess reserves, and multiplier adJustments to
the original path (to take account of changes in required reserves for a
given level of deposits) that turned out to be incomplete.
2.
Econometric evidence from simulations of monthly money market
models carried out with various reserve measures as operating targets
(nonborrowed and total reserves and the monetary base), given the
existing institutional framework, buttresses indications from actual
experience last year that the relationship between reserves and money
is relatively loose in the short run.
Over the one-year period since
October 1979, the mean absolute error of misses in the level of M-lB
relative to target path during the 4- to 7-week operating periods between
FOMC meetings was a little over 0.6 of a percent.
This degree of
variability was in line with--in some cases less than and in some cases
more than--model simulation results (holding various reserve measures at
predetermined target levels for the simulations) . .!./
In comparing the
models and the reserve technique actually used, it should also be
observed that model simulations generally implied more interest rate
variability last year than proved to be the product of the technique
actually in use.
1/
The root mean square errors of actual misses and simulated model
misses ranged around .7 to .8 of a percent over short-run operating
periods of a month or so. This would mean that, with disturbances
similar to last year's, two-thirds of the time M-lB would generally
come within plus or minus .7 to .8 of a percent of the intermeeting
target path over approximately a one-month period (or, expressed in
annual rate terms, within a range of plus or minus 8 to 10 percentage
points over such a period).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A43.
In the model simulations of the past year, control of money supply
through strict adherence to a total reserves or total monetary base
target produced more slippage than control through the nonborrowed counterparts of each.
This phenomenon largely reflects the presence of multiplier
disturbances on the qupply side that would be generated, for example, in
the current institutional environment by changes in deposit mix and hence
in required reserves for any given level of money supply.
In the model
simulations, use of total reserves or the total base as an invariant target
over the control period does not permit these disturbances to be cushioned
by changes in borrowings.
4.- Judgmental predictions of the multiplier relationship between
reserves or hase measures and money made since the shift in operating
procedure were generally superior to, though on a few tests not
significantly different from, forecasts derived from econometric-models.
5.
Over a longer period than a month (or than an intermeeting period)
errors in the predicted relationship between money and reserves may be
expected to average out--that is, over time, errors in one direction tend
to be offset by errors in the other.
Simulations of the Board's monchly
model suggest that such a process is at work.
In actual operations over
a one-year period since October 1979, the absolute miss in the level of
M-lB when individual misses relative to the short-run target paths are
averaged over three or four intermeeting periods was reduced from a
little over 0.6 of a percent (reported above) to over 0.4 of a
percent.
This represents a somewhat s111aller reduction than would have
been expected from certain results, and may have reflected the nature of
unusually large, unanticipated successive month-to-month changes in money
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A5demand last year, first in one direction and then in the other.
These
changes were related in part to identifiable special factors such
as the imposition and subsequent removal of the credit control program.
Accommodation to such special and temporary factors, as they emerged,
might tend to lengthen the period over which deviations from monetary
targets could be expected to average out, but would, by the same token,
tend to dampen fluctuations in interest rates that would not have
contributed to better control of money over time.
B.
Variability in money growth
1.
Evaluation of the variability of money supply series is
importantly affected by the seasonal adJustment process.
Seasonal factors
applied during a current year are unable adequately to reflect changing
seasonal patterns in the course of that year; after a year is over,
therefore, reestimation of seasonal factors often tends to smooth
variability.
Based on current seasonal adjustment factors for the year
just past (that is, factors before seasonal revisions that take account
of the influence of a~tual experience this year), variability in weekly,
monthly, and quarterly growth of M-1 (and also M-2) was substantially
greater than in any year during the past decade.
However, when the
variability in money growth during the year from October 1979 to October
1980 is compared with variability in earlier years--with earlier years
adJusted_,using seasonal factors that were current in those years--nearly
all of the heightened variability in weekly growth of M-1 and a sizable
portion of the monthly and quarterly variability are removed.
While
this comparison makes it seem prohable that seasonal factor distortions
are overstating variability in the year just past, the extent cannot be
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A6assessed with confidence until a number of years have passed.
In
general, it would appear that money has been more variable over the past
year, especially on a monthly and quarterly basis--though so far as can
be judged from the available data, still generally well within the range
of foreign experience with money supply volatility.•
2.
The variability in money growth of the past year appears to be
related to an unusual combination of circumstances:
a.
There were large swings within the year in the demand for
money resulting from sharp short-run variations in economic activity
caused in large part by factors independent of the new monetary control
procedure, such as the imposition and subsequent removal of the credit
control program.
The imposition and subsequent removal of the credit
control program may have also increased the variability of money growth
through a more direct channel, as the·associated large variation in
bank loans was accompanied by temporary changes in demand deposits-for example, as large loan repayments were initially made from existing
demand balances.
b.
In addition, econometric evidence from a variety of models
suggests that there were "unexplained" factors other than economic
activity and interest rates causing substantial fluctuations in money
demand.
In particular, money levels fell considerably short of model
simulations (given GNP and interest rate~) in the second quarter,
when money growth was negative.
Relatively rapid growth in subsequent
quarters reflected in part a tendency for money levels to move back
toward more normal relationships with GNP and interest rates.
3.
T~e money targets on which reserve paths were based reflected the
intention to return money over time to the long-run obJective following
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A7divergence~.
In 1980 the target for narrow money in the month following
the F0MC meeting typically implied making up about 30 percent of the
difference between the projected level of the money stock in the month of
the meeting and the long-run target path.
If disturbances in 1980 had been
more representative of those prevailing in the 1970s, simulations using
•
the Board's monthly model suggest that the reserve operating technique
would have kept money closer on a month-by-month basis last year to longrun objectives than actually was the case.
These simulations also indicate
a distinct trade-off between variability of the federal funds rate--and
money market rates generally--and the speed with which attempts are made
to return the money stock to its longer-term path once it moves off path.
The more rapid the attempted return to,path, the larger are the implied
fluctuations in money market rates.
4.
Interpretation of money supply volatility is complicated by the
large amount of noise in weekly and monthly changes in first published
figures for the narrow monetary aggregates (and for monthly changes in M-2)
resulting from transitory variation and seasonal factor uncertainty.
Based
on data for the 1973-79 period, the estimated standard deviation of the
noise factor for monthly changes in M-lA and M-lB is about $1.5 billion
(4-1/2 percent at an annual rate), and about $3.3 billion for weekly
changes.
~or M-2, the estimated standard deviation of noise in monthly
growth rates is 3-1/2 percent at an annual rate.
The noise factor declines
for growth rates over longer periods of time.
c.
Variability of interest rates
1.
As had been expected, the federal funds rate has been more variable
on an intra-day, intra-weekly, and inter-weekly b~sis since the new procedure was implemented.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Intra-day and day-to-day variability has tended
-A8to be at least twice as large as before, as have weekly changes after
adjusting for trend.
This greater variability of the federal funds rate
reflects the role of nonborrowed reserves as an operating guide for the
Desk.
2.
There has also been heightened variability of interest rates on
Treasury securities of all maturities following adoption of the new
operating procedure.
Based on data from which cyclical movements were
removed, the variability in Treasury yields measured on a weekly average
basis has been at least twice as large as before October 1979.
1.
The relationship over interest rate cycles between the federal
funds rate and yields on Treasury securities of all maturities has been
essentially the same before and after October 1979, suggesting that the
underlying linkage between the federal funds rate and other market rates
has remained about unchanged.
At the same time, however, correlations
between very short-run nonsystematic movements in the funds rate and
other market rates have increased substantially since the new procedure
was implemented.
This higher correlation possibly reflects the sensi-
tivity of market participants to day-to-day changes in the funds rate in
the uncertain environment that prevailed last year but possibly also
reflects concurrent adjustments in market interest rates generally,
particularly short rates, that tend to occur as closer control is sought
over the money supply, given variations in money demand.
D.
Effects on domestic financial markets
The swings in interest rates last year, and the high levels reached,
clearly affected behavior in financial markets.
It is difficult to
isolate the role of the new operating procedure, as such, in contributing
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A9to interest rate swings or changes in market behavior.
It is likely that
large cyclical variations in interest rates would have developed last
year in any event if the basic monetary aggregate targets were pursued by
other operating techniques in the face of cyclical variations in money and
credit demands that were exceptionally large and compressed in time.
And
adjustments that took place in financial market behavior last year largely
represented adaptations that would have been expected on the basis of
past cyclical experience--for example, constraints on housing finance-or were related to the special credit control program.
Market adjustments
that might have primarily reflected adaptations to the new procedure as
such are likely to be those more associated with a perceived greater
continuing risk of short-term interest rate volatility--adjustments
that would be difficult to detect in an environment like that of last
last year, which was dominated by cyclical changes in credit flows, a
credit control program, and inflationary expectations.
1.
Mortgage markets.
Greater interest rate volatility since October
1979 may have hastened the trend in process for a number of years toward
more flexible mortgage instruments, such as variable-rate, renegotiable,
and equity participation mortgages.
In addition, mortgage bankers and
other originators in their commitment policies appear to have attempted
to avoid qome of the risk of interest rate changes occurring between the
time a commitment is made and funds are extended.
They have done so by
setting rates or points at the time of closing, shortening the period
for guaranteed fixed-rate mortgage commitments, and by imposing large
nonrefundable commitment fees to discourage cancellation if rates should
decline.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-AlO2.
Dealer market for Treasury and Agency Securities.
Wider bid-ask
spreads on Treasury bills appear to have emerged last year.
Evidence on
such spreads for coupon issues is difficult to interpret; spreads rose
considerably a few months prior to introduction of the new procedure, and
thereafter remained wider than in earlier years.
Greater uncertainty
about interest rates may have influenced dealers to maintain leaner inventory positions relative to transactions; turnover of dealer inventories
rose last year as a very large expansion in gross transactions outpaced
the rise in the level of inventories.
3.
Underwriting spreads on corporate bonds.
Underwriting spreads on
corporate bonds issued on a negotiated basis did not widen, on balance,
over the year since October 1979.
However, data on competitively bid
issues suggest that spreads on such issues have widened.
This might tend
to raise bond costs, but any such effect last year would appear to have
been very small relative to the more basic supply and demand conditions
affecting markets.
4.
Commercial bank behavior.
Bank behavior last year was strongly
influenced by a number of factors other than the new procedure, such as
the imposition and removal of the special voluntary credit restraint
program, marginal reserve requirements on managed liabilitie9, and
increasing reliance, especially by small banks, on money market certificates as a source of funds.
It is difficult to detect changes in behavior
associated with the new procedure per~•
There appears to have been
some increased reliance on floating-rate loans, especially for term
loans, but this trend was evident prior to October 1979.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-All5.
Futures markets.
Futures market activity expanded rapidly in the
period following October 1979, raising the possibility that the new
procedure led to an increased desire to hedge against expected greater
interest rate fluctuations.
~owever, the expansion in activity represented
a continuation of the trend of recent years, as has been the case with
other market adaptations noted above.
It is virtually impossible to
separate growth in futures activity arising from attempts to reduce exposure
to interest rate risk in the new environment from underlying trend growth
connected with increasing familiarization by the public with the variety
of financial futures instruments that are becoming available.
6.
Liquidity premiums.
An attempt was made to detemine whether
there was an increase last year in liquidity premiums, manifested by a
rise in long-term rates relative to short-term rates.
Such a result
might be expected if risk-averse financial market participants attempted
to protect themselves from a perceived risk that the new procedure would
make for greater interest rate variability and hence greater risk of capital loss on holdings of longer-term issues.
There appears to be little,
if any, evidence that liquidity premiums became greater last year--although
as noted in paragraphs 2 and 3 above there may have been some increase of
transactions costs in financial markets.
E.
Exchange market and other external impactg
1.
The spot value of the dollar appreciated by more than 5 percent
in the 14-month period subsequent to late September 1979, though there
were pronounced cycles that coincided with intermediate-term movements of
interest rates in the United States.
2.
Day-to-day movement in money market rates related to the new
procedure could have had some influence on very short-term exchange rate
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A12~
volatility.
Spot rates have displayed more variability on a daily basis
since the new procedure was adopted, reflecting greater daily variability
of interest rate differentials between U.S. dollar and foreign currency
assets.
The evidence on weekly and monthly exchange rate movements also
suggests more variability, but the evidence is not so conclusive as that
for daily variability.
3.
There _is little evidence of a significant increase in the
variability of foreign interest rates, apart from in Canada~ on a monthly
basis related to the new procedure as such.
Some countries, especially
developing countries with currencies tied to the dollar and with inflexible interest rate structures, appear to have experienced some tech~ical
difficulties over this period connected, for example, with the impact of
interest rate variability-on financial flows.
4.
The evidence does --not suggest that the new operating procedure
has contributed to the variable nature of gross U.S. international capital
flows since the fall of 1979.
Significantly greater contributing factors
were the credit control program and marginal reserve requirements on
managed liabilities.
5.
The proposition that more short-term variability of exchange
rates could have adverse effects on the domestic price level, because
price increases caused by currency depreciation would not be fully offset
by the reverse effect of currency appreciation, is not supported by
econometric evidence.
Therefore, the short-term variability of exchange
rates since October 1979 would not itself appear to have raised the
domestic price level.
Meanwhile, the underlying trend toward appreciation
since that time would have had a favorable effect on the price level.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A13F.
Economic activity
1.
Assessing the contribution of the new procedure as such to the
pattern of economlc activity and inflationary expectations is complicated-as noted-at other points in this paper--by the force of other factors
that were importantly influencing the markets for goods and services over
the recent period, including the effect of the basic money supply targets
themselves.
Certain "fundamentals"--such as the previous sharp increase
in oil prices, the relatively low saving rate, and the illiquid balance
sheet of the household sector-suggest that economic activity would have
contracted in any event in 1980.
In addition, prices and real economic
activity were strongly influenced by the highly sensitive state of
inflationary psychology, the imposition and removal of the credit control
program that lasted from mid-March to early July 1980, and erosion of
fiscal restraint.
2.
Nevertheless, to the extent that the new control procedure
encouraged more prompt interest rate adjustments in response to cyclical
fluctuations in money and credit demands, it probably exerted some
influence on the pattern of economic activity.
It may have hastened the
slowdown in economic activity-especially in housing and possibly consumer
durables--in early 1980 and also hastened the recovery in the summer, as
interest rate~ advanced rapidly to peak levels and then contracted sharply.
Psychological reactions to the credit control program, however, may have
been an important influence on the depth of the recession and the promptness
and strength of the subsequent rebound.
There was a sharp contraction in
spending following introduction of the program, and relief on the part of
both financial institutions and borrowers as the program was phased out
probably encouraged a sizable resurgence of spending.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-Al43.
In view of the lags in the response of capital spending plans to
changes in credit conditions, the new procedure does not appear to have
exerted much influence on plant and equipment spending during the past
year.
The timing of inventory movements, by contrast, may have been
altered to the extent that the new procedure had effects on the pattern
of final sales and on movements in short-term financing costs.
4.
The new control procedure was adopted in part to provide more
assurance that inflation would come under control (as money growth was
restrained), and thereby to reduce inflationary expectations.
It is
difficult to measure inflationary expectations, let alone to attribute
changes to a technical change in monetary control procedures in so highly
unsettled a period as last year.
Indirect evidence about inflation
expectations based on changes in interest rates is obviously difficult to
interpret, since interest rates are also influenced by other factors.
Some direct evidence about consumer expectations of inflation can be
gleaned from the Michigan survey.
No clear improvement in inflationary
attitudes is evident until into the spring, probably related in large
part to the sharp contraction of economic activity in the qecond quarter.
There did not appear to be any significant worsening of expectations, as
judged by the Michigan survey, in the latter part of the year as the
economy strengthened.
5.
The Board's large-scale quarterly econometric model, as well as
two other much more simplified models used for comparative purposes, were
employed to help evaluate the extent to which the actual fluctuations in
money and interest rates affected economic activity in the course of the
year.
These models, of course, all suffer from an inability to take
account adequately of attitudinal changes and other behavioral factors
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-Al5related to the special conditions of a particular year, including any
attitudinal changes that might be occasioned by the shift in operating
procedure.
Simulation results suggest that, because of long response
lags, the pattern of economic activity last year would not have been'
particularly sensitive to efforts at smoothing the quarter-to-quarter
pattern either of money growth or of interest rate variations~ though
smoothing money growth had slightly more impact.
The smoothing of
money growth would have been at the cost of even greater interest rate
variability than was actually observed over the last five quarters.·
II.
General Considerations
Evaluation of the current and alternative operating techniques
to be discussed in section III depends very much on the role accorded
intermediate targets, particularly the monetary aggregates, in the
formulation of monetary policy.
This section examines advantages and
disadvantages involved in employing monetary aggregates, or for that
matter interest rates, as intermediate targets, and also examines'certain
limitations on the feasible range of target settings.
A.
Advantages and disadvantages of monetary aggregates as intermediate
targets
1.
Advantages
a.
Money stock control tends to work toward stabilizing GNP when
the economy is buffeted by disturbances to spending on goods and
services and shifts in inflation expectations; such factors appeared
to be an important influence on economic and financial behavior last
year.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
If spending surges unexpectedly, for example, as it did in the
-Al6second half of 1980, adherence to a money stock target would automatically lead to tighter financial markets, tending to offset some
of the surge in spending.
Similarly, if spending were to weaken
unexpectedly--and very substantial weakness developed in the second
quarter of last year--efforts to hold to a money stock target would
lead automatically to lower market rates of interest, which would
tend to partially restore spending to desired levels.
b.
Current approaches emphasizing control of monetary aggregates
rest on the proposition that planned deceleration in monetary growth
will lower inflation over time by limiting funds available to
finance price increases and encouraging expectations and behavioral
patterns consistent with reduced inflation.
c.
By clearly communicating to the public the Federal Reserve's
obJectives for monetary policy, a monetary aggregates targeting
procedure enables private decision-makers to better plan their
activities and to make wage and price decisions that are more
hamonious with noninflationary growth in money and credit.
d.
Targeting on monetary aggregates involves adjustments of
market interest rates, in response to underlying changes in demands
for credit, that might otherwise be unduly delayed, either on the
down- or up-side.
2.
Disadvantages
a.
Looseness in the relationship between money demand and
nominal GNP reduces the significance of monetary aggregates as a
target, particularly in the short run.
Unexpected shifts in this
relationship lead to undesirable interest rate movements with strict
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-Al7adherence to money supply targets.
Last year, there was evidence
of looseness in this relationship.
For example, as noted earlier,
econometric models suggest a sizable downward shift in the demand
for money in the second quarter, given actual GNP and interest
rates.
b.
Attempts to achieve steady growth in monetary aggregates
on a month-by-month or even quarter-by-quarter basis can lead to
large interest rate fiuctuations, given the high degree of
variability in short-run money flows and the relatively interestinelastic demand for money over the near term.
Large fluctuations
in interest rates have certain risks; for instance, they might endanger financial institutions that are unable to make timely compensating adjustments in their balance sheets, adversely affect the
functions of securities and exchange markets, and lead to confusion
about the basic thrust of policy.
c.
Money supply targeting procedures might themselves introduce
recurrent cyclical responses of economic activity following an
economic disturbance.
Whether this is a realistic risk depends on
the nature of response functions in the economy.
It would be a high
risk in the degree that: (1) money demand was very insensitive to
interest rate changes (and thus interest rates would need to change
sharply to maintain steady money growth in response to an exogenous
disturbance fron the goods market), and (2) there was no significant
current impact on spending from such changes in rates but impacts
were felt over later periods.
It would be difficult to attribute
the cyclical behavior of economic activity over the past year to
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-Al8-
such a process, though, given model estimates of the interestelasticlty of money demand and of relatively long lags between
interest rates and spending (with such lags implying a longer cycle
than observed last year).
d.
The concept of money is elusive, and is becoming more so
as new substitutes evolve for traditional transactions media and as
improvements in financial technology facilitate the ability of
the public to shift funds about for payments purposes.
B.
Interest rates as targets
1.
Advantages
a.
Control over total spending can be strengthened by greater
emphasis on stabilizing interest rates when disturbances stem
mainly from the monetary sector rather than from markets for goods
and services.
b.
Control over rates might make for greater short-run stability
in ~inancial markets, since market institutions might be relatively
certain about the terms and conditions under which they can "safely"
meet near-term credit demands.
2.
Disadvantages
a.
It is very difficult to determine the appropriate interest
rate level, particularly in an inflationary environment in which
shifting expectations of inflation are continuously altering the
relationship between real and nominal market rates of interest.
b.
Efforts to stabilize interest rates tend to amplify economic
cycles ~temming from cyclical variations in the demand for goods
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-Al9and services, since by stabilizing rates, pro-cyclical growth in
money and credit would be heightened.
An upswing in the demand for
goods and qervices, for example, would be accompanied by an expansion
in the volume of money and credit.
By contrast, with a money stock
targeting procedure, resistance would be introduced automatically
through increases in interest rates ..!/
c.
While interest rate targets could in concept be adjusted
promptly so as to minimize the likelihood of a pro-cyclical monetary
policy, in practice the institutional decision-making procedure often
limits the ability to make sizable adjustments in the target.
This
could constrain interest rate variations when rates are taken as the
intermediate target of monetary policy.
c.
Limitations in the targeting process
Regardless of whether monetary aggregates or interest rates are
selected as intermediate targets, there appear to be a number of limitations
on the monetary authority's range of choice of the particular target
setting and ,the precision with which the target is pursued.
1.
The particular target setting must take into account the capacity
of the economy and financial markets to adjust to the targets, and the
degree to which the implications of those targets can be understood by and
are acceptable to the larger public whose behavior patterns are involved.
Inflexibilities in wage and price determination, for example, have implications for the degree to which monetary targets can be reduced, without
risking unduly adverse implications for economic activity in the short
1/
Even with a money stock procedure such resistance may not be sufficient
to hold nominal GNP down to a previously desired level if the upward
shock in demand for goods and services involves a rise in velocity--as
it well might if it resulted from, say, expansion in federal spending.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A20run.
This would be less of a limitation to the extent that attitudinal
shifts--either in response to announced monetary targets or other/factors-brought upward wage and price pressures down in line with monetary targets.
Experience of the past year has not yet provided a basis for believing
-
J
that the lengthy lags between money growth and price changes have been
shortened significantly or that inflation expectations have begun to
respond more rapidly to the money control procedure per~2.
The question may arise as to whether disturbances in domestic, or
foreign exchange, markets may on occasion require short-run departures
from intermediate-term targets of monetary policy.
However, these markets
appear to have adjusted to a subgtantial degree of interest rate or
exchange rate fluctuation during the past year.
3.
Precise month-by-month control of money does not seem possible,
given existing behavior patterns in the economy and financial markets and
institutional factors.
Nor is there evidence that such close control is
needed to attain the underlying economic objective of encouraging noninflationary economic growth.
Statistical investigation suggests that
"noise" alone accounts for substantial variation in monthly money growth
rates.
Moreover, model simulations indicate that variations in money
growth above or below targets lasting a quarter or so are not likely to
have substantial economic effects.
4.
Uncertainties involving the relationship between money demand and
GNP--as evidenced by unexpected variations in such demand last yearsuggest the need for a degree of flexibility in target setting (ranges
may be preferable to point estimates), and also suggest the possibility
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A21that, at times, ,there may be a need for large deviations from predetermined targets or for changes in the targets.
On the other hand,
deviations from target ranges involve the risk of changes in market
expectations that are counter-productive (for example, when money supply
runs strong relative' to target, inflationary expectations_may be heightened, compounding the difficulties of controlling inflation).
In general,
though, in the degree that there is success in achieving targets over
time, expectations are less likely to be adversely affected by short-run
deviations in money growth.
III.
Evaluation of Operating Procedures
Because the past year was in many ways exceptional--and because
a year, or 15 months, in any event is too short a time frame within which
to judge whether observed relationships are accidental to the period or are
lasting--evaluation of the new control procedure, and of possible alternatives, must at best be quite tentative.
The choice of operating proce-
dure would be influenced by the predictability of certain financial and
economic relationships and by the capacity of markets to adjust to operating techniques without severe distortions--evidence about which was
presented in section I.
In addition, the desirability of retaining the
present reserve procedure (with or without possible modifications), of
shifting to an alternative reserve
procedure, or indeed of shifting back
I
entirely to a federal funds rate operating guide depends in part on the
value to be placed on relatively tight short-run control of money, given
uncertainties about the likely sources of potential disturbances in
economic and financial conditions.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A22If there were complete certainty about economic relationships,
the choice of operating procedure would not be particularly critical, for
a given money stock target would be associated with unique, known values
for the federal funds rate, nonborrowed reserves, and the monetary base.
And the monetary authority could achieve its objectives no matter which of
these instruments was selected for operating purposes.
In practice, however, markets are continually subject to disturbances that are not knownl n advance.
The principal kinds of disturbances
are those occurring in overall spending (the market for goods and services),
those occurring in the demand for money (independently of GNP and intereqt
rates), and those affecting the supply schedule for money (such as deposit
mix or banks' demand for excess reserves).
Moreover, such disturbances-
all of which were evident last year--can be of a temporary or self-reversing
variety, or they can be permanent.
Alternative operating procedures tend to produce different outcomes for the pattlrn of interest rates and money growth in the face of
these disturbances.
With some procedures, and depending on the source of
the disturbance, interest rates would be changed more, while with others
the money stock and other financial quantities would absorb more of the
impact.
The choice of operating procedure therefore involves, among
other things, judgments about whether there is more risk to monetary
policy's ultimate objective of noninflationary growth from procedures
that·'tend -to emphasize interest ,rates ... as,, oper.atinghtarge,ts.,-wit;,h.-som,e~,,~.
implication of a relatively gradual change in rates, or from those that
tend to work more directly against money supply variations.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A23A.
Assessment of present operating procedure
The present reserve operating procedure proved flexible enough
to permit some accommodation in the short run to unexpected shifts in
money ~emand, given GNP and interest rates, that occurred last year.
At
the same time, the procedure worked to limit the extent to which changes
in demands for goods and services (and thus in transaction demands for
money) were reflected in actual money growth.
Actual money growth devi-
foted from short-run targets last year, but there were large accompanying
changes in interest rates that tended, over time, to set up forces bringing money back toward path.
Nonetheless, money growth over time deviated
more from path than might have been expected relative to the average
degree of looseness that seems to exist in reserve-to-money relationships.
While the experience of last year may have been atypical because
of the nature of disturbances during the year, still a number of modifications to the operating proced11re used since October 1979 might be
considered for their potential value in reducing slippage in money relative
to reserve paths.
These modifications all have certain disadvantages,
however, that need to be weighed aginst their varying advantages for more
precise monetary control, to the degree that closer control in the shortrun is considered desirable.
1.
Evidence of the past year suggests that during an intermeeting
period relatively prompt downward (or upward) adjustments in the original
'''"'~'
nonborrowed reserve path may be.needed, tn an effort,t~
8[f~~~~
~Y~! t!m~,.
increased (or decreased) demand for borrowing when money is strengthening
(or weakening) relative to ,target.
As an alternative, more prompt upward
(or downward) adjustments in the discount rate would tend to discourage
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A24(or encourage) borrowing over time (in practice the actual level of
borrowing will not change until money demand changes sufficiently to
alter reserves demanded to meet reserve requirements).!./
These adjust-
ments run the risk of increasing the volatility of short-run interest
rate movements in view of the transitory fluctuations often experienced
in short-run money demand.
However, they could also dampen the amplitude
of longer-term swings of interest rates by more promptly leading to adjustments by banks that bring money growth back toward path.
2.
More fundamental changes in the administration of the discount
window and in the way discount rates are structured and varied could be
considered for strengthening the relationship between reserves and money.
a.
At an extreme, discount window borrowing might be limited
to emergency needs.
This is tantamount to adhering to a total reserves
or monetary base path.
However, this would eliminate the valuable
buffering function of the discount window.
The window buffers the
money stock (and the markets) from disturbances affecting the supply
of money (such as changing demands for excess reserves and changes
1/
Expe~ience has demonstrated that it is difficult to determine in advance
the appropriate level of borrowing to be employed in constructing the
nonborrowed reserve path consistent wtth the -;hort-run noney supply
target. This level of borrowing would depend on a projection of market
interest rates consistent with the money supply target path and knowledge
of depository i~stitutions' willingness to borrow, given the spread
between market rates and the discount rate, and could differ significantly
from borrowing levels based on or ranging around recent experience. In
attempting to forecast borrowings, evidence from models may be usefully
weighed along with judgmental assessment of particular conditions at
the time. However, in view of considerable uncertainties about interest
rate projections, the high degree of year-to-year variability in the
success with which models proJect economic and financial relationships,
and the heightened variability in denands for discount window credit
evident last year, projections of borrowing demand from interest rate
forecasts and past bank behavior are subject to a considerable degree
of error.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A25in the deposit mix affecting required reserves).
Its role in that
respect was evident from the results of model simulations showing a
weak relationship between total reserves or the monetary base and
money (when reserves or the base are treated as exogenously determined).
In addition, the discount window cushions markets from the
full impact of variations in money demand that may be transitory or
which the FOMC may wish at least partially to accommodate.
Finally,
lagged reserve accounting requires access to the discount window in'
the short run on occasions when required reserves run above the nonborrowed reserve path (if that path is to be maintained).!/
b.
Another approach to consider would be to eliminate administrative
guidelines at the discount window and to substitute a graduated discount
rate schedule for adjustment credit--in contrast to emergency and other
longer-term types of discount window credit--based on, say, size of
borrowing.
This approach would tend to make the relationship between
borrowing and short-term market rates more'certain by eliminating from
the decision to borrow the uncertainties connected with administrative
guidelines.
It also thereby transforms th~ highest discount rate on
the schedule into an upper limit for the federal funds rate.
There
are, however, legal questions about the System's ability to use size
of borrowing as a criterion, administrative problems in overseeing
the adequacy of collateral and the financial condition of a vast
number of potential regular borrowers, and difficult questions with
' '
re~ard to the appropriate gradient for the discount rate schedule.
1/
Even with contemporaneous instead of lagged reserve accounting, it is
by no means clear that banks would be able to make needed adjustments
reducing their required reserves within a statement week--except at the
expense of relatively extreme interest rate movements.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A26Too steep a gradient risks undue market interest rate fluctuations,
particularly at times when borrowing demands may be changing for
transitory reasons, while too flat a gradient-and at the limit a
perfectly flat one--would tend to eliminate the incentive of banks
to make portfolio ad1ustments that would bring money supply
back to targe~.
c.
The recent policy of applying a surcharge above the basic
discount rate for frequent borrowing (by larger banks) represents
a step toward a graduated discount rate structure within the present
administrative guidelines and tends, when applied, to speed up the
response of market rates to overshoots or undershoots of money
relative to path.
This approach has the dttraction of flexibility,
but in practice it has proved difficult to assess because of the
limited experience with it thus far.
d.
Another approach to speeding up the response of banks within
present administrative guidelines would be to tie the discount rate
to market rates, either as a penalty rate or not.
However, this
approach tends to limit flexibility and raises the danger of upward
or downward ratcheting of market rates in the short run that may be
excessive for monetary control needs and unduly disturbing to the
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A27'-
functioning of markets ..!/
While a tied rate accelerates the response
of market rates, the change may be counterproductive-particularly
if money behavior were going to reverse itself naturally or if the
rise in borrowing were needed to moderate shocks from the supply sideand could intensify short-run money supply and interest rate cycles.
3.
A closer snort-run relationship between reserves and money could
be attained by measures that strengthen the link between required reserves
and deposits in the particular money stock that is being controlled.
One
such measure would be a shift from lagged reserve accounting (LRA) to
contemporaneous reserve accounting (CRA), which the Board has already
announced that it is contemplating.
Such a shift would make the link
between current reserves and current deposits stronger, though -there
still would be relatively sizable slippage between reserves and money
from other sources.
The monetary control advantages of CRA apply
particularly to the short run.
They have to be weighed against (1) the
benefits of LRA for reducing the cost of reserve management by the
banks, (2) the contribution of LRA to the Trading Desk's ability to assess
reserve supply conditions, and (3) judgments about the adequacy of
monetary control under LRA over a longer-term period.
1/
This danger is greatest tn the degree that the disco~nt rate is tied
to a current or very recent market rate. If required reserves expand
rapidly in the current week, banks will have to borrow the added required
reserves that are not being accommodated by the nonborrowed reserve
target. As a result market rates must rise to the point at which banks
are willing to borrow from the discount window. With an attempt to
maintain a "penalty" discount rate, the new market rate would therefore
have to move temporarily above the discount rate, which could not be
maintained, in those circumstances, above current market rates. Market
rates would go up by the amount needed to reestablish the normal
spread of market rates over the discount rate (that emerges from
pressures generated by discount window administration and banks'
reluctance to borrow). ~ut this rise in rates may well bring about a
further rise in the discount rate if an attempt is made to reestablish
a "penalty" rate, entailing yet a further rise in market rates, so long
as required reserves remain at an advanced level.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A284.
The present relatively complicated reserve requirement structure,
even apart from LRA, makes for considerable slippage in the relation
between reserves and money.
While the Monetary Control Act has tended to
simplify the required reserve structure, it will be a number of years
before the new structure is fully phased in.
Because of the unpredic-
tability of shifts in deposit mix, in the ratio of currency to deposits,
as well as in banks' demand for excess reserves, judgmental multiplier
adjustments to original paths were made week-by-week last year as new
information was obtained.
Model simulations suggest money-reserve rela-
tionships would have otherwise been more variable on average.
Thus,
there is no reason not to continue making such adjustments, though it
remains unclear, because multiplier changes are so erratic, whether full
adjustment should he made to each week's added infonnation.
5.
It appears from tentative results based on the Board's monthly
money market model that the faster the FOMC attempts to move back toward
the longer-run target for money, once off target, the more likely is the
long-run target to he hit, assuming no federal funds rate constraint.
However, these results al~o suggest that the more quickly a return to path
is sought, the more substantial fluctuations in money market rates are
likely to be.
And experience of the past year suggests these more
substantial fluctuations would be transmitted broadly through the rate
structure.
Moreover, for a more rapid return beyond a certain speed--per-
haps around three months--it seems as if the gain in reducing the chance
of departures from longer-term money targets is small compared with the
increasing chance of a wider range of variability in money market rates.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A29,B.
Assessment of other targeting procedures
1.
Monetary base or total reserves
A.
The principal reason for adopting these measures as day-to-
day operating guides would be to ensure more precise control of money.
However, there is no clear evidence that money can be controlled more
closely through use of a strict total reserves or monetary base
operating procedure under the present institutional framework than
through current procedures.
Indeed, most of the evidence suggested
that these measures could produce more slippage because of supplyside shocks to the money multiplier.
These shocks tend to be
partially offset by changes in borrowing with a nonborrowed reserves
day-to-day operating target.
Under a total reserves or base target,
there would not automatically be an offsetting tendency.
In practice,
though, the precision of a total reserve or base target would be
improved through judgmental adjustments to the reserve path that
offset multiplier shifts.
Improvements could also be effected, and
the need for judgment reduced, by further simplification of the
reserve requirement structure (such as removal of the reserve requirement on nonpersonal ttme deposits if the FOMC wishes to control mainly
narrow money) and by a return to CRA.
While such changes would
tjghten the linkage between reserves and money, shifts between currency
and deposits would still tend to be a factor causing slippage--with
model simulations indicating greater slippage with the monetary base
as the operating target (whlch is essentially currency plus total
reserves) than with total reserves.
With a monetary base target,
short-run volatility in currency would lead to large variations in
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A30money supply because changes in the public's holdings of currency
would need to be offset by equal changes in bank reserves; and these
changes in reserves would, given the fractional reserve system,
force a multiple change of deposits in the money supply.
With a
reserves target, the changes in money supply would be no larger than
the currency variation; consequently, money supply would be less volatile with a reserves target.
b.
In any event, strict adherence to total reserve or base targets
appears to be impractical over short-run operating periods in the
current institutional setting.
clearly not feasible.
With the present LRA system, it is
If CRA were adopted, such targets might become
somewhat more practical, though efforts to attain them would accentuate
short-run interest rate fluctuations.
Such fluctuations, given the
inelasticity of money demand relative to interest rates over the short
run, would stem from the inability of the reserve supply to provide
at least partial accommodation to transitory money demand variations,
and would also result from remaining multiplier slippage.
In the process,
borrowing at the discount window would fluctuate widely, as banks reacted
to efforts by the Open Market Desk to reach the total reserve target.
c.
While there are practical questions about the feasibility of
targeting on total reserves (or the base) on a day-to-day or week-toweek basis, in a longer-run context a path for such reserve aggregates,
properly adjusted for multiplier shifts, could serve as a general guide
in helping to make adJustments in the nonborrowed reserve path or in
indicating the need for a change in the basic discount rate--as is,
in fact, present practice.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
For example, when total reserves are
-A31running strong relative to its adjusted path, this can be taken as
an indication to hold back on the supply of nonborrowed reserves
relative to its path (in order over time to offset the rise in borrowing)
or to raise the discount rate (in order over time to discourage a rise
in borrowing).
2.
Federal funds rate target
a.
Model simulations, given existing institutional arrangements,
indicated that in concept slippage in short-run money stock targets could
be little different on the whole under a funds rate targeting regime
than under a nonborrowed reserves regime.
However, in practice--to be
reasonably certain of attaining its long-run target--the FOMC would need
to be willing to move the funds rate quite actively when it was the
operating instrument and be able to predict fairly well the appropriate
extent, and indeed the direction, of the required change.
Uncertainties
in those respect9 of course were runong the factors leading to a shift
toward reserve targeting.
b.
A federal funds rate operating target would have advantages
if the FOMC wished to provide more scope for being accommodative to
variations in money demand, either because of uncertainties about the
proper path of money growth within its longer-run target band or
because of a belief that money demand disturbances are more likely
to occur than disturbances in the market for goods and services.
c.
The federal funds rate range under the current reserve
operating procedure has been much wider than under the earlier funds
rate targeting regime.
Moreover, the range under the new procedure
has generally been changed as the limits were approached--a practice
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A32that has been consistent with evidence quggesting that a wide range of
variation in the funds rate is a by-product of efforts to attain tight
control of the money supply.
In that context, a relatively narrow
acceptable funds rate range would only have advantages in the degree that
the FOMC (1) felt more qcope could be given in a particular period, for
one reason or another, to variations of money from a pre-set target,
or (2) felt that narrow funds rate limits provided a device that,
given the need to make judgments about sources of economic and monetary
disturbances, would prompt further assessment of underlying monetary
and other conditions by the FOMC in the interval between meetings.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-A33-
Monetary Control Project Staff Papers
Davis, ~ichard. "~onetary Aggregates and the Use of Intermediate Targets
in Monetary Policy."
Enzler, Jared.
"Economic Disturbances and Monetary Policy Responses."
---------,.,..
and Lewis Johnson.
"Cycles Resulting from Money Stock
Targeting."
Greene, Margaret. "The New Approach to Monetary Policy--A View from the
Foreign Exchange Trading Desk."
Johnson, Dana, and others. "Interest Rate Variability under the New
Operating Procedures and the Initial Response in Financial Markets."
Keir, Peter. "Impact of Discount Policy Procedures on the Effectiveness of
Reserve Targeting."
Levin, Fred,and Paul Meek. "Implementing the New Procedures: The View
from the Trading Desk."
Lindsey, David, and others.
Opera ting Procedures."
Pierce, David.
"Monetary Control Experience under the New
"Trend and Noise in the Monetary Aggregates."
Slifman, Lawrence, and Edward McKelvey. "The New Operating Procedures and
Economic Activity since October 1979."
Tinsley, Peter, and others.
Procedures."
"Money Market Impacts of Alternative Operating
Truman, Edwin M., and others. "The New Federal Reserve Operating Procedure:
An External Perspective."
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Monetary Aggregates and the Use of
nintermediate Targets" in Monetary Policy
January 12, 1981
Paper Written for a Federal Reserve
Staff Review of Monetary Control
Procedures
by
Richard G. Davis
,\.
Ir.troduction
The"obJective of this paper is to reconsider the rationale underlying the use of long-term monetary growth objectives as
"intermediate targets" in U.S. monetary policy.
to survey the following major areas:
(1)
The paper seeks
The nature of "inter-
mediate targets" and their use by the Federal Reserve.
(2) The
question of whether intermediate targets in general are an aid or
a hindrance
to effective policymaking -- the "decision theory"
approach to intermediate targeting.
(3)
The "monetarist" case
for money stock measures as intermediate targets.
( 4) Money stock
targets conceived as the centerpiece of monetary policy's contribution to a long-term anti-inflation strategy.
(5)
Circumstances
under which it might be necessary or desirable to suspend or modify efforts to track money stock targets -rate objectives?
(6)
a role for interest
Alternative money stock measures as long-
term intermediate targets -- reserve and monetary base measures,
credit measures.
(7)
"Mixed strategies" in which long-term
aggregate targets are blended with shorter-term focus on interest
rates and financial market conditions.
(8)
Structural features
of the economy limiting the Federal Reserve's flexibility in setting long-term aggregate targets.
Each of these areas has been the subJect of extensive
analysis both within and without the Federal Reserve over a period
of years.
While there can probably be said to be a fairly broad,
if shifting consensus on some of these topics, virtually all the
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 2 significant issues remain controversial to at least some degree.
Under these circumstances, the aim of the present paper is mainly
to lay out in one place the main strands of argumentation that
surround the use of ~oney and credit aggregate measures in an
"intermediate" target role, suggesting only as possible and appro, priate where the weight of relevant evidence may lie.
B.
Some Definitions
It is necessary to begin the discussion of "intermediate
targets" by trying to establish some terminology.
tinction needs to be
First, a dis-
made between the ultimate "goals" of policy,
such as price and output behavioi;which policy may be able to influence but which it cannot directly control, and the "instruments"
of policy over'which the policymakers do exert direct control.
Strictly speaking, the main instruments
of monetary
policy are confined to the discount rate, reserve requirements,
deposit interest rate ceilings,and the level and composition of
the System's open market portfolio.
It does not seem very useful
or convenient, however, to define the open market operations instrument in terms of the System's portfolio, even though moving
to broader definitions quickly encounters problems.
purposes, nonborrowed reserves and
For present
the nonborrowed monetary base
will be defined as potential policy instruments.
This seems
appropriate since short-term obJectives for them can be hit subject to correct forecasts of the behavior of the "operating
factors" affecting reserves, factors that are themselves
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 3 -
essentially independent of the level of open market operations.
At the same time, experi~nce has
amply shown that the Desk can
control the weekly average level of the federal funds rate with
,a,very high degree of precision in most weeks. Thus the funds
•
rate can clearly be included as a potential "instrument.."
"Intermediate targets, 11 as the term suggests, fall somewhere between the measures that represent ultimate policy goals
and the measures that can be treated as policy instruments.
~
Such
measures cannot ~e directly controlled in the sense that policy
instruments can be, but they are usually presumed to be much more
closely subJect to control through manipulation of the instruments
than are the goal measures.
The term implies that for some period,
short or long, policy instruments will be adJusted in line with
the intermediate target,which, in turn, will itself be readJusted
at longer intervals as required by pursuit of ultllllate policy goals.
Since the Open Market Committee meets at discreet intervals rather than continuously, it is useful
11
ro
distinguish between
short-term 11 intermed1ate targets, defined as any intermediate tar-
get set at each-Open Market Committee meeting to be pursued over the interval
between meetings, and "long-term" intermediate targets, which are expected to
be maintained over a multiTheeting horizon.
It may be helpful to use these various distinctions to look briefly at
how the Open Market Committee has used instruments and intermediate targets over
the period since the accord.
Until the first tentative use of money stock
targets in the early 1970s, the Committee did not, in fact, make use of long-term
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
intermediate targets at all.
,
Various measures of money market
conditions, net free reserves, the federal funds rate, and the
Treasury bill rate were used alone or in combination as 'both the
open market,policy 'instrument-and the short-term intermediate
target.
Indeed, until the advent of monetary targets there was
really no need for the distinction between short-term "instruments"
in the sense used here and the concept of "intermediate targets. 11
In the earlier years before money and credit targets, the settings
of the various money market instrument/target measures were customarily readJusted at each meeting,and there was no particular
expectation on anyone's part that a target setting voted at one
meeting would be maintained at'subsequent meetings unless economic
conditions just happened to make this appropriate.
This picture
was complicated slightly for a time by the introduction of a socalled "bank credit proviso clause"
1960s.
in the directives of the mid-
Under this procedure, a forerunner of the later "tolerance
ranges, 11 the Desk was instructed to adJust the money market conditions target in art appropriate direction if current bank credit
growth rates appeared to be deviating substantially from proJections.
Thus, the currently projected
rate of growth of bank
credit became part of a 11 feetloack 11 mechanism that could lead to
a resetting of the money market conditions target under certain
conditions.
In the early 1970s, the Committee began to orient its
decisions regarding money market conditions around mu1tiperiod
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- s objectives,stated in
(
the directive in qualitative terms,
va~ious measures of the monetary aggregates.
for
While such aggre-
gate targets could,of course,be reset each month, they began to
take on some of the characteristics of true long-term intermediate targets in the sense that prospective failure to achieve
them became a basi~ for readjusting the open market instruments
at subsequent Committee meetings.
The use of long-term inter-
mediate targets became more formalized under HR 133 in 1975.
The
targets announced under this r.esolution were set for a fourquarter horizon and were to be restated each quarter.
Given the
frequent phenomenon of "base drift, 11 it is clear that in dollar
terms, at least, if not in growth rate terms, the targets were in
fact readJusted each quarter, but at
least they functioned as
long-t~r!U intermediate targets over the multimeeting period
within the quarter.
The intention to make use of long-term intermediate
targets becomes even clearer under the provisions of the Full
Employment and Balanced Growth Act of 1978 since one-year growth rate
targets have subsequently been set only twice a year, always using
the actual level of the long-term target in the fourth quarter of
the previous year to define
rates.
the base for the targeted growth
In principle, the FOMC may readJust its long-term targets
at each meeting in light
of changing economic conditions.
If this
were,in fac½ done, the long-term targets would, in effect, be converted into short-term intermediate targets.
But it is reasonably
clear that in operating under the act, the Committee will consider
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 6 -
changes in previously announced targets only for very substantial
reasons.
Thus, both in practice and in theory, the present con-
duct of monetary policy does revolve around the setting of open
market instruments to achieve long-term intermediate targets that
are, in turn, readjusted only relatively infrequently in line with
changing prospects and objectives for the ultimate goal measures.
C.
The Case Against Intermediate Targets
Somewhat paradoxically, as the use of long-term inter-
mediate targets has become an increasingly dominant feature in the
formulation of monetary policy, one strand of thought in the
technical literature has arisen that_regards the use of intermediate targets as simply an inefficient or "sub-optimal" decisionmaking procedure.
For want of a better term, this line of thought
will be dubbed "decision theoretic" since it is concerned with
"optimal" decisionmaking per se rather than with specific views
on how the economy works.1/
(As used here, the term is intended to
include but not to be confined to the work of 11 optimal control"
theorists. )
l/ See, for example, BenJamin Friedman, "Targets, Instruments, and
Indicators of Monetary Policy, 11 Journal of Monetary Economics, vol. l
(1975), pp. 443-73; William Poole, 0ptimal Choice of Monetary
Policy Instruments In a Simple Stochastic Macro-Model, 11 Quarterly
Journal of Economics, May 1970, pp. 197-216; Jack Kalchbrenner
11
and Peter Tinsley,
0n the Use of Optimal Control in the Design of
11
~onetary Policy, Special Studies Paper 76, (Board of Governors of
the Federal Reserve System, July 1975); Benjamin Friedman, "The
Inefficiency of Short-Run Monetary Targets for Monetary Policy,
Brookings Papers on Economic Activity, 1977:2, pp. 293-335.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
11
11
- 7 Fundamentally, the argument made in this literature is
that since movements in the intermediate measure are of no intrinsic significance to the policymakers and since the relationship of
the intermediate targets to ultimate goals is likely to be subJect
to various kinds of perturbations, the pursuit of intermediate
targets is not an effective way to achieve the desired path of the
goal variables.
According to this view, it would be more efficient
to set forth values of instrument variables such as nonborrowed reserves or short-term interest rates directly in terms of the desired
behavior of such goal measures as nominal income.
In this concep-
tion, movements in such "intermediate" measures as the money supply
would be treated as, at most, one among many sources of "information" on possible needed resettings of the instrument variables.
Further, the presumption would be that while the Committee might
set provisional multiperiod target paths for such
11
i:nstrument 11
measures as nonborrowed reserves, these paths would normally be
revised at each meeting in light of incoming information.
There
would be no desire or expectation that any particular growth rate
for either an instrument like nonborrowed reserves or an intermediate measure such as the money stock would be sought for its own
sake.
The case against using such measures as the money stock
as longer-term intermediate targets in the context of this kind
of analysis is easily stated.
As long as the relationship between
the path of the intermediate target appropriate to achieving the
ultimate goals of interest is subJect to constant potential change,
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 8 -
and as long as there is a more or less steady inflow of information making possible a reassessment of this relationship, then
the kind of long-term mu1 tiperiod intermediate targeting currently in use is hard to justify.
Indeed, even as applied to short
periods such as the period between FOMC meetings or the somewhat
longer period between receipt of quarterly readings on GNP, the
use of intermediate targets is, according to this line of analysis,
generally not efficient.
The use of such short-term intermediate
targets implies, for example, that if a target measure (say M-1B}
were drifting off course, the instrument measure, say nonborrowed
reserves, would be adjusted to bring the intermediate target
measure back on course.
~.
But the critics of this approach note
that there may be many reasons for an off-target drift of the
M-18 intannediate measure.
In one simple but widely used analyti-
1/
cal framework,- the possible sources of off-target drift in the
money stock can be separated into three categories:
(1) the ex-
pected relationship between nonborrowed reserves and M-18 (the
"supply of money function") may have shifted, (2) the demand for
M-18 (\n relation to intQw.e and interest rat~s) may have shiftad~
or (3) the demand for goods and services (the neo-Keynesian "IS"
curve) may be stronger or weaker than expected, pulling up (or
down) both the levels of M-'iB and of i'llterest rat.es assochta.:i ,11ith
the initial nonborrowed reserve path.
1/ See Friieduuat1,,
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Iniefficie!!'lcy of Short-Run Monetary 'Targets. 11
- 9 -
As this line of analysis points out, it is only ,r. the
first case, where there is, in effect, a shift in the demand for
'
reserves, that it will prove optimal to adhere to the intermediate
money stock target.
The reason is that it is only in this case
that readjusting the supply of reserves to keep money on target
will, at the same time, also keep nominal income and related macroobjectives on target.
(Indeed, it should be noted, shifts in the
demand for reserves are automatically accommodated under a federal
funds rate instrument whether or not money stock measures are being
used as intermediate targets.)
To the extent that the second case
prevails, however, a shift in the demand for money, it is appropriate to accommodate the shift rather than to resist it.
In this
case, adhering to the intermediate money stock target by tightening
up on the reserves (or interest rate) instrument variable will
merely serve to deflect the ultimate goal variables from their desired path.
In the third cas~, a shift in the demand for goods and services, it would of course be more appropriate to maintain a money stock
target unchanged than it would be to maintain an interest rate target unchanged.
The reason is that adherence to a predetermined
money stock target would at least permit interest rates to move in
o/
a direction that would partially, but only partially, offset the
e~fect of the initial shift in aggregate demand.
However, because
the 9ffset would be only partial, it would be necessary to move the
money stock target itself to keep aggregate demand on target.
in the case of a shift of aggregate demand, the use of a money
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Thus
- 10 -
stock intermediate target, while preferable to an interest rate
target,is nevertheless not an optimal procedure to follow.
\
Given the various possibilities, decision-theoretic
analysis argues that movements in an intermediate measure such as
M-18 cam at bes,
D'.!
.a :source of inforo.in:ion, ar,d on1/ -0.,e .among
many, as to how changing relationships may require a resetting of
the instrument to keep the goal variables on,trackG
stronger than expected.
For example,
'
But given the highly erratic performance
of monetary variables in the short run, and given the ready and
prompt flow of direct information on the real and
financial sides
of the economy in the United States, it is difficult to see how
movements in the money stock measures can even be
7egarded as a
very useful source of information on the, economy.
The net con-
clusion of this line of analysis, then,would seem to be that intermediate targeting on the monetary aggregates is inef~icient.
While the line of analysis that criticizes the use of
intermediate targets from the point of view of decision theory
I
appears entirely correct on its own terms, it is by no ~eans the
last word on the subJect.
First, it proceeds from completely
general assumptions about the structure of the economy in the
sense that it assumes no specific restrictions on the relative
sources of instability in the economy or on the size of cricital
parameters such as the interest elasticity of the demand for money.
If more specialized assumptions are made (as is suggested below
in discussing one strand of "monetarist" support for the use of
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 11 money stock targets), its skeptical conclusions about intermediate
targets are undercut.
'
More-importantly, however, the somewhat
narrow focus of ~he decision-theory view of intermediate targeting fails to address some of the larger concerns and issues that
have,in fact,generated widespread support for the use of long-term
monetary targets over the past decade or so.
D.
The Case for Monetary Aggregate Intermediate Targets-Correcting ~r~,;1cJic;i1 Si~S:8!, iri •lJ,'-.:t_i t ?,D°:"'C'f
One of the earliest arguments for the use of the mone-
tary aggregate targets antedates both the increasing professional
and public interest in "monetarism" of the:1960s and the emerging
problem of accelerating inflation in the 1970s.
thought,
This line of
which clearly continues to have some relevance,
argued that the use of money market conditions instruments (whether
net free reserves or short-term interest rates) in the absence of
explicit attention to growth rates in measures of money and bank
tions in money and reserve growth and can lead to systematic misjudgments of the impact of monetary policy around cyclical turning points.
The argument starts from the proposition that the Open
Market Committee will tend to show a natural caution in changing
its money market conditions obJectives in the face of cyclical
changes in the demand fo~ money and credit.
The result, accord-
ing to this view, is that such a targeting approach leads more or
less inevitably to an acceleration in money and credit growth in
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
the ea~ly stages 'of economic expansi'.o:n 'as intere'st rate 'objective·s
are
not perm'.I. tted to rise rapi'.dl.y ''eno'tigh to ':modera'te the 'cyclical
'strength'eni'ng 'of money and credit demand and as "the rise in money
market rates '(and correspond.1.n:g reduction in net free reserves)
that does take place is misconstrued as a firming policy.
The
mechanism suggested by this argument does not explain how upper
turning points in the business ·cycle occur, but it do'es suggest
that once the economy begins to weaken, declining demands for
money and credit will push down interest rates.
If, using money
market rate targets, the FOMC does not allow rates to fall rapidly
enough, again misconstruing the decline in rates as a sign that
policy is moving in an appropriately "easing" direction, the result will be a deceleration in money and reserve growth, or even
outright declines in these measures.
Thus, contrary to belief and
intention, the actual impact of policy proves to be procycl ical,
worsening the economic downturn.
Something of this sort was alleged
to have happened in early 196~when, with the economy moving into
recession, the Open Market Committee progressively relaxed its
money market conditions objectives but apparently not rapidly
enough to prevent several months of outright declines in money and
bank reserves.
It may be, as this view argues, that the use of money
market conditions targets produces problems of this sort if there
is a systematic tendency for the Committee to adJust its targets
"too slowly. 11
However, it does not follow from this that the
proper remedy is to move to long-term intermediate monetary
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 13 -
aggregate targets.
In the language of the decision theory critics
I
of intermediate targeting already cited, an equally plausible
conclusion might be that current and recent money and reserve
behavior ought simply to be one of the "information variables"
the Committee should take into account in setting its money market
conditions obJectives.
E.
A "Strong Monetarist" Case for Intermediate Money Stock Targets
What might be called a "strong monetarist" case for the
use of money stock measures as intermediate targets can be derived
in various ways by restrictions on the general assumptions of the
"decision theory" critics of intermediate targeting.
Basically,
of course, the end product is simply the proposition that movements in monetary growth rates call the tune for movements in nominal demand.
In the usual exposition of this view, the first im-
pact of variations in nominal demand is assumed to be on real
output, with the ultimate impact concentrated entirely on prices,
but only after a substantial lag.
More recently, under the in-
fluence of the "rational expectations" theorists, some monetarists
seem to have taken the position that the entire impact of variations in money growth rates that are nonrandom {and hence anticipated by the public) is on prices even in the short run.
In this
variant, only unanticipated changes in money growth rates have any
effect on the real economy even in the short run. This "rational
expectations" wrinkle, however, is inessential to the "strict
monetarist" prescription of money stock intermediate targets in
view of money's presumed dominating effect on nominal aggregate
demand.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 14 -
A "strict monetarist" case for money stock intermediate
targets can be constructed even in the face of decision theory
criticism of such targets in various ways.
One possibility is to
assume that the demand for money is essentially stable (stable "LM")
and that the interest elasticity of the demand for money is negligible (vertical LM).
These two assumptions are
sufficient to
assure that aggregate demand will track money quite closely.
turbances in the demand for
Dis-
money are ruled out by assumption,and
any disturbances in the market for goods and services (the "IS"
function) will not cause aggregate demand to stray from the level
implicit in the money stock target.
Further, since disturbances
in the market for goods and services cannot affect actual aggregate
demand, it is apparently unnecessary to continually readJust money
targets in light of incoming information in the manner suggested
by the decision-theoretic critics of monetary targeting.
Alternatively, some monetarists seem to argue that the
private economy is "inherently stable" in the absence of monetary
shockso
Such a view suggests the interpretation that in practice
disturbances in the market for goods and services are likely to be
small.
Such a conclusion, in turn, again in qonjunction with the
assumption that the demand for money is stable, would also logically imply that the use of money targets is an effective way to
regulate aggregate demand.
Of course, "extreme" assumptions,such
as absolute
stability in the demand,for money or completely interest-inelastic
money demand,should perhaps be regarded as limiting cases.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The
- 15 majority of "strict monetarists" would perhaps regard them merely
as approximations, but approgimations sufficiently close to reality
to Justify the use of money stock targets even in the face of their
nonoptimality under more general assumptions.
The "strong monetarist" case for intermediate targeting on money stock measures must,of course,rest essentially on
empirical evidence.
Over the years, evidence that nominal income
dances to the tune of money growth rates has been offered both
from examination of the cyclical movements of money and the economy
and from examination of the relationship of changes in nominal GNP
to current and lagged changes in money.
In addition, a number of
economists have attempted to provide evidence on one of the main
constituent arguments of the "strong monetarist" position, the stability of the demand for money.
Both the particular results and
the general methodological procedures involved in producing these
variouE strands of evidence remain a source of substantial dispute
in the economics professiono
And in recent years, the case for
stablemoney-demand has come under·particrirarly intense attack as
a result of evidence that standard relationships began to break
down in a rnaJor way around the rnid-1970s (as discussed further
below).
It should be noted that with some additional assumptions,
the strong monetarist case can be extended to a case not only for
long-term monetary targets but, indeed, for a monetary "rule"
where the intermediate target
once set and achieved-would never
be changed except perhaps in the face of long-term structural
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 16 -
changes in the economy affecting the trend rate of growth in output or in the demand for money.
One such additional assumption is,
of course,the familiar argument that while the economy dances to
the tune of money, the effects of money on the dancer are uncertain in the sense that they operate with long and variable lags.
In this case, it is argued, discretionary movements in the intermediate money target should be replaced with a fixed growth rate
target adjusted ultimately to a rate of growth consistent with a
zero (or perhaps slightly negative) rate of inflation.
To be sure,
faced with unfavorable initial conditions, such as a high embedded
rate of inflation, such a final fixed setting of the intermediate
target might have to be approached only gradually over time.
An alternative assumption that also seems to produce a
case for a monetary "rule" is the rational expectations argument'
that systematic (and therefore anticipated) movements in the money
1/
stock affect only prices.- In this case also, there is no room for
(systematic) discretionary adjustments in the money stock inter~.mediate target, which might g_ust as well-be fixed at·a-,·rate pro1
ducing the desired behavior of prices.
Indeed, under the rational
expectations argument, in the absence of long-term contracts, the
intermediate money stock growth target could be moved right away
1/ See Thomas J. Sargent and ~eil Wallace, "'Rational' Expect2tions,
the Optimal Monetary Instruwent, and the Opt1rral Money Surply Rl :e, 11
Journal of Political Econorey, vol. 83 {April 1975), pp. 241-54.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-·u·
- 17 to its desired long-term setting -- at least if the authorities
announce their intentions and are believed so that the shift in
policy generates no "real" effects.
F.
A "Practical Monetarist" Case for Long-Term Money Stock
Targets
A different approach to the use of monetary inter-
mediate targets starts from the search for a monetary policy
strafegy to deal with the chronic inflation problem of the past
15 years.
This alternative route to monetary targeting has been
termed "practical m'onetarism, 11 although the main thing it shares
with the "strong monetarist" approach is the conclusion that the
Federal Reserve should indeed make use of long-term money stock
objectives as intermediate ta,rge;ts.
Practical monetarism takes no particular position on
the extent to' which short-term movements in aggregate demand a.re
dominated by the behavior of the money supply.
But it does accept
the view very widely (if not quite universally) held that in the
long run, inflation rates will be no greater than are accommodated
by the long-run rate of monetary growth.
Under this view, excessive
monetary growth, is a necessary condition for inflation, and reductior.
in monetary growth is--a necessary condition for restoring reasonable
price stability.
A corollary of this view is that in the long run,
monetary growth rates affect only nominal magnitudes (including the
level of nominal interest rates) and have little or no effects on
real output, real interest rates,or the composition of output-as between consumption and investment, for example.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Consequently,
- 18 -
if long-run money growth targets are to be set, their ultimate
objectives should be framed entirely in terms of the desired
behavior of the price level.
It should be noted that in accepting excessive money
rowth as a necessary condition for inflation in the long run,
his view by no means rules out other possible sources of inflation over shorter periods and is therefore not committed to any
tight relationship between the path of inflation and the path of
monetary expansion.
Moreover, it does not even rule out the
possibility that nonmonetary forces may play a major role i,n initiating inflationary pressures, to which, in effect, money growth
must be accommodated if the authorities are to avoid sharply adverse short-run effects on output and employment.
The implication
of these po~nts, in turn, is that monetary policy by itself may not
be able, in any realistic sense, to bring inflation under control
without appropriate cooperation from other types of policies and
perhaps even from some "good luck"' in avoiding nonmonetary inflationary shocks.
Nevertheless, the clear agenda for monetary policy implied by
0
practical monetarism" is that whatever else may be need-
ed to bring inflation under control, the role of monetary policy
will have to be to reduce monetary growth rates over the long run
\
to rates compatible with' approximate price stability.
Indeed, to
the extent that the long-run real growth potential of 'the economy
can be estimated and trend behavior of money velocity guessed at,
it may be possible to quantify in rough terms the long-run obJective of policy in monetary growth rate terms.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Moreover, to the
- ;I.9 -
extent that current inflation ra;tes are at least partly "embedded"
in ,the economy in the form of -long-term contracts (including the
nomipal interest rates on outstanding long-term debt) and in
JKJ~J·
•form of inflationary expectations that are sub~ect only to gradual
change, it follows that monetary growth rates will have to partially
.. ...
~
accommodate the embedded inflation rate in the short run.
Thus the
objective of "practical monetarism" is to reduce money growth rates
to noninflationary levels, but only over time.
In effect, then, practical monetarism lays out a longterm strategy for monetary policy for dealing with inflation, a
strategy that is compatible with a very wide range of views about
the inflationary process and about the determinants of short-run
economic fluctuations.
The potential value of long-term inter-
mediate targets for monetary policy couched in terms of monetary
growth rates has several dimensions in this strategic context.
First,it provides an internal guideline for the Federal Reserve
itself against which shorter-term developments and decis~?M can
be viewed, creating, in effect, a kind of self-disciplining procedure for keeping the longer-run objectives in view.
Second,
the use of money stock targets in the contex:t of winding down excessive monetary growth over time provides a means of communicat'
ing~the objectives of policy with the rest of the government and
with the public.
To the extent that the Federal Reserve comes
to provide a credible record of achieving its announced intentions,
the very announcement of progressively lower monetary targets should
have an independently favorable effect on inflationary expectations.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 20 '
And
an improvement in inflationary expectations, in turn,
would be' a major means for reducing the cost of reducing_
inflation in terms of temporarily-reduced o~tput levels.
The use of monetary targets as a tool of communication might also be important in clearing up the•·
wiaespread public confusion about the relationship between
anti-inflationary monetary policies,· interest rates, and
the· actual performance of prices.
If it is understood that
the objective of anti-inflationary policy is to reduce monetary growth and ultimately
nominal interest rates, then the
frequently 'Voiced complaint that "tight money" raises interest
costs and thus actually abets inflation may become widely seen
as a misconception.
It should be noted that the possibility of defining
an anti-inflationary strategy in terms of a long-term path for
intermediate money growth rate targets, with its attendant ~dvantages for internal and external communication, apparently
has no analog in interest rate targets.
There is seemingly no
satisfactory way to state a long-term anti-inflation strategy
in terms of nominal or real interest rates as can be done in
the case of money growth targets.
As inflation comes down,
nominal interest rates would undoubtedly also decline.
But
to aim at a policy of "gradually reducing nominal interest
rates over time" to some presumed noninflationary 1sve1
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 21 -
{that is, to some long-terni 11 natural 11 real rate of interest)
appears to involve major risks that precisely the opposite
of the intended results will occur.
Reductions in nominal
rates might prove to be exactly the wrong prescription in
any particular period, and in the case of long rates, this
might well be unachievable.
Moreover, we have no way to
estimate what the real long-term "natural" rate of interest
may be or, finally, any particular reason to assume it is
constant over time.
It is, to be sure, possible to put a constraint
on policy in terms of "real" interest rates such as "nominal
rates should always be high enough to make real interest
rates positive. 11
Indeed ,it seems reasonably clear that nega-
tive real interest rates would be incompatible with price
stability and probably even with a nonaccelerating rate of
inflation.
But there seem to be serious practical measure-
ment problems in connection with the real interest rate concept that would have to be faced if it were to be used as
a formal,quantitative intermediate target.
These problems
stem from difficulties of defining and measuring expected
rates of inflation over the various relevant time horizons
and the problems of determining appropriate tax adjustments
for borrowers and lenders in different tax situations.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 22 -
As indicated, the "practical monetarist" case
for long-term money stock intermediate targets arises out
of the desire to find an appropriate strategy for dealing
with secular inflation.
One implication of this might
seem to be that if and when inflation were substantially
eliminated, long-term money targets would no longer be
needed -- and, of course, the Federal Reserve, in fact, had
no such targets during the long period between the accord
and the early 1970S3 when inflation was generally low by
present standards.
An alternative interpretation, however,
I
is that our recent experience with inflation has shattered
the public's confidence in the long-run stability of the
price level.
If so, any actual return to price stability
might prove very precarious because of the speedy and
sharply unfavorable effects on inflationary expectations
that any short-term re-emergence of inflation might produce.
Thus a commitment to the maintenance of monetary
growth at rates compatible with rough price stability might
continue to be needed as a financial anchor.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-
G.
23 -
Problems With Implementing a Strategy of Gradual Monetary
Deceleration -- Is Rigorous Adherence to Money Targets
Desirable?
While the "practical monetarist" position clearly im-
plies-the desirability of setting and-achieving money grewth
targets over some "long-run, 11 it does not necessarily imply that
such targets should be followed rigorously over shorter periods
of time.
Indeed, a number of questions can be raised about the
desirability of seeking to continuously reset instruments to keep
money growth along some predetermined track.
(Questions about the
feasibility of rigorously tracking money targets are not discussed
in this paper.)
The questions about the desirability of attempting
rigorously to adhere to money growth targets are well known.
In
the first place, rigorous control of money growth implies sacrifice
of any ability to influence interest rates, let alone to seek
specific interest rate objectives.
There are many reasons why the
_ Federal Reserve may have a legitimate concern for. interest rate
behavior per se.
On the international side, foreign exchange rates
are highly sensitive to interest rate differentials.
times when the policymakers
There may be
may prefer to constrain interest rate
movements at the expense of a temporary loss of control of money
growth to prevent adverse developments in the exchange markets.
Similarly, on the domestic side, the performance of
interest rates per se can be of immense importance to the cyclical
performance of the economy and to the functioning of financial
markets and to financial health generally.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
There is, for one thing,
- '24 a whole group of credit market "stress" phenomena that can -arise
when sharply rising open market interest rates collide with various
kinds of institutional or other frictions.
These phenomena include
_disintermediation,, .credit "-c:c:unches, 11 and other types of market,
"failures" that may result in part from institutionally created
frictions (such as interest rate ceilings on deposits or lending
rates) or simply from the difficulties financial intermediaries
may encounter in coping with sharply negative yield curves that
often materialize in "tight money" periods.
This whole array of
phenomena may, as some believe, have major significance for the
way monetary policy impacts the economy in the cyclical context,
and, of course, it is clearly relevant to the "lender of last
resort" function in the face of threatened financial breakdown.
Given such financial problems, or the risk that they might emerge,
concern with the behavior of open market interest rates might well
at times supersede the desire to keep monetary targets on track.
~To bersure, some proponents of money growth targets
would argue tha-c in the fot~rest of foreign exchange market stability
interest rate stability and the avoidance,of financial breakdown
are best served in the long run by stable monetary performance.
This may well be true, but it provides no guidance on the extent
to which monetary targets should be superseded in the short-run
in the event of imminent or ongoing foreign exchange or domestic
financial market problems.
Perhaps more subtle and pervasive questions about the
desirability of pursuing rigorously monetary growth targets are
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-
25 -
raised by questions about the stability of the "demand for money" --
that is, the stability of the amount of monetary assets (however defined) the public will seek to hold under given interest rate and
aggregate demand conditions.
The extent to which the demand for
money can be regarded as "stable" is controversial and is deeply
enmeshed in unsettled econometric issues of considerable complexity -- although there is at least agreement that conventionally
formulated demand functions for "money" in any of the thenconventional definitions did break down in a serious way in 1974
and for some time thereafter.
It may be useful to think of problems imposed by potential money demand instability in short, intermediate,and secular
terms.
Consider first a "short-run," defined, for example, as
movements within the calendar quarter, the usual unit of observation for econometric money demand functions.
Even if the demand
for money is "stable" in the usual sense, normal stochastic
developments not accounted for in the equation may lead to substantial deviations of actual from predicted money demand in the
, short runo
To ,be sure, if the function is . properly formulated
and 1s truly
II
stable, 11 these errors will be random and will tend
to wash out over time.
But this may not prevent these movements from presenting significant practical problems for policytT1akers.
In
theory,
once unexpected deviations in money growth are known to represent
genuine "shifts" in money demand, it will probably become appropriate to accommodate to them by readjusting money targets
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 26 -
accordingly.
The difficulty for the policymaker
lies in the
fact that if the normal random errors in money demand functions
are large (relative to the size of target ranges), it may take a
large number of observations of money demand "errors" before
"permanent" shifts in demand can be identified by the usual
statistical tests.
Thus in the presence of errors that may or
may not prove in retrospect to have been random and selfreversing, the policymaker
is faced with uncertainty as to
whether and how to respond.
Given the difficulties of identify-
ing a "shift" with the usual statistical techniques until several
quarters of data are available, this problem has., in fact., been
troublesome on a number of occasions.
In the short run, a shift of money growth from expected values is automatically accommodated under a federal funds
(
rate target unless it is substantial enough to trigger the sort
of feedback mechanism formerly provided by the "tolerance range"
errtployed by the FOMC in conjunction with its federal funds rate
targetso
Under a nonborrowed reserves approach, a movement of
money growth (for whatever reason) from the rate implied by th~-intermeeting nonborrowed reserve path will automatically set off
changes in interest rates.
The point is that under either approach,
off-target behavior of the money stock will at some point force a
decision to be made as to whether the behavior represents a demand
shift that should be accommodated. A federal funds rate approach
tends to err on
the side of assuming (implicitly) a demand shift
and accommodating it.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A nonborrowed reserve approach, on the other
- 27 -
hand, never fully accommodates unexpected money behavior whether
due to a temporary shift demand or to some other cause since
it does set off partly offsetting movements in interest rates.
As already suggested, the sort of long-term shift in
the demand for money that apparently occurred around 1974 requires
either that the money targets be changed to a~commodate the shift
or, where appropriate, that "money" be redefined and new targets
set.
Again the problem is that knowledge that the demand curve
has shifted is never absolutely firm and can, at best, only emerge
gradually over time.
And again the point is that judgment must be
made whether to suspend or modify pursuit of existing target growth
rates.
More generally, the experience of recent years has
suggested to many that forces are in place that tend to generate,
however irregularly, a long-term downward drift in the demand for
money that will at the least complicate the implementation of the
"practical monetarist" strategy.
One source of such long-term
forces would be delayed institutional, technological, and capital
~
investment responses to rising incentives to economize on the use
1/
of transactions balances paying below-market rates of interest.Another possible source of long-term shifts in the demand for
money is the increasing sophistication of computer and electronic
l!
See Thomas D. Simpson and Richard D. Porter, 11 Some Issues Involving the
Definition and Interpretation of the Monetary Aggregates,' 1 a paper presented
at the 1980 Federal Reserve Bank of Boston Conference on Controlling the
Monetary Aggregates, Bald Peak, New Hampshire, October 7, 1980.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 28 -
technology.
Technological ~dvances have encouraged the, cr~~tion
of new, quasi-transactions instruments paying market interest rates
as well as new means of minimizing holdings of the more traditional instruments.
The future progress of such development~ is of
course unknown.
But the prospect is that emerging alternatives
to transactions instruments that pay below-market rates will be
exploited, possibly at an accelerating rate.
Obviously, such a
prospect makes more uncertain the path toward noninflationary
monetary growth envisioned by practical monetarism as well as the
ultimate distination of such a path.
Indeed, it raises questions
about the very future of any transactions in$trument that must
face the competitive hurdle of a reserve requirement 11 tax. 11
H.
Which Aggreg~te To Use?.
Different approaches to the use of monetary aggregate
measures as long-term intermediate targets may imply different
answers
to the question of just which aggregate or aggregates
to target.
As noted, the "decision theory" approach in general
concludes that any intermediate target would be inefficient. It
does suggest, however, that the relative value of alternative
measures as monetary "indicators" should be evaluated in terms of
their relative value as sources of information on current movements in the goal variables.
The strong monetarist view, of course,
implies t h a t ~ money stock measure should be used, but monetarists (like other economists) have been divided over the years
as to Just which money measure might be best.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 29 The "practical monetarist" approach does not appear
to imply any particular measure of money.
Indeed, it does not
even seem to rule out the possibility that some nonmonetary
financial measure might be used to define the long-term strategy
for restoring price stability.
In principle, long-term targets
defined in terms_'of credit or reserve aggregates, for example,
might serve as well as money stock measures.
As a matter of ab-
stract theory, it appears that the stabilization of any financial
magnitude in a deterministic model of the economy will stabilize
y
the price level in the long run.
The practical question would
seem to be which measure or measures have the most stable demand,
have the tightest relationship to nominal GNP and other major macrovariables, have the closest relation to price performance in the
long run; are most clearly subject to Federal Reserve control,
and have the greatest value in terms of their impact on expectations and communication.
Unfortunately no single measure, whether
monetary or otherwise, seems to emerge as the clear choice given
this multiplicity of crite~ia.
/
J
Numerous-attempts have_been made over. the pas±.decade
or so to discriminate among various definitions of money (and
bank credit) on the basis of the closeness of the relationship of
nominal GNP to current and lagged values of the various competing
1/ See Franco Modigliani and Lucas Padademos, "The Structure of
Financial Markets and the Monetary Mechanisrr., 11 paper presented at
the Federal Reserve Bank of Boston Conference on Control 1ing the
~onetary Aggregates, 1980.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 30 -
aggregates.
The results, in general, have failed to turn up clear-
cut winners, and they have proved sensitive to apparent 11 deta1ls' 1
of equation specification and estimation period.
With respect
to money demand equations, conventional econometric equations
for all the new and old money stock measures apparently underwent
a shift in the mid-1970s.
Moreover, t~e econometric properties
of the demand equations for all the various measures have also
apparently deteriorated in recent years, producing implausible
coefficient values for individual explanatory variables.
With
respect to the breakdown of the equations after the middle of
the decade, broader aggregate measures that include the new instruments, such as 1~0W accounts and money market funds, show substantially smaller cumulative drift than narrow measures.
However,
there seems to be little agreement as to the precise source of
the better performance and hence some problem in prescribing the
best measure for future use.
One recent survey of the econometric
properties of the various money stock measures concluded that
"one unambiguous result these statistical exercises provide is that
money demand and reduced-form equations ..,, have
no power - to discrim"'
--- ...
inate among alternative aggregate definitions. all While this may
overstate the case, it is clear that the econometricians have
not yet been able to resolve the problem in any decisive way.
1/ Neil Berlanan, "hbandoning Monetary Aggregates, 1 paper presented to the Federa 1 Reserve ~ank. of Boston Conference 011 Colltro 11 i ng the
Monetary Aggregates, 19809 p. 22.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 31 -
A priori considerations also seem to fail to point to
a clear choice.
It has been suggested from time to time, for
example, that money stock measures ought to make better policy
targets than bank credit measures -- to which they are obviously related on the supply side -- because there are allegedly
fewer substitutes for "money" (at least as defined in transactions terms) than there are for "bank credit" and because
"money" as an aggregate is allegedly more homogenous than
"bank credit. 11
The implication presumably is that money mea-
sures should have more stable demand functions than bank credit
and that money demand is less likely than the demand for bank
credit to be disturbed by financial market innovations and by
market developments affecting relative interest rates.
If
these propositions were true in the past, their plausibility
would seem to have been considerably weakened by the financial
innovations of recent years.
With regard to choices among particular definitions of
the money stock, a priori arguments again seem to fail to provide decisive answers.
On the surface, rece.at experience
with changes in the utilization of conventional transactionstype instruments and the invention of new ones would seem to
point to a preference for broader money stock measures that
include the new instruments and away from narrow measures
such as M-18.
Other considerations tend to undereut this
conclusion, however.
(1)
As Regulation Q is phased out, a
substantial proportion of the broader measures will become
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 32 -
more competitive with market instruments, increasing uncertainty
about the interest rate sensitivity of the demand for the broader
1/
measures over the transition period.-
(2)
The demand for the
broader measures may be relatively sensitive to minor changes in
relative yields -- such as the rate on money market funds versus
open market rates, for example -- and hence may show a relatively
unstable relationship to
the average level of rates.
(3)
Growing
portions of the broader aggregates are not currently subJect to
reserve requirements, creating control problems in a reserveoriented tactical approach.
(4)
Finally, there is a good deal
of uncertainty as to just how broad a measure to use once the
decision moves beyond M-1B.
Why exclude·(include) large CDs, for
,,
example?
Such decisions•could make a substantial difference in
the short- to medium-term if only a single measu~e were to be
targeted.
Broad credit aggregates have been studied far less extensively in the literature than money stock measures.
The available
evidence suggests that a:: least some broad ~redit measures may be as
2/
closely or more closely related to GNP than the money measures~
What, if anything, this may mean in "cause and effect" terms is
unclear from the existing research.
1/
For no apparent reason,
See Simpson and Porter, "Some Issues, 11 p. 14.
2/ See Richard G. Davis, "Broad Credit Measures as Targets for
Monetary Policy, 11 Federal Reserve Bank of New York, Quarterly
Review, Summer 1979, pp. 13-22.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 33 -
moreover, other broad credit measures evidently perform worse
than the standard money measures and all the caveats associated
with the methodology of "reduced form" equations and the sensitivity of their results to details of specification and choice
of sample period need to be kept in mind.
With broad credit
measures, moveover, the problem of control by the Federal Reserve
would presumably be even more difficult than it is in the case of
money measures.
Successful use of funds rate approach to finan-
cial aggregates implies the existence of a reasonably stable demand function for the aggregate.
But no one has even suggested
that the demand for broad credit totals should be a stable function of aggregate demand and short-term rates, let alone provided
evidence that such a relationship actually exists.
A reserve
tactical approach would appear completely irrelevant in the case
of broad credit measures.
Given the problems with money stock and credit measures,
especially with respect to control, it is tempting to ask whether
a measure such as total reserves or the total monetary base might
not prove a suitable intermediate target?
In some respe~ts the
control problem is clearly easier to solve for these measures than
for the money stock measures.
The only slippage between a non-
borrowed reserves (or nonborrowed base) instrument and a total
reserves (or total base) target is obviously borrowings.
Ignoring
the short-term problem created by lagged reserve accounting, this
clearly would be easier to deal with than the multiple slippages
that exist between these instrument measures and the money stock
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 34 -
measures.
In the case of a funds rate approach, however, con-
trol problems for total reserves would be similar to those encountered with money stock objectives.
Apart from the controllability question there are some
respects in which the total reserves or the monetary base seem
inferior to the money stock measures, or at least no improvement.
The available evidence does seem to indicate fairly decisively
that the relationship of nominal GNP to current and lagged
measures of total reserves and the base is substantially less
1/
close than the relationship of nominal GNP to money stock measures7
Moreover, any shifts in the demand for money will be transmitted
into corresponding shifts in the demand for reserves and the
monetary base to the extent that there are corresponding shifts
in the demands for required reserves and/or currency.
Finally,
total reserves and the base have the potential for some special
problems of their own that would tend to destabilize their relationshjp to aggregate demand and thus weaken their value as
intermediate targets.
These include shifts in the public's
desired currency/deposit ratio, shifts in the demand for excess
reserves, and shifts in the demand for nonmonetary reservable •
1/ See Richard G. Davis, "The Monetary Base as an Intermediate
Target for Monetary Policy," Federal Reserve Bank of New York,
Quarter!~ Review, Winter 1979-80, and Carl Gambs, 11 Federal Reserve
Intermediate Targets: Money on the Monetary Base?" Federal
Reserve Bank of Kansas City, Economic Review, January 1980, pp. 3-15.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-
liabilities.
35 -
In short, targeting on the base or reserves risks
transmitting to the economy at large shifts not only in the demand
for money
but also
11
supply-side 11 shifts emanating from changes
in the demand for reserves for given levels of the money stock.
I.
The Use of Mixed Strategies
The preceding discussion suggests that even granting the
desirability of setting, and meeting, long-term monetary targets,
there remains a problem of bridging the gap between short-run
decisionmaking
and long-run strategy.
If one takes the view that once long-run money targets are
set, short-term decisionmaking
should concentrate solely on the
task of hitting them,-the problem of meshing short- and long-run
decisionmaking
is, to be sure, considerably simplified.
But even
\
in this case, the impossibility of hitting money stock targets
precisely in the short-run requires consideration of how short-run
tactics should be designed.
Thus consideration has to be given to
how the setting of short-run tactical instruments should respond
to misses in keeping money growth on track.
In particular, over
how long a time horizon shou1d the Federal Reserve aim to restore
money growth to its target when such misses have occurred?
If one goes beyond the pure money control approach and
takes the view that there may be good reasons for suspending pursuit of money targets from time to time or for changing them, the
problem of meshing short-run decisionmaking
strategy obviously becomes more difficult.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
with longer-term
- 36 -
Given the uncertainties in identifying demand shifts, for
example, there are risks both in under-reacting to, off-target
movements in money supply -- perhaps the natural bias of afeder~
funds rate approach -- and of ~-reacting, perhaps the natural
bias of a nonborrowed reserves approach~
And the same kinds of
problems of meshing short- and long-run decisions will, of course,
arise at any time interest rate levels (or their volatility) become
themselves a source of direct concern and thus a reason to
suspend or alter money growth targetso
In the face of issues of this kind an obvious suggestion
is the adoption of some sort of "mixed strategy" combining, in
some manneru the use of long-run intermediate targets with shortrun flexibility to respond to perceived shifts in money demand and
'
the impact of interest rates on international and' domestic
finan-
cial markets.
Such a strategy would attempt to achieve the main
objective of the "practical monetarist" program:
An orderly pro-
gressive slowing over time in the monetary aggregates as a necessar:
component of an overall anti-inflation strategy.
It would start
from the premise that whatever uncertainties there may be about
'
shifts in the demand for money and about long-term changes in the
trend growth of velocity, money growth rates, however measured and
defined, have been too high in recent years and perhaps too variable
over the business cycle.
On the other hand, this approach would
also take the view that money targets are not to be pursued mechanically, but that, as indicated, there will be, from time to time,
good and sufficient reason to suspend or modify them.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 37 -
This sort of "mixed strategy" is clearly open
to competing dangers.
On the one hand, overly rigorous
concentration on sticking to money targets exposes policy
to potential risks from shifts in money demand, from interest rate movements with undesirable market impacts,
and from the generation of instability in interest rates
and perhaps in the economy itself.
The other side of the coin, however, is that
the compromising of long-term money objectives risks continual postponement of movement toward a noninflationary
financial environment and it risks the loss of Federal
Reserve credibility.
It thus threatens to throw away the
potential expectational and communications benefits of
monetary targets and the anti-inflationary strategy based
on their use.
Obviously it is easier to state these opposing
dangers than it is to point a course between them.
The
attempt to achieve a reasonable balance of opposing risks
requires consideration of the choice of operating tactics
and of institutional arrangements, neither of which are
within the preview of this paper.
In any case, unless
the Federal Reserve moves toward a rigidly inflexible
pursuit of monetary aggregate targets, it will probably
,
not prove possible to solve the dilemmas of relating
short-run decisions to long-run strategy in purely technical terms.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 38 -
J.
Structural Limitations on the Setting of Intermediate Policy
Targets
In concluding this discussion of intermediate targets it
is important to emphasize an issue touched on at various points
earlier in the paper:
namely, the existence of structural fea-
tures in the economy that may impose significant limitations on
the settings of intermediate targets, however formulated.
The
issue of such limitations arises in part from the apparent fact
that monetary policy initially impacts mainly on real variables
and only more gradually on prices.
To the extent this is true,
'
settings of intermediate
targets have to take into account limits
on the tolerable short-run impact of such settings on output and
other real variables and on a potentially fragile financial system.
The existence of such limits has already been alluded to in de-
scribing the practical monetarist_program as one of gradual reduction in monetary growth.
The case for such gradualism rests on a
desire to accommodate partially the embedded rate of in'flation and
.
thus to reduce short-run impacts on real variables and the finan-
cial system.
The speed with which inflation can be expected to respond
to monetary policy, the ext~nt to which policy affects the real
economy in the short run, and therefore the extent of the potential
flexibility to adjust intermediate monetary targets are the subjects
of debate.
Monetary restraint (however indexed) is generally
viewed as affecting inflation by,restricting aggregate demand.
Lowered aggregate demand is visualized as creating market slack
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-
39 -
that leads to slower average rates of increases in prices and
wages.
In most versions of this process, the resulting experience
of slower inflation gradually lowers the public's expected rate of
inflation along with the actual rate.
Tpe gradual decline in
the
expected rate of inflation, in turn, makes possible a progressive
slowing in actual inflation as long as some level of slack in markets is maintained.
The bulk of the econometric evidence (and indeed of more
"casual" sorts of evidence) from the postwar period, however,
tends to be pessimistic about the speed with which inflation can
be slowed through this process and about the costs entailed in
1/
terms of less-than-potential levels of real activity.- This
evidence suggests that given (1) existing rigidities in the pricing mechanism, (2) little or no direct favorable impact of policy
actions on inflationary expectations, and (3) no assistance from
other government policies, anti-inflationary policies that operate
through aggregate demand alone could well require several years of
slack to achieve major reductions in the underlying rate of inflation.
Taken by itself, this evidence therefore suggests that
efforts to reduce inflation through progressive reductions in
targeted money growth would have to proceed quite gradually to
avoid major adverse effects on real activity.
1/ A summary of findings on this matter is given in Arthur M. Okun,
"Efficient Disinflationary Policies, 11 American Economic Review, vol. 68,
(r1ay 1978), p. 348.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 40 -
There are, to be sure, plausible grounds for a more optimistic view of the problem than these results by themselves seem
to suggest. For one thing, it is possible that other kinds of government policies (such as policies to reduce/rigidities in the pricing mechanism and tax policies to reduce costs or to provide incentives to limit wage and price increases) could alter the structure of the economy in a favorable direction.
'
Such changes could
speed up the responsiveness of prices and wages to market conditions.
Such a speed-up, in turn, would make possible a quicker re-
duction in inflation at smaller cost, even without any favorable
alteration in the way price expectations are formed.
These develop-
ments could thus make possible a more rapid implementation of the
practical monetarist program.
Further, it is possible that monetary policy itself could
have a directly favorable impact on inflationary expectatio~s that
would also speed up the process.
The available econometric evi-
dence generally does not allow for such an impact.
Instead,it
assumes that inflationary expectations are the mechanical product
of past inflation history.
This is a major factor accounting for
the sluggishness with which inflation rates apparently respond to
aggregate demand policies in this literature.
Yet one of the in-
tended potential benefits of defining a monetary strategy in terms
of progressively reduced targets for monetary growth is precisely
that such an announced policy, if credible, could by its very
announcement lower inflationary expectations.
Such an impact on
expectations, in turn, would make a possible quicker lowering of
actual inflation than is implied by the existing evidence.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 41 -
In the limit, as some versions of the rational expectations theory suggest, the impact of an announced policy of monetary reduction might produce a more or less immediate correspond1/
ing slowdown in inflation with little or no loss of output.Such an outcome is at least imaginable to the extent that (1) money
does affect prices proportionally in the long run and is neutral
as to real effects, (2) the targeted slowdown in money growth is
announced and the announcement is believed, (3) the public quickly
foresees the ultimate price effects of the targeted money slowdown,
and (4) markets are free to respond without frictions to the implications of these changed expectations.
This extreme "optimistic" case, however, also seems open
to serious' questions. The questions stem from the apparent existence
of rigidities in the pricing mechanism and the likelihood that the
generally expected rate of inflation may be subject to a substantial degree of inertia.
First, there are significant institutional
In "The Ends of Four Big Inflations," Working Paper 158, (Federal
Reserve Bank of Minneapolis), Thomas Sargent cites the hyperinflations of four central European countries in the early 1920s
that ended abruptly with, apparently, no more than relatively
moderate and temporary adverse effects on real output following
the announcement of policy reforms. The relevance of this experience to the present U.S. situation can, of course, be
questioned because of thd differences in circumstances. Of particular importance to the present discussion is the fact that the
policy changes made in these four countries included sweeping and
binding restrictions involving both monetary and fiscal policy, in
two cases imposed under agreementwith foreign creditors. These
were clearly much more far-reaching changes than anything contemplated in the CGntext of the intermediate monetary target discussion.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
characteristics of the economy that would seem to prevent any
immediate full translation of slower monetary growth into price
effects almost whatever its impact on expectations.
Thus there
are many long-term contracts in the economy that implicitly embody
some ongoing expected rate of inflation.
Examples include multi-
year wage agreements, utility rate schedules, rents subject to
fixed-term leases, and fixed interest rate debt instruments.
In
markets where long-term contracts exist, changes in demand conditions are likely to have implications for the level of real
activity since prices cannot irmnediately adjust.' Moreover, any
actual change in the rate of inflation shifts the real terms of
trade between the contracting parties and is thus likely to have
ieal effects.
'I
Apart from long-term contracts embodying existing inflationary expectations, the vast majority of labor and product
markets depart substantially from the textbook model of organized
financial and commodity markets in which prices react continuously
.
1/
and sensitively to changed conditions.Instead, for whatever
reasons, most prices and wages are set relatively infrequently so
that demand pressures are likely to affect volume before they
affect prices -- though, to be sure, wage- and price-setting practices are·themselves probably responsive to the rate of inflation
in the longer run.
1/ See, for example, Arthur M. Okun, "Inflation: Its Mechanics
and Welfare Costs, 11 Brookings Papers on Economic Activity, 1975: 2,
pp. 351-90.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 43 -
With respect to inflationary expectations, the precise
exte~t ,to which annottnced targeted reductions in monetary growth
c9uld be expected directly to lower the expected rate of inflation is conjectural.
The whole postwar economic experience might
• well suggest to a "rational" public that there is an-inflationary
bias in modern industrial economies and, indeed, in government
policies, whatever their announced intentions.
Moreover, the
actual per,formance of monetary aggregate behavior relative to
Federal Reserve targets since they were
has itself been somewhat mixed.
first announced in 1975
Thus the direct impact on in-
flationary expectations 0£ an announced reduction in targets for
monetary aggregates may not be very large -- at
And it is possible
least not by itself.
to doubt, even under the most favorable circum-
stances, whether inflationary expectations for the public at large
are importantly influenced by announcements of monetary policy intentions.
tions
That such announcements might loom large in the
expecta-
of participants in sensitive and sophisticated financial
markets is certainly plausible.
That they would loom equally large
in the wide range of less sophisticated markets seems much more open
to doubt.
Thus, it seems likely that,
11
rational" or not, the public's
expectations of continuing inflation are not likely to be easily
overcome.
And for this reason as well, it appears that intermediate
monetary targets settings would have to accommodate ongoing inflation
to some degree to avoid unacceptable effects on real output.
Finally, the point already mentioned that the feasible
settings of intermediate targets, of whatever kind, may be
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 44 -
constrained at times by structural features of the financial system also needs emphasis.
Reductions in monetary growth rates, to
the extent they lower the rate of inflation, will also lower nominal
interest rates over time.
But for reasons already noted, the full
I
1
For
effects ~n inflation are likely to be ~elt only gradually.
related reasons, the immediate effects of slower monetary growth
may well be higher rather than lower interest rates -- or
at least
in the short-term sector where prospective longer-term beneficial
effects on the inflation premium loom less large.
Clearly, for
reasons mentioned earlier, there may be limits to the extent to
which interest rates can rise in the short run without creating
serious problems for financial institutions and the functioning of
markets and thus ultimately for the "real" economy. Such limits
would therefore also restrict the extent to which financial aggregate targets could be lowered at any given time.
In summary, there appear to be important structural characteristics of the economy that put
,
'
limits on the feasible settings
'
of potential intermediate targets measures however defined.
To be
sure, the use of such targets may have directly favorable effects
on inflationary expectations.
To that extent, they can lessen the
problems created by these structural features.
Nevertheless, the
structural rigidities themselves would remain, whatever intermediate target or targets may be used.
Thus the choice of whether
to use such targets at all and, 1f so, which target or targets to
use seems unlikely to dispose of significant limitations on the use of
monetary policy.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
ECONOMIC DISTTJRBANCES AND r,,1'0NETARY POL I<~Y RESPONSES
Paper Written for a Federal Reserve
Staff Review of ~onetary Control
Procedures
by
Jared J. Enzler
Contributions to the paper were Made by J. Herry, n. 8attenberg, C.
Corrado, and W. Davis from the Hoard staff; by B. Figgins and V. Roley
froM the Kansas City Rank; and by J. TatoM from the St. Louis Bank.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
(
ECONOMIC DISTURRANCES AND MONETARY POLICY RESPONSES
Section I:
Introduction
The central topic of this project is an evaluation of the rela-
tive merits of using the federal funds rate or a bank reserve measure as
the monetary policy operating instrument.
While much of the project is
concerned with the ability to achieve intermediate monetary-aggregate targets, this paper is mainly concerned with the ability of policymakers to
achieve their overall objectives for the economy.
Using both reduced
forms and the Board's quarterly structural econometric model, the consequences of various disturbances to the economy under selected monetary
policy operating rules are examined.
It should be noted at the outset that the models used in this
paper, which are quite representative of the types in current use, do not
take into account the recently popular notion that the structure of the
economic system may depend on the existing policy rules.
Practical models
which incorporate this notion are not yet available; if they were, they
would undoubtedly produce quite different results from the ones reported
below.
Whether good models should incorporate this notion in a very power-
ful way is, of course, still a matter of great dispute.
Section II is backward looking.
The year since October 6, 1979,
has been characterized by dramatic fluctuations in both short-term interest
rates and money ~rowth.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The consequences of following a smoother money
-2growth rate policy over this period are investigated using both the quarterly model and some reduced-form models.
Then the results flowing from a
smoother interest rate policy are examined.
Section III addresses the consequences of deviations in monetary
policy instruments from desired paths.
In particular, it examines the
effects of various re-entry paths for policy instruments which have strayed
from planned values.
Once again, the question is addressed using both the
quarterly and reduced-form models.
The analysis presented in Section III turns out to be somewhat
unsatisfying.
It is argued in Section IV that in analyzing the consequences
of deviations of policy instruments from planned paths, it is very important
to understand what caused the deviations.
It is demonstrated, using the
quarterly model, that the consequences of an economic disturbance depend
both on the nature of the disturbance and on the policy response.
It
follows that the choice of policy responses should depend on historically
based knowledge concerning the relative frequencies and severities of the
different kinds of disturbances, and also should take into account any infor-,
mation available concerning the sources of the disturbances as they occur.
Section II:
Smoothing the Policy Instrument Path
In the past year both money growth and the federal funds rate
have been extremely volatile.
Table 1 reports quarterly average historical
values for the federal funds rate and the growth rates of various money
stock concepts used by the models analyzed in this section.
The funds
rate was even more volatile than the quarterly average numbers reported
in the table suggest.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The average rate for the month of April 1980 reached
-1-
a cyclical high of 17.6 percent.
By July the rate was down to 9 percent,
and by December it had climhed to lR.Q percent.
Table 1
1979
Federal funds rate
Quarterly average
M-lAl
End-of-quarter
M-lA 1/2/
Quarterly-average
M-1R 1
1980
03
04
03
Q4
01
02
10.9
13.6
15.0
12.7
9.8
15.9
R.R
4.7
4.8
-3.9
11.()
8.7
6.3
4.4
.1
2.1
14.4
o.oe
10 .1
'i. 3
-2.4
13 .s
11.4
e = estimated
1/ Percent increase at compound annual rates.
2/ End of quarter coMputed as the average of the two months surrounding
the end of the quarter.
It is natural to aqk what the economy would have looked like had
the path of one or the other of theqe instruments been smoothed.
terly econometric model can he used for this purpose.
The quar-
It can be simulated
taking either the funds rate or money growth as a policy-determined instrument.
models.
It 1s also poqqible to adrlress the question using rerluced-form
A number of rerluced-forfll. models which take money growth as the
policy-determined variahle are available.
them.
He have choqen to use two of
The first was developed by John Tatom of the St. Louis Federal
Reserve Rank.
It will hereafter he referred to aq SL-Ml.
T~e second was
developed by Byron Higgins and Vance Roley of the Kansaq City Federal
Reserve ~ank and is lahelerl KC-Ml.
The only availahle reduced-form model
which takes interest rates as the policy instrument waq also rleveloped by
Higgins and Ro]ey.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
It is referred to in this paper as KC-"t;'F.
-4SL-Ml is a rather stanrlard reducecl-form equati,on which has been
modified to take account of the relative price of energy. 1 , The SL-Ml model
is
4
4
400 MnGNP = , wiA400tnMlt-i + E vi A n Et-i
i=O
i=O
6
e
- .462 s + E ai 400 A 9,,n Pt-i + 2.59Q,
i=O
wo = .310
vo = .075
a.o
w1 = .421
v1 = .038
a.1 =
wz = .320
vz = -.023
a2 = -.031
w3 = .110
v3 = -.056
a3 = -.018
w4 = -.024
V/+
= -.043
a4 = -.060
= -.054
.039
a5 =
.025
% =
.1no
where GNP is in current dollars, Eis current-dollar federal expenditures, S
is a strike variable, pe is the relative price of energy, and Ml refers to
old M-1 prior to 195() and M-l'R thereafter.
The KC-Ml model is simpler, containing only money erowth:
4
400 tiGNPt/GNPt-J = 3.17R9 + E Bi 400 tilllt-i/Mlt-i-1
0
Bn = .52?3
83 = .430R
61 = .1868
84 = -.3304
82 = .1741
Ml in this case refers to
t,f-11~.
1/ SL-Ml is presented by John A. Tatom, "Energy Prices and Economic
Performance: 1979-81," working paper (Federal Reserve Board of St. Louis,
1')(11).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-5-
The KC-FF model isl
24
400 ~GNPt/GNPt-1 = 14.3168 +
r
Yi400*MFFt-i/RFFt-i-1
1
YI = -.0147
Yq
-.0263
Y17 = -.0233
Y2 = -.0218
YlO = -.0276
Y18 = -.0209
Y3 = -.0243
Yn = -.0288
Yl9 = -.0186
Y4 =, -.0247
Y12 = -.0295
Y20 = -.0166
Y5 = -.0243
Y13 = -.0296
Y21 = -.0151
Y6 = -.0240
Y14 = -.0290
Y22 = -.0136
Y7 = -.0243
Y15 = -.0276
Y23 = -.0115
YB = -.0251
Y16 = -.0257
Y24 = -.0076
=
RFF is the federal funds rate.
It should be noted that the functional form of the KC-FF model
is difficult to justify (its authors are fully aware of the problem).
It
appears that no standard structural model of the economy in which prices
are affected by monetary policy can be collapsed to the KC-FF reduced form.
The shortcomings of KC-FF can be seen easily by noting what happens if' for
example, the funds rate is either lowered to 1 percent and kept there or
raised to 100 percent and stabilized. Either policy eventually leads to a
growth rate of 'nominal 'GNP of -14~.•3· percent~· ·Obviously, in -a,·world •where inflation rates can change, a drastic lowering of the funds rate would eventually lead to very high nominal GNP growth rates as inflation spiraled
1/ The KC-Ml and KC-FF models can be found in Byron Higgins and Vance
Roley, "The Sensitivty of Nominal Spending to Interest Rates and Monetary
AgP,regates: Evidence from Reduced Form Equations~' (Federal Reserve Bank
of Kansas City, 1981).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-6upward, and a high enough interest rate setting would lead to the elimination of inflation and thus nominal GNP growth rates much below 14.3 percent.
Nevertheless, if prices react rather sluggishly to monetary policy, the
KC-FF model may be adequate for short-run (not more than a few quarters)
analysis.
The money demand equations in the quarterly econometric model
use end-of-quarter values to preserve certain end-of~quarter balance-sheet
identities.
However, because most discussion of monetary policy issues,is
now conducted in terms of quarterly average values, staff work done with
the model is usually based on end-of-quarter values interpolated from
quarterly averages.
While this procedure is sufficiently accurate for
most purposes, it leads to difficulties in describing simulation results
when the focus of attention is on the quarter-by-quarter paths of money
and interest rates.
Consequently, end-of-quarter money stock values are
used in this paper for the quarterly model work.
Table 2 reports the simulated consequences for the 1979Q4-1981Q4
period of eliminating the fluctuations in money growth in the year after
October 6.
The consequences were estimated using the structural quarterly
model, the SL-Ml model, and the KC-Ml model.
were done.
In each case two simulations
In the first simu~ation the relevant exogenous M-1 concept-was -
held at historical values in the 197904-1980Q3 period and at proJected
'
values thereafter.
In the second simulation money growth was held at
its average value for 1979Q4-1980Q3 and at the same projected values for
subsequent periods that were used in the first simulation.
Table 2 reports differences between the two simulations for each
model.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Smoothing the money growth path requires increasing the money growth
-7' Table 2
Simulated Effect of Stabilizing M-1, Growth, 1979Q4 to'1980Q3
1979
ci4
Ql
Q2
1980
Q3
Q4
Ql
Q2
Y
1981
~Q_3__Q-"--4_
M-1 (percent increase in level)
Quarterly model
(end-of-quarter M-lA)
.2
1.4
2.3
.o
.o
.o
.o
.o
.o
SL-Ml
(quarterly average M-lB)
.1
.o
2.0
.o
.o
.o
.o
.o
.o
KC-Ml
(quarterly average M-lB)
.1
.o
2.0
.o
.o
.o
.o
.o
.o
Nominal GNP (percent increase in level)
Quarterly model
SL-Ml
KC-Ml
.o
.o
.o
.3
.6
.7
.5
.4
.4
.3
.3
.1
.7
.8
.6
.2
-. 1
.o
1.0
.4
.3
.8
-.7
.o
.o
.o
.o
-.6
-4.1
-3.7
3.6
2.5
.6
.1
.7
.8
.s
.5
.4
.4
.3
.2
Federal funds rate (points)
Quarterly model
Real GNP (percent increase in level)
Quarterly model
.o
.2
1/ For comparison purposes the level and annual rates of growth of nominal GNP
for the 1979-Q4 period are listed here:
Nominal GNP
Level (billions of
dollars)
Percent increase
(annual rate)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Ql
Q2
Q3
2546.9
2520.8
2521.3
2586.5
10.S
10 .8
.1
10 .8
Q4
-8"J;able 3
Simulated Effect of-Holding Fu~ds Rate at Average 1979Q4 to 1980Q3
'
Levels'during That Period
1979
Q4
Ql
Q2
1980
Q3
!/
1981
Q4
Ql
g2
Q3
Q4
Federal funds rate (Eoints)
,.
.
Quarterly model
-.8
-2 .2
.1
3.0
0
0
0
0
0
KC-FF
-.8
-2.2
.1
3.0
0
NR
NR
NR
NR
Nominal GNP (percent increase in level)
Quarterly model
.o
.2
.3
.2
.1
.2
.2
.2
.3
KC-FF
.o
.1
.4
.6
.4
NR
NR
NR
NR
.8
.4
-.6
.o
.o
.1
.2
.3
.2
.1
.1
.2
.2
.2
.2
M-1 (percent increase in level)
Quarterly model
.2
Real GNP (percent increase in level)
Quarterly model
.o
.1
N.R. Not reported. See discussion of KC-FF model, p.5, and footnote 1, p. 9.
1/ For historical GNP levels, see footnote 1, Table 2.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-- -
- -
-9rates in each of the first three quarters of the period and reducing it
sharply in the last quarter.
On average the level of money is higher.
The
smooth money path does appear to diminish the output fluctuations that
occurred, but the various models agree that the impact on nominal output
would be relatively small compared to the size of the recent recession.
The quarterly model channels from money and interest rates to output contain relatively long lags so the reaction of simulated output is slow and
fairly small.
The monetarist reduced-form models show a somewhat greater
tendency for the smoothed money path to smooth GNP.
In the SL-Ml case,
output growth in 1980Q2 is about 2-1/2 percent greater at annual rates; in
the KC-Ml case it is about 4 percent stronger.
Table 3 reports the simulated effects, using both the quarterly
and the KC-FF models, of smoothing the path of the federal funds rate by
holding it at its historical 1979Q4-1980Q3 average over that period.
This requires holding the federal funds rate down early in the period, and
increasing it in the later part.
The changes in the funds rate needed to
accomplish the smoothing (Table 3) are on average smaller than those simulated by the quarterly model when the money growth path was smoothed (Table
2).
As a result the simulated effects on GNP are smaller.
Like the monetarist reduced-form models, the KC-FF model (which
is certainly not monetarist) shows a somewhat larger change than does the
quarterly model as a result of the smoothing.1
Once again, however, both
1/ Smoothing the interest rate path causes the KC-FF model to produce
higher output in 1981. This appears to be a result of the fact that the
KC-FF model works off percentage changes in the funds rate. A volatile
rate path needs bigger percentage increases than decreases on average if
it is to be trendless. It seemed doubtful that this characteristic of the
model corresponded to the true structure of the economy. Consequently,
KC-FF results for 1981 were not reported.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-10models show simulated effects which are relatively s~all compared to the
size of the 1980 recession.
These results should be accepted only with great caution.
Eco-
nometric relationships are the result-of theory and of average historical
relationships.
Average relationships do not hold at each point in time.
Some argue that behavioral relationships (as they are usually measured econometrically) vary with the public's perception of economic policy rules.
It may be that if the-alternative money growth or funds rate paths examined
above had been followed, public perception of policy would have been very
different.
A substantial number of observers appear to believe that a
policy like the steady funds rate policy would have resulted in higher inflation expectations in this period and that the downturn in production
which occurred in early 1980 might not have taken place.
turn, would have been worse.
Inflation, in
Whether these views are correct is beyond the
scope of this paper.
Even if one r~jects the caveat of the previous paragraph, one
should not conclude from the simulations presented in Tables 2 and 3 that
all temporary deviations in policy instruments from planned paths have
negligible effects on output.
In the exercises performed above the levels
of the policy instrument variables differ from quarter to quarter, but on
average over the one-year period they are the same.
If a policy instrument
variable were to deviate from its planned path, and if its return to the
planned path were to be delayed or not compensated for by an offsetting
deviation on the other side, the effect might be considerably larger.
possibility is the subject of the next section.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
That
-11Section III:
Policy Instrument Deviations and the Return to Targeted Paths
In recent years the Federal Open Market Committee has set longrun target ranges for monetary aggregates.
For a variety of reasons, the
aggregates do not grow smoothly along the target paths.
It will be argued
in the next section that the consequences of any such instrument deviation
depend on the deviation's causes.
In this section, however, we will assume
that the deviation is deliberate (at least in the sense that the FOMC or the
Desk acquiesced in the deviation).
long-run target path for money.
For example, suppose the FOMC adopted some
The FOMC would have some expectations about
the path of interest rates, prices, and output.
Further suppose a situation
arose which caused the Committee to wish temporarily to miss its targets.
The
size of the effects would depend on the ma~nituae and duration of the policy
variable deviations.
This section explores the consequences of different re-
entry paths for the policy instrument variables.
It should be stressed that the instrument deviations analyzed in
this section are not the usual sort which are caused by either the level of
economic activity of money demand being stronger or weaker than expected.
That subJect will be addressed in the next section.
The deviations studied
here are ones in which the Committee or the Desk have deliberately allowed
the instrument to stray from its long-run target path. 1
Table 4 presents the results of a set of simulation experiments
in which the money stock was caused to rise 1 percent above target in the
1/
There is one realistic type of unintentional deviation to which the
analysis of this section applies. If the Committee has a long-run aggregate
target which it attempts to achieve using a bank reserves instrument, unexpected values of the reserves/deposit multiplier could cause deviations to
which this section is applicable.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-12first quarter and then returned to the target path along a variety of traJectories.
In the first trajectory examined, the money stock is returnerl all the
way to the target path in the second quarter.
The quarterly model shows
little effect on nominal output of the deviation in the initial quarter, but
the effect builds to about 0.2 percent after four quarters, then starts to
decline.
The monetarist reduced-form equations show a sharper initial effect.
The SL-Ml model gives a 0.29 percent response in the initial quarter.
The
effect builds to 0.4 percent in the second quarter, then falls rapidly to
zero.
The KC-Ml model shows an even sharper initial effect.
Output is about
1/2 percent higher in the first quarter in this model, then returns by an
erratic route to very near the undisturbed path in the fifth period.
Table
4 also shows simulated effects on the federal funds rate taken from the
quarterly model.
The second and third traiectories investigate the simulated effects
of more persistence in the overshoot on money.
In the second trajectory the
overshoot is not eliminaterl until the third quarter.
the overshoot lasts three full quarters.
In the third trajectory
As one would expect, increasing the
persistence of the overshoot increases the size and prolongs the duration of
the simulated effects on nominal GNP.
In the case of the three-quarter over-
shoot (traiectory 3) the quarterly model estimate of the maximum effect is
about 0.6 percent after five quarters.
The monetarist reduced forms show a
percentage increase in nominal output nearly as larP,e as the percenta~e
increase in money by the third overshoot quarter.
The fourth and fifth tra1ectories examine the effects of returning
money gradually to the target path after the initLal overshoot.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
As expected,
-13Table 4
Effect of Temporary neviations of M-1 from Planned Path
2
3
4
5
6
7
0
0
0
0
0
0
.19
. 30
. 17
.21
.16
. 40
. 18
• 11
.oo
.oo
-1.8
.5
M-1 Deviation (percentage
points from desired level)..!/1.0
1. 0
1
(1)
M-1 Deviation (percentage
points from desired level)_!/1.0
8
-,
- O·
Percentage Effect on Level of Nominal GNP
Quarterly Model
SL-Ml
KC-Ml
Effect on Funds Rate
(points, ()uarterly
Model)
( 2)
. 08
. 29
.49
• 18
.21
.oo
.45
• 21
-.02
-.30
.14
.01
.01
.01
.5
.4
.2
.4
.2
.2
0
0
0
0
0
0
.35
. 70
.34
.3Q
. 41
.57
.40
. 08
.10
.38
-.02
-.30
.oo
.oo
.01
.01
Percentage Effect on Level of Nor111nal GNP
Quarterly Model
SL-1'11
KC-Ml
Effect on Funds Rate
(points, Quarterly
l-fodel)
(3)
. 08
. 29
.49
. 25
. 69
.67
.32
• 24
-1.8
-1.3
1.0
.9
.5
.4
.5
.4
Deviation (percentage
points from desired leveJ.J../ 1. 0
1. 0
1.0
0
0
0
0
0
.57
.80
.74
.58
. 38
.57
• '50
• 08
.oo
• 26
• 10
-.02
-.30
.6
.7
}1-l
Percentage Effect on Level of tTominal GNP
Quarterly Model
SL-Ml
KC-Ml
Effect on Funds Rate
(points, ()uarterly
Model)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
• 08
. 29
. 4g
-1.8
• 25
• 69
.67
-1.3
.44
. 99
.83
-.8
1.5
1. 1
.8
.40
.01
-14Tabl'e 4 (cont'd)
Effect of Temporary Deviations of M-1' from Planned Pa,tb
1
(4)
M-1 Deviation.(Eercentage
points from desired level).:!/'1.0
2
3
4,
5
6
7
8
.67
.33'
0
0
0
0
0
.32
.66
.44
.41
.44
.57
.39
.15
.02
.38
.02
-.07
-.09
.oo
.oo
.2
.9
.6
.s
.4
.4
0
0
0
0
0
0
.27
.50
.25
.30
.26
.49
.30
.02
-.10
.29
-.01
-.15
.oo
.oo
.oo
.oo
Percentage Effect on Level of Nominal GNP
Quarterly Model
SL-Ml
KC-Ml
Effect on Funds Rate
(points, Quarterly
Model)
(S)
.08
.29
.49
.23
.59
.so
-1.8
-.7
,,
M-1 Deviation (percentage
Eoints from desired level).!/1.0
.5
.32
.oo
.2s
Percentage Effect on Level of Nominal GNP
Quarterly Model
SL-Ml
KC-Ml
Effect on Funds Rate
(points, Quarterly
Model)
(6)
.08
_zq
.49
.21
.54
.42
.23
.17
-1.8
-.4
.8
.6
.4
.4
.4
.4
M-1 Deviation (Eercentage
points from desired level)l/1.0
-.5
.o
.o
.o
.o
.o
.o
.14
.25
-.06
. 1()
.10
.09
.13
-.04
.33
.11
-.07
-.49
.11
.02
.16
.oo
.oo
.01
.01
1.4
.2
.1
.o
.2
.2
.2
Percentage Effect on Level of Nominal GNP
Quarterly Model
SL-1'11
!ZC-Ml
Effect on Funds Rate
(points, Quarterly
Model)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
.08
.zg
.49
-1.8
.07
.04
-15Table 4 (cont'd)
Effect of Temporary Deviations of M-1 from Planned Path
1
(7)
M-1 Deviation (Eercentage
Eoints from desired level).!/1.0
3
4
5
6
7
8
0
0
0
0
0
0
.10
.10
-.30
.02
-.09
.oo
.oo
.03
-.19
.26
-.12
-.68
.01
.03
.33
-.03
.01
.02
-.03
.01
2.4
.o
-.1
-.2
.1
.o
.o
2
-LO
Percentage Effect on Level of Nominal GNP
Quarterly Model
SL-Ml
KC-Ml
Effect on Funds Rate
(points, Quarterly
Model)
!/
.08
.29
.49
-1.8
.02
Deviations are for end-of-quarter M-lA in the case of the Quarterly Model, and for
quarterly average M-lB in the cases of the SL-Ml and KC-Ml models.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-16the effects on output are in both cases greater than when the overshoot was
corrected promptly and smaller than when it was allowed to oersist for more
than two quarters.
All the above trajectories lead to output being higher on average
than would have been the case if money had remained on the target path.
The
simulated effects on output can be reduced by balancing the initial money
overshoot with an offsetting later undershoot.
In traJectory number 6, the
1 percent excess of money in the initial quarter is followed by a 1/2 percent
deviation in the opposite direction in the succeeding quarter.
This largely
eliminates significant output responses in the quarterly and SL-Ml models,
but at the cost (according to the quarterly model) of a rather choppy interest
rate path.
The uneven lag structure in the KC-Ml model leads to a somewhat
irregular output response.
Trajectory 7 shows the simulated effect of creating an instrument deviation equal and opposite in sign in the second quarter to the deviation in the first.
This damps even further the output responses of the
quarterly and SL-Ml models.
The KC-Ml response is now still more erratic.
Table 5 reports the results of similar experiments using the federal
funds rate as the policy instrument.
model only.1
Results are reported for the quarterly
In these simulations it is assumed that the VOMC has set a long-
run target path for the funds rate and then for some reason allows a temporary
deviation.
The deviation amounts to 300 basis points in the initial quarter.
Five trajectories are then examine<l which take the funds rate back to target
levels by different paths.
1/ The ~C-VF model was simulated but the results were not reported in
view of the problem described on page 5 and 10 the footnote on page 9.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-17Table 5
Effect of Temporary Deviations of Federal Funds Rate From Planned Path
Quarter
(1)
1
2
3
4
5
6
7
8
3.0 '
.o
.o
.o
.()
.()
.o
.o
-.1
-.3
-.5
-.6
-.7
-.8
-.9
-1.0
-1.8
-.5
-.6
-.6
-.5
-.6
-.6
-.7
3.0
3.0
0
0
0
0
0-
0
-.1
-.5
-.8
-1.1
-1.3
-1.5
-1.7
-1.8
-1.8
-2.2
-1.1
-1.1
-1.0
-1.1
-1.2
-1.3
3.0
3.0
3.0
0
0
Quarterly Model _
-.1
-.5
'-.9
-1.4
-1.8
-2.2 - --2.5
'
-2.7
Percentage Effect
on Level of M-lA
(Quarterly Model)
-1.8
-2.2
-2.7
-1.7
-1.6
-1.6
-1.9
Funds Rate Deviation
From Desired Level
(points)
Percentage Effect
on Level of
Nominal GNP
Quarterly Model
Percentage Effect
on Level of M-lA
(Quarterly Model)
(2)
Funds Rate Deviation
-From Desired Level
(points)
Percentage Effect
on Level of
Nominal GNP
Quarterly Model
Percentage Effect
on Level of M-lA
(Quarterly Model)
(3)
Funds Rate Deviation
From Desired Level
(points)
0
0
0
Percentage Effect
on Level of
Nominal GNP
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-1.7
-18Table 5 (cont'd)
Effect of Temporary Deviations of Federal Funds Rate From Planned Path
'
Quarter
(4)
Funds Rate Deviation
From Desired Level
(points)
- . -- --
1
2
3
4
-
5
6
7
8
3.0
2.0
1.0
.o
.o
.o
.o
.o
-.1
-.4
-.7
-LO
-L3
-L5
-L7
-L8
-L8
-L6
-L4
-Ll
-LO
-Ll
-L2
-L3
3.0
L5
.o
.o
.o
.o
.o
.o
-.1
-.4
-.6
-.8
-LO
-L2
-L3
-L4
-L8
-L3
-.8
-.8
-.7
-.8
-.9
-LO
3.0
-LS
.o
.o
.o
.o
.o
.o
-.1
-.3
-.3
-.4
-.4
-.5
-. 5
-.5
-L8
-.2
-.3
-.3
-.2
-.4
-.4
-.4
Percentage Effect
on Level of
Nominal GNP
Quarterly Model
Percentage Effect
on Level of M-lA
(Quarterly Model)
(5)
Funds Rate Deviation
From Desired Level
(points)
Percentage Effect
on Level of
Nominal GNP
Quarterly Model
Percentage Effect
on Level of M-lA
(Quarterly Model)
(6)
Funds Rate Deviation
From Desired Level
(points)
Percentage Effect
on Level of
Nominal GNP
Quarterly !fodel
Percentage Effect
on Level of M-lA
(Quarterly Model)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-19-
Table 5 (cont'd)
Effect of Temporary Deviations of Federal Funds Rate From Planned Path
1
(7)
Funds Rate Deviation
From Desired Level
(points)
2
3
4
Quarter
5
6
7
8
3.0
-3.0
.o
.o
.o
.o
.o
.o
-.1
-.2
-.1
,-. 2
-.1
-.1
-.1
-.1
-1.R
1.2
-.1
.I)
.1
-.1
-.1
-.1
Percentage Effer.t
on r.evel of
Nominal GNP
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Quarterly Model
Percentage Effect
on Level of M-lA
(Quarterly MoQel)
---
-20The first trajectory returns the funds rate to the target path in
the second quarter of the simulation.
this is a rather serious deviation.
The quarterly model indicates that
Nominal output is lowered by about 1/2
percent after one year and nearly a full 1 percent after two years.
This is
in some contrast to the results for the one-period deviation from money
targets shown in the first panel of Table 4.
difference.
There are two reasons for'the
The first is merely a matter of scale--the initial deviation
being examined is nearly twice as big.
The deviation of 300 basis points on
the interest rates initially cause~ money to deviate 1.8 percent from its
expected path, whereas in Table 4 money stock deviations of 1 percent were
being analyzed.
The second reason, however, is fundamental.
est rate is raised, output is reduced.
When the inter-
When the rate is then returned to
the long-run path, output remains lower and consequently so does money.
The
consequences for the money stock (as simulated by the quarterly model) can
be seen in the last line of the first panel of Table 5.
all the way to its undisturbed value.
It does not return
Thus the interest rate traJectory
analyzed here is the equivalent of a more persistent deviation of money from
its target path than anything presented in Table 4.
This suggests that
"misses" from an interest rate target path must be eliminated more quickly,
or be more completely compensated for by making offsetting "misses" on the
other side, than is necessary for "misses" from a l'lOney stock target.
Trajectories 2 and 3 investigate the effects of more persistent
deviations of interest rates from target paths.
The effects on nominal out-
put are roughly proportional to the number of quarters the deviation is
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-21allowed to persist.
Trajectories 4 and 5 simulate gradual returns to
target after the initial miss.
TraJectories 6 and 7 examine the possi-
bilities of creating offsetting deviations from the funds rate target.
Only in the case of an equal offsetting deviation (trajectory 7) is the
effect on nominal output largely eliminated.
Section IV:
Policy Responses to Economic Disturbances
With perfect knowledge of the working of the economy the conduct
of monetary policy would be relatively simple.
The range of feasible
outcomes for ultimate targets such as output, employment, and inflation
would be known; and it would be possible to achieve a desired path for any
particular single target variable (such as a time path for output or inflation), or to attain the most desirab1e of the many feasible combinations
of ultimate targets (such as output and inflation).
Unfortunately, knowledge of the economic system is far from perfect.
While the main determinants of the value of each economic variable
may be known, there is a remaining unknown component.
Economists may say
that real consumption depends on current and recent real disposable income,
for example, or that the quantity of money demanded depends on income, the
price level, and interest rates.
But while these variables may explain most
of the variation in consumption and money holdings, they will not explain
all of it.
The remaining variation is due to other causes, and is referred
to as the uncertainty in economic relationships.
Policymakers must attempt
\
to achieve their goals in the presence of this uncertainty.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
\
\
'\
\
\'
'
'
-22Through the use of simulations it will be demonstrated below that
the consequences of a surge in the unpredictable component of any relationship will depend on which relationship suffered the disturbance, and on the
reaction of the monetary policymakers to that disturbance once it happens.I
We will demonstrate that a particular policy response pattern which has
relatively favorable consequences in the face of one type of disturbance
may have unfavorable consequences in the face of a different disturbance.
Since past experience suggests that the uncertainty component of some relationships is larger than for others it follows that some policy reaction
rules will on average work better than others.
In the material that fol-
lows, various simple monetary policy reaction rules will be hypothesized.
Then, given an empirically based structural representation of the economy
(i.e., the quarterly model) the consequences of various economic disturbances will be simulated, tabulated and discussed.
The Policy Rules
Three simple policy rules are considered in this paper.
Under
the first rule it is assumed that a path for the Federal funds rate is
chosen, and that the path for this instrument is adhered to regardless of
any events which may later occur but which were not expected at the time
the funds rate path was chosen.
Obviously this policy rule is something
of a straw man.
1/ Readers familiar with the literature in this area will recognize
that the analysis developed here is descended from the work published by
William Poole in "Optimal Choice of Monetary Policy Instruments in a Simple
Stochastic Macro Model," Quarterly Journal of Economics, vol. 84 (May
1970), pp.197-216.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-23Under the second rule, the central bank is assumed to set and
achieve a path for M-1 regardless of subsequent events.
This policy rule is
considerably more realistic than the first and in fact has many advocates.
The third policy rule is a combination of the first and second.
Policymakers are assumed to have a target path both for the funds rate and
for M-1.
The interest rate path is adjusted upward or downward whenever
money exceeds or falls short of its target path.
Specifically, this rule is
tnRFFt = inRFFt-1 + A tn(M/Mtarget) + 8 ln RFFtarget,
where ~FF is the funds rate and Mis M-lA.
As A approaches infinity, this policy rule approaches a money
fixed rule (rule 2).
As A approaches zero it becomes a fixed path for
interest rates (rule 1).
In the simulations reported below A= 15 so that
a 1 percent deviation in money from its target value causes the funds rate
to be increased by 15 percent (say from 10 to 11.5).
The Economic Shocks
Four different shocks were tested under each policy rule.
1)
Real consumption was depressed by 1 percent.
Specifically,
this means that the consumption function of the model was altered so that
whatever the values of the usual determinants of consumption, the consumption function produced 1 percent less consumption that it normally
would given those determinants.
2)
cent.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The demand for demand deposits was shocked upward by 2 per-
-243)
The level of wages was depressed by 1 percent.
4)
The level of prices was depressed by 1 percent.
In each case the shift in the relationship was assumed to be
permanent.
Table 6 shows the simulated result of lowering consumption by
1 percent relative to its usual determinants.
This weakens real GNP, of
course, although that may not be immediately observable.
The monetary
policymakers are first likely to observe a tendency for short-term interest
rates and money to fall below their expected levels.
What happens next
depends on how the policymakers react.
If the policy reaction is to keep the federal funds rate on its
predetermined path, the simulated results are those given by the top line
in each panel of Table 6.
The funds rate is, of course, unchanged.
Real
GNP and nominal money balances both fall steadily further from the values
they otherwise would have had.
In the case of this particular shock, a
fixed interest rate path turns out to be the worst policy examined.
Holding nominal money balances at their predetermined levels works
out much more satisfactorily.
in output.
The funds rate falls and this resists the fall
These first two simulations support the usual contention that
monetary aggregates are a superior instrument to interest rates in the
presence of shocks to aggregate demand.
The last policy examined assumes the policymakers change interest
rates in response to deviations of money from the target path, but not by
enough to make up all the discrepancy.
This policy turns out to be less
stabilizing than the second one but more stabilizing than the first.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-25Table 6
Simulated Effect of Sustained 1 Percent Reduction in
Consumption Relative to its Determinants
Policy
reaction
function
1
2
Quarter after shock
3
4
12
8
Effect on GNP (billions of 1972 dollars)
(1)
Funds rate fixed
-9.6
-18.0
-n.3
-25.5
-36.8
-43.0
(2)
M-1 fixed
-9.4
-17.1
-18.9
-20.5
-20.1
-7.8
(3)
Money target
-9.6
-17.7
-20.2
-23.0
-23.8
-10.5
Effect on M-lA (billions of current dollars)
(1)
Funds rate fixed
(2)
M-1 fixed
(3)
Money target
-.3
-1.2
-2.6
0
0
0
-.2
-.7
-3.9
-9.3
-16.7
0
0
0
-1.2
-1.4
-.9
-.7
Effect on federal funds rate (percentage points)
(1)
Funds rate fixed
(2)
(3)
y
0
0
0
0
M-1 fixed
-.4
-.8
-1.1
-1.3
-1.7
-3.0
Money target
-.1
-.4
-.7
-LO
-1.8
-3.2
0
0
Contro-1 simulation values of the variables whose responses are tablulated above
are as follows:
Quarter
1
GNP (1972 dollars)
M-lA
Funds rate
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
1220.0
285.7
6.2
2
1227.9
288.6
5.4
3
4
8
12
1259 .5
293.3
4.8
1267.4
296.2
5.2
1331.2
317.0
5.2
1395.2
342.9
7.3
-26Table 7 shows the results of the same experiment for a permanent
upward shift in the demand for demand deposits.
Unlike the case of the
consumption shock, the first' observable consequences of this shift are
likely to be a tendency for interest rates and money balances to rise above
expectations, although,as it is with the consumption shock, the effect on
aggregate spending is downward.
Policy 1--the fixed path for the funds rate--works quite well in
this case while, as expected, the fixed M-1 path works poorly.
Policy 3,
like the money fixed policy, does not perform well because it reacts to a
change in money demand that should be accommodated.
Table 8 reports the results of a sustained downward shift in
prices and profit margins.
ways in the short run.
money stock.
This shift increases simulated real output two
First, the reduction in prices increases the real
Second, the reduction in profit margins shifts real income
from businesses to wage earners.
ginal propensity to spend.
Businesses have a lower short-run mar-
The first effect the monetary policymakers
would observe is a tendency for interest rates and money balances to fall
below expectations.
The proper response is less clear in this case than it was in the
case of the shocks to consumption and deposit demand.
Presumably, the
authorities would, if they had full information about what was happening,
prefer to keep output relatively unaffected and accept the reduction in the
price level.
This could be accomplished by raising the funds rate for a
period of time.
The policies which emphasize aggregates (2 and 3) tend to reduce
the funds rate, which is inappropriate.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Holding to the planned funds rate
-27Table 7
Simulated Effect of Sustained 2 Percent Upward Shift
in the Demand for Demand Deposits l/
Policy
reaction
function
1
2
Ouarter after shock
3
4
12
8
Effect on GNP (billions of 1972 dollars)
(1)
Funds rate fixed
(2)
M-1 fixed
-.o
-.o
-.o
-.o
-.2
-.6
,-.9
-2.6
-4.9
-7.4
-15.~
-17.R
-.4
-.8
-2.4
-4.1
-13.4
-18.4
~
(3)
Money target
-.
Effect on M-lA (billions of current dollars)
(1)
Funds rate fixed
(2)
M-1 fixed
(3)
Money target
2.2
4.3
4.4
4.4
4.6
4.6
0
0
0
0
0
0
1.4
2.2
1.1
-.7
-.9
.3
Effect on federal funds rate (percentage points)
(1)
Funds rate fixed
(2)
M-1 fixed
(3)
Money target
)j
0
0
0
0
3.1
1.9
1.2
1.0
.3
-.7
.9
1.4
1.4
1.5
.8
-.2
0
0
Control simulation values of the variables whose responses are tahlulated above
are as follows:
Quarter
1
GNP (1972 dollars)
M-lA
Funds rate
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
2
3
1220.0
285.7
1227.9
288.6
6.2
5.4
1259. 5
293.3
4.R
4
1267.4
296.2
5.2
8
12
1331.2
13Q5.2
317.0
5.2
342.9
7.3
-28Table 8
Simulated Effect of Reducing Prices 1 Percent
by Reducing Markups over Costs 1/
Policy
reaction
function
2
1
Quarter after shock
3
4
8
12
Effect on GNP (billions of 1972 dollars)
(1)
Funds rate fixed
2.1
3.3
3.3
J.Q
8.5
15.6
(2)
M-1 fixed
2.6
4.6
5.7
7.4
15.5
20.5
(3)
Money target
2.4
4.0
4.8
6.5
15.7
22.5
Effect on M-lA (billions of current dollars)
(1)
Funds rate fixed
(2)
M-1 fixed
(3)
Money target
-.8
-1.5
0
0
-.5
-.8
-1.6
-2.0
-1.8
0
0
0
0
- .1.
-.1
.4
.7
-1.5
Effect on federal funds rate (percentage points)
0
0
0
0
0
M-1 fixed
-.1
-.6
-.4
-.3
-.o
.a
Money target
-.3
-.4
-.4
-. 5
-.2
.6
(1)
Funds rate fixed
(2)
(3)
1/
0
Control simulation values of the variable_s___whose responses are tabluldted above
are as follows:
Quarter
GNP (1977. dollars)
M-lA
Funds rate
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
1
2
3
4
8
12
1220.0
285.7
1227.9
288.lj
5.4
1259.5
1267.4
2'lJ .3
2%.2
4.8
5.2
1331.2
317.0
5.2
1395.2
342.9
7.3
6.2
-29path 1 works better.
None of the strategies considered makes the appro-
priate response.
Table 9 presents the simulated result of shifting wages down by
1 percent.
This shift has two major short-run effects.
wages lowers prices and increases real money balances.
is shifted temporarily from wage earners to businesses.
First, lowering
Second, real income
In the short run
this depresses output.
Once again, a value judgment is needed in determining the appropriate response.
Policymakers might wish to resist the initial fall in
output, but it seems unlikely that they would wish to take output above
its undisturbed levels and thus dissipate the gain the shock produced in
fighting inflation.
The aggregates policies (2 and 3) resist the initial decline in
output by reducing interest rates.
If allowed to stay in effect, however,
they would eventually increase output enough to return prices to their undisturbed levels, which seems undesirable.
The funds rate policy, 1, does
not resist the decline in output, but has the desirable property that it
does not reflate.
Thus again, none of the simple response rules tested
works very well.
A more satisfactory response rule would be more complex
and would take into account knowledge about which relationship had suffered
the disturbance and the sign and magnitude of the disturbance as quickly
as that knowledge became available.
Conclusions
This section has looked at simulations of the effects of a number
of possible shocks to the economic system under a number of different mone-
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-30Table 9
Simulated Effect of Reducing Wages by 1 Percent
Policy
reaction
function
Quarter after shock
1
3
2
4,
12
8
Effect on GNP (billions of 1972 dollars)
(1)
Funds rate fixed '
-1.4
-,2. 7
-3.3
-{.,.2
(2)
M-1 fixed
-1.2
-1.7
-1.0
(3)
Money target
-1.3
-2.5
-2.3
-,4.0
-.1
-7.2
"
5.7
-1.8
3.8
15.9
,
I
._. I
15.5
Effect on M-lA (billions of current ' dollars)
(1)
Funds rate fixed
(2)
M-1 fixed
(3)
Money target
-.5
-1.3
-2.0
-2.6
-s.1
-8.0
0
0
0
0
0
0
-.6
-.8
-.8
-.7
-.1
.5
Effect on federal funds rate (percentage points)
(1)
Funds rate fixed
0
0
0
0
0
0
(2)
M-1 fixed
-.s
-.7
-.7
-.8
-.8
-.8
(3)
Money target
-.2
-.4
-.s
-.7
-1.0
-1.1
1/ vControl simulation values of the variables whose responses are tablulated above
are as follows:
Quarter
GNP (1972 dollars)
M-lA
Funds rate
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
1
2
1220.0
285.7
6.2
1227.q
288.6
5.4
3
1259.5
293 .3
4.8
4
1267.4
296.2
5.2
8
1331.2
317.0
5.2
12
1395.2
342.9
7.3
~
-31tary policy response rules.
It has found that some of the policy responses
are better for one type of shock, other responses better for another type.
None of the policies examined is clearly dominant.
The material provided
here suggests that a good policy rule--which would necessarily be much
more complex than any considered here--would take into account both the
relative likelihood of the variou~ kinds of shocks in the period before the
nature of the shock could be determined, and the nature and magnitude
of the shock once the determination could be made.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
CYCLF.<; RESULTING FROM MONEY STOCK TARGETit-IG
Paper Written for a Federal Reserve
Staff Review of Monetary Control
Procedures
by
Jared Enzler and Lewis Johnson
The assistance of Nancy Artz is gratefully acknowled~ed.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
CYC:LES RESULTI~G FR.OM MONEY STOCK TARGETPTG
Section I.
Introduction
This paper examines the possibility that money stock targeting may
introduce a cyclical mechanism into the economy._!_/
The hypothetical mechanism
contains the following elements:
1.
A standard IS-LM macroeconomic system in which economic
activity responds to interest rates, but only with a la~,
2.
A money demand function with a low interest elasticity
of demand for money and a substantial ~ncume elasticity,
3.
A monetary policy which consists of raising or lowering
interest rates whenever money exceeds or falls short of
some target path.
It is easy to see that this set of assuMptions could lead to a
cyclical mechanism.
Suppose some disturbance were to cause aggregate output
to fall below levels the monetary policymaker had expected.
Because the
demand for money responds to falling income, money balances would fall below
target.
The postulaterl policy response would be a reduction in interest
rates.
Eecause of a low interest elasticity of money demand, anrl lags in
the effect of interest rates on income, this policy would initially have
negligible impact on money balances, leading the policymaker to further reduce
interest rates.
Eventually, the initial reduction in interest rates would
return output to its original level, which would return money balances to the
1/ The possibility was suggested by Governor Gramley, and this investigation
was originally undertaken at his request.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 2 -
target level.
Subsequently, the later interest rate reductions would cause
income to rise and money balances to exceed target.
As will be seen, the
resulting cycle could he either damped or undamped.
To examine the nature of this cyclical process, a small macro model
was created and sinulated.
The sensitivity of the amplitide and periodicity
of the cycle to variations in parameters was examined to shed light on three
issues:
(1) Would plausible values of the system's parameters result in
cyclical responses to disturbances under a monetary policy of the sort
described in assumption 3?
(2) Could this mechanism be invoked to explain
the pattern of aggregate output in the United States in 1979-80?
(3)
Could
alternative monetary policies be formulated which would eliminate or greatly
attenuate the cycle for sets of parameter values which lead to cycles?
In Section II the simulation model 1s presented.
It 1s a simple
dynamic IS-LM model extended by the addition of an expectations-augmented
Phillips curve and a policy function.
It should be noted that this model
contains relatively sinple adaptive expectations, explicitly for goods prices
and implicitly for interest rates.
The incorporation, instead, of the widely
discussed "rational" expectations mechanism, would certainly alter and probably eliminate cyclical characteristics from the model.
For a particular set of parameter values, the model presented 1s
shown to have cyclical responses to exogenous disturbances.
It exhibits (at
least) two basic cycles -- a short cycle (of 3 to 5 years) resulting from
the interaction of the hypothetical monetary policy with the IS and LM curves
and a long cycle (more than 15 years) resulting from the interaction of
monetary policy with the Phillips curve.
In Section III the nature of the short cycle is explored in isolation
by "disconnecting'' the model I s Phillips cttrvE!'.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The SenHtivity of tl'f'e snort
- 3 -
cycle's duration and degree of damping to changes in behavioral parameters
and policy reponses is examined.
In Section IV the performance (both short run and long run) of
various policies is considered with the Phillips curve re-activated.
Policies
based on final targets (income and inflation) are found to be reasonably
effective in damping the cycle.
Results are summarized and conclusions drawn in Section V.
It is
concluded that while plausible parameters do indeed give rise to cycles, the
parameters needed to ~enerate a cycle as short as the one observed between
late 1979 and late 1980 are not plausible.
II.
The Model
The analysis will be of a simple dynamic IS-LM model, generalized
by the addition of a Phillips curve and a policy function.
The model is
fully specified by equations (1) through (4):
n
+ a 1 lny_1 - 1.ai(r - Y)_i + e y
1
p
(1)
lny
=
Ct
(2)
lnM
=
B0 + B1lnM-1 + Bzlny - B3lnr + R4lnP + eM
(3)
P/P
0
m
(4)
p
+ Un(y/y*) + ep
lnr_l + Y1ln(M/M*) + YzMn(M/M*)
lnr
=
y
y*
= real output
= "capacity" output
where
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
r.ci (f)_i
1
r
(corresponding to the "natural"
unemployment rate)
= the "short" nominal interest rate
- 4 -
P
= the
= the
= the
= the
price level
inflation rate
M
money stock
M*
target monev stock
ey, em, ep are disturbances
h indicates a first difference
P/P
Equation (1) represents an adjustment process for the goods markets
(it is a dynamic analogue of the IS curve).
The aggregate demand for goods
presumably depends upon income and the "long" real rate of interest -
assumed
to be generated by a distributed lag on short nominal irtterest rates less oneperiod inflation rates.
This equation may be interpreted as postulating that
the current rate of growth of output will be proportional to the preceding
period's excess demand for goods.
possible.)
equation.
(An equilibrium interpretation is also
The coefficients ai are based on estimates of a somewhat different
Simulation of the MPS quarterly model reveals implicit coefficients
roughly comparable to those used here.
The coefficients a 0 and a1 were
arbitrarily chosen to yield a plausible steady-state real interest rate of
about 1.2 percent.
Equation (2) is a conventional money demand equation, with desired
real balances (real because the long-run price elasticity is assumed to be 1)
depending upon real income and the short nominal interest rate.
Actual money
holdings are adjusted to desired levels by a partial-anjustment mechanism.
The long-run real income elasticity of money demand is assumed to be 0.75.
The remaining parameters of this equation will be varied over plausible
ranges subject to the constraint that in the lon~ run the demand for nominal
money is always proportional to prices.
equation (3) is analogous to an expectations-augmented Phillips
curve.
The distributed lag on past inflation rates represents the expected
rate of inflation.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The ci are geo~etrically declining and sum to 1.0.
(The
- 5 -
long-run Phillips curve 1s vertical.)
Excess rlemand is measured by the ratio
of output to capacity output and it~ coefficient, A, has been set at 15
broadly consistent with typical estimates of the responsiveness of the ~hillips
curve.
In this model expectations of both future short-term interest rates
(equation 1) and future inflation rates (equation 3) have been representen as
distributed lags on past rates -- expectations are assumed to be formed adaptively.
In this respect, our model is representative of all maJor econometric
forecasting models.
A currently fashionable alternative is to assume that
expectations are formed in such a way as to be consistent with the model's
own predJctions.
The behavior of the model would likely be dramatically
different under this "rational" expectations alternative.
Finally, equation (4) specifies the policy followed by the Federal
Reserve.
This particular specification would have the Fed attempt to achieve
a target money growth path by raising the short-term interest rate when money
is above target and lowering the rate when money is below target.
Alternative
policies will also be considered.
Ranges of parameter values used in simulations of this model are
shown in Table 1.
It should be noted that in equations (1) and (2) the "dependent"
variable appears lagged once on the right-hand side of the equation.
This
means that if an explanatory variable is chan~ed permanently the effect of
that change will cumulate over time.
For example, if income were permanently
increased by 1 percent in equation (2), the demand for money will be increased
by 82 percent in the first quarter (8z is sometimes called an impact
elasticity).
In the second quarter money demand will be greater not only
because income is higher but also because money demand was greater by~~
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 6 -
in the preceding quarter.
Accordingly, money demand will be inLreased by
(B2 + B182) perLent in the second quarter after the change.
The cumula-
tive effect after income has been at its higher level for a very long time
approaches [82/(l-B1)J -- referred to as a long-run elagticity.
It is
ranges for long-run elasticities that are shown in Table 1.
Table 1
Ranges Considered for Parameters of Simulation Model
IS curve:
. 7*
fa
.029*
i
Money demand·
81
.25
8
2
(l-A1)
.7*
B
3
(l-B1)
.02
to
.058
to
.S*
to
.1()*
,~
R
(1-B1)
Phillips curve:
1.0*
A
Policy function:
1.0*
0
to
15*
0
to
10
The values with a(*) supersLript are approximately the magnitudes usually
found in empirical estimates.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 7 -
To simulate this model, lagged values of income, interest rates,
inflation rates, and the money stock are required as inputs as well as the
paths of the target money stock and of capacity output.
As is evident by inspection, this model has a steady-state solution
(for A/ 0) if the target money stock follows a constant growth path.
In
that steady state the rate of inflation equals the target rate of money
growth, real output equals capacity output, and interest rates are constant.
Our procedure is to examine the responses of the system when it is disturbed
from a steady state in which capacity output is constant at the level of
actual output in 1980 Q2 and money growth is targeted at 7 percent per annum.
The steady-state values of endogenous variables are shown in Table 2.
Table 2
Steady State of Simulation Model
1524.5
y
r
8.3%
P/P
7.0%
368.9
M/P
billion measured in
period 1 dollars
billion measured in
period 1 dollars
Using these steady-state values for all lagged variables, and a 7
.
percent money growth target, the system will merely remain at its steady state
if undisturbed.
The dynamics of the model may be prob~d by examining the time
paths of deviations of key variables (usually y) from ~teady-state values in
response to a shock.
demand (ey = .01).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Consider a permanent 1 percent increase in aggregate
- 8 -
Ry equation (3), it can he seen that inflation will proceed at a
constant rate only when output is equal to capacity (y
tion, ,ci
= 1.
=
y*) since, by assump-
In the steady state, interest rates must be constant.
Suh-
stituting capacity output for y and Y-1• equation (1) may be solved for the
equilibrium interest rate.
It can he shown that the rate must be higher by
cey) to offset the increase in aggregate demand. Real balances must, of
[ai
course, be constant and lower because of the higher interest rate (equation 2).
By equation(~) interest rates wlll be constant only when money demand equals
the target monev stock~*, which is growing at a 7 percent rate.
For real
balances to be constant, prices must also be rising at a 7 percent rate.
While the system may be solved for its steady state, it will not
necessarily approach that steady state when disturbed; the system may be
unstable for some combinations of parameter values.
unstable with fairly plausible parameter values.
In fact, this model is
Figure 1 displays the time
path of deviations of output from the initial steady-state level (the output
"multipliers") resulting from a permanent 0.01 inc.rease in aggregate demand
with S1
=
.5, [ai
=
.0088, $3
=
.025, Yl
= 15.
The only "unlikely" value
in this set is the one for ~3 (the impact interest elasticity of money demand),
which is generally thought to be closer to 0.05.
a 75-year period.
The necessary compression of the time axis makes cycles
appear unusually sharp.
The vertjcal scale is truncated, which ' further
exaggerates the appearance of the cycle.
Due care must be taken in examining
Figures 1 and 2 to note the scales on the axes.
cycle emerge from Figure 1.
18 quarters duration.
Two striking features of the
First, there 'is a sharp, undamped cycle of about
Second, there appears to be a muc.h longer c.yc.le --
about 60 quarters in duration -- as well.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The simulation was run over
..
0
Figure 1
0
0 _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _...,.,
ro
..q'
0
0
0
\D
1'-
0
0
'0
0.1
<D
0
0
0
0
t.O
I
I.O
I
0
0
0
0
o
-
-
--
O
I - I
N-+----..-------,----.------.----~----r------,-.--~---...-------1'
-0 00
40 00
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
80 00
t 20.
oo
, 60. oo
200 00
Quarters
240 00
iso oo
320
oo
360
oo
AOO
co
Figure 2
0
0
--
I
�
g1
0
0
-
\
rJ)
0
0
�
0
..n
u'J
0
0
�IV
Ll...g I
::i:�0
o-
0
I
\
I
I
I
\
�
-
\
\I
I
\
1
I
I
I
I
Zo
-o
I
I
l
v
I
I
I
-�
I\
�
I
I
\
\
I
l
I
\
t-'
0
\
I
\
\
\
-
\
0
0
\
-
\.{'I
�')
\
I")
-,....
V
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
(""f'i
V'-'
rn oo
&
80 00
120 00
200 00
160 00
Quarters
�
240 00
280 00
•
}2'.),00
3GO O'.)
4C)0 C ",
11 -
Much can be learned about this system by examining the dynamics of
this same model with the price equation "disconnected"
A set to zero.
In
this case the expected and actual inflation rates will he permanently set at
7 percent.
Figure 2 shows the output multipliers for the same 0.01 shock to
aggregate demand.
The shorter cyLle is retained, virtually unaltered, but
the long cycle has disappeared.
To focus our analysis on the effect of
policy on the shorter Lycle, we suppress the effect of output on prices
(A= 0) for a series of simulations discusserl in the next section.
The analysis to follow will he restricted to the effects of shocks
to the aggregate demand for goods (changes to ey)•
The reason is that it is
for this type of shock that policies focused on monetary aggregates are
supposed to be desirable.
Shocks to the demand for money, on the other hand,
can (in this model) be completely absorbed by changes in money holdings with
no effect on output if interest rates are simply held constant.
III.
The Behavior of the Models without Price Effects
This section examines the properties of the shorter cycle of the
standard model, abstracting from the longer price-expectations cycle by
breaking the link between output and the price level (A= 0).
This is purely
a matter of analytical and expositional convenience and does not imply any
judgment about the significance or relevance of the price-expectations cycle.
In fact, as will be seen, ignoring the price cycle when making policy decisions
can have disastrous consequences.
"Before examining alternative policies, it is useful to learn more
about the nature of the cyclical responses of the model and the sensitivity
of the model rlynamicg to certain key parameters.
Table 3 displays the resultg
of varying the interest elasticities of money demand (~3) and aggregate
demand (ra 1 ).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The table shows the turning points in the time path of output
TARLE 3
Effect of Interest Elasticities on Output Hult1.pliers for a Sustained 1 Percent Increase in Aggregate Demand
A =
(3
3
1:a
i
y
n, B1
=
l/
.5
Values of output multipliers at turning points
1
1
.05
.0088
15
(5)
37.5
(13)
6.1
(23)
23.2
(32)
14.6
(41)
lQ.2
16.7
.025
.0088
15
(4)
35.7
(13)
2
-2().6
(21)
43.0
(2Q)
-29.7
(47)
50.9
(45)
-40.3
(54)
59,4
3
.01
.0088
15
(4)
34.6
(12)
-70.6
(21)
133.2
<2'n
-451.4
(41)
294.1
(48)
-1207.1
(60)
-93.5
4
.05
.0088
30
(5)
36.4
(13)
7.6
(2?)
21.7
(31)
15.6
(40)
18.5
(4q)
17.1
(58)
17.8
(60)
17.7
5
.025
.0088
30
(5)
31.8
(11)
-15.8
(lQ)
34.7
(26)
-15.0
(34)
32.7
(41)
-12.8
(48)
30.9
(56)
-10.7
(4)
6
.01
.0088
30
31.2
(11)
-65.0
(18)
115.0
(26)
-423.6
(35)
172.3
(43)
-948.9
(56)
29.0
(60)
-1150.9
7
.05
.0176
15
34.3
(12)
-13.S
(lQ)
33.2
(26)
-10.3
(34)
29.8
(42)
-6.9
(49)
26.8
(57)
-4.1
(60)
8.3
8
.025
.0176
15
(4)
32.3
(11)
-45.9
(18)
93.1
(26)
-184.8
(32)
232.1
( 40)
-614.1
(49)
287.7
(57)
-910 .3
(60)
-644.1
(4)
(11)
(18)
(26)
9
.01
.0176
15
10.1
-96.1
250.3
(39)
-724 .8
(47)
-1524.S
(60)
-581.5
10
.05
.0176
30
( 4)
31.0
(11)
-10.3
27.8
(25)
-3.3
(32)
21.4
(39)
2.4
(47)
17.1
(54)
s.7
(60)
14.4
(3)
(10)
( 2())
.025
.0176
30
30.4
-37.4
(16)
72 .1
('B)
1r
-122.3
161.9
(36)
-365.0
( /43)
248.1
(50)
-647.1
179.9
.01
.0176
30
(3)
28.9
(9)
12
(15)
195.4
(22)
-1100.0
(34)
467.?
(42)
-1524.S
(4)
-88.8
(18)
-1122
(SO)
(60)
-42.0
(60)
13.6
....N
(60)
512.44
The autonomous shock to aggregate demand is 15.24 billion "dollars. "
The number in parentheses above a value is the quarter in which that value occurs.
All simulations are660 quarters long. The value for quarter 60 does not necessarily represent,a turning point.
1/
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
(58)
- 13 multipliers during a 60-quarter simulation in which there has been a permanent
1 percent increase 1n aggregate demand.
behavior of the model with R3
demand of 0.1) 11nder a Yl
= 15
=
The first row of Table 3 snows the
.05 (a long-run interest elasticity of money
policy.
The result is a noticeable but well-
damoea cycle of about 18 quarters duration.
LowerinP, R3 to 0.025 (row 2)
shortens the cycle to about 16 quarters and makes it undamped.
ing R3 to 0.01 (row 3) results in a violently unstable cycle.
three rows of Table 3 repeat this experiment under a Yl
= 30
Further lowerThe next
policy.
Again,
the cycle becomes more violent as the interest elasticity of money demand is
reduced.
Reducing the interest elasticity of money demand increases the
magnitude of interest rate response through the policy function to movements
in output.
It might be expected that making output more responsive to interest
rates (but holding the lag structure constant) would have much the same
effect -- and in fact it does, as is shown in the last six rows of Table 3.
Doubling the sum of the ai coefficients makes the cycle less damped, Just as
reducing the interest elasticity of money demand did (compare row 7 to row 1,
row 8 to row 2, and so on).
The first six rows of Table 4 show the result~ of shortening the lag
structure on interest rates in the aggregate demand equation, from a mean lag
of 6.4 quarters to a mean lap, of 4.0 quarters.
r.omparing these responses to
the first six rows of Table 3, it ts evident that the main effect is to
shorten the duration of the cycle from about 18 quarters to about 13 quarters.
Also, the cycle becomes somewhat less damped.
Rows 7 throu~h 11 of Table 4 show the effect of an increase ln the
speed of adJustment of money demand (~1 is reduced from 0.5 to 0.25) by comparison to rows 1 through 5 of Table 3.
cycle with little impact on its duration.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
This change tends to dampen the
TARLE 4
The Effect of Lag Structure on Output Multipliers for a Sustained 1 Percent Increase in Aggregate Demand
).. = 0
s3
Me,qn
lag
'Y
1
s1
----
.OS
4./)
15
.5
2
.025
4.0
15
.5
(4)
3?.l
(10)
-19 .2
(17)
47.1
(23)
-43.3
(29)
73. 5
(36)
-86.4
(42)
115. "i
(49)
-161.2
.s
(4)
30.R
(10)
-59 .4
(17)
129.6
(23)
-389.4
(32)
33Q.1
(38)
-1179.4
(51)
397.7
(57)
-1376.4
(l•)
(10)
.5
32.8
7.9
(16)
23 .s
(23)
13 .9
(29)
19.9
(35)
16.1
(41)
18.5
(48)
17.0
(54)
17. 9
(60)
17.3
.s
(3)
30.l
(9)
-14. 5
(15)
39.5
(21)
-26.Q
(26)
52.5
(32)
-48.Q
(38)
73.3
(43)
-77 .9
(49)
97.2
(55)
-127.0
(3)
(9)
-53.9
(14)
117 .5
(20)
-417.6
(27)
265 .8
(33)
-11Q9.9
(43)
277 .o
(49)
-1416.8
(60)
266.0
.01
4.0
15
4
.OS
4.0
1()
s
6
.025
4.0
.01
4.()
30
30
.s
78
.s
(-,. 0
(17)
?5. 5
(56)
18. 5
1
3
(11)
Values of output multipliers at turning points
(SO)
(24)
(30)
(3)
(43)
11.Q
21.5
14 .8
19.5
16.2
(4)
34 .1
(SO)
16.8
~
+'
(51)
16.Q
(60)
17.9
34.7
(45)
-15.1
(53)
34.3
-270.3
(38)
216.0
(45)
-772 .2
(60)
338.1
(22)
21.4
(3?)
15.8
(41)
18.4
(SO)
17.2
(59)
17.7
(11)
(l'l)
-13.1
10.3
(26)
-8.1
(34)
25.5
(41)
-3.4
(48)
21.9
7
.07 5'};_/
6.4
15
.25
37.1
(13)
7.2
(23)
22.4
(32)
15.1
18.8
8
.0375]:/ 6. 4
15
.25
(4)
35.0
(12)
-16.7
(20)
35.4
(28)
-15.7
.015.:U
15
.25
(12)
-57 .4
(?.7)
6.4
(4)
33.2
(20)
q
103.3
. or-,J:./
(S)
10
6 .4
30
.25
36.1
(13)
8.1
11
.037 5Jj 6.4
30
.25
(4)
33.5
(5)
1/
2/
(42)
(37)
(60)
-12.6
(56)
.1
(60)
11.5
The number LTI narentheses ahove a value is the quarter in which that value occurs.
The values of B3 preserve the long-run interest elasticities of money demand found in the first fLve lines
of tl-i.1s table.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- l"i -
To summarize the results thus far, we have seen that t~is model
generates endo~enous cycles when perturbed ann that these cycles are reasonably well damped when the model coefficients are of the magnitude ordinarily
estimated.
However, the model produces undamped cyclical responses if the
interest elasticity of money demand is much lower or the interest responsiveness of aggregate demand 1s much higher than usually estimated.
The lag
structure of interest rates in the aggregate demand equation affects the
duration of the cycle with "reasonable" lag struLtures resulting in a cycle
of from three to five years duration.
or Yl = 30 monetary policies.
All of these results are for Yl
=
15
We turn now to consideration of a broader
range of alternate policies.
A policy of attempted control of the monetary aggregates can, of
course, be pursued more or less actively.
In equation (4), ~e can consider
responses with Yl varying from n to arbitrarily large.
These extremes cor-
respond to fixed interest rates and constant money growth respectively,
The
effect of varying Yl on the cycle is shown in rows 1 through 6 of Tahle 5.
In the case of Yl
=0
(fixed interest rate9) output monotonically approaches
a value "il.672 billion period-one dollars higher than the steady-state value.
Examination of equation (1) will convince the reader that this must be the
case.
While this policy would eliminate the short cycle it would also be
explosive when price linkages are present (X
> 0).
This should serve to
remind the reader that these simulations without price linkages (X = 0) are
being examined to facilitate learning about the nature of the short cycle
and that any policy conclusions reached must be re-examined in the full model.
It is apparent in Table 5 that for Yl
= 15
and above, increasing the
strength of policy response tends to damp the cycle; with Yl
longer undamped even in the case where f33
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
=
.025.
=
30 it is no
However, as may be seen in
TABLE 5
Output Multipl1.ers for a 1 "Percent Sustained Increase in Aggregate Demand, Alternative Policies
;>..
=0
l/
1
y
1
y2
y
3
Values of output multipliers at turning points
l
.05
0
0
0
Asymptotically approaches 51.672
(26)
.05
15
0
0
(5)
39.9
(16)
2
2.3
27.7
3
.05
30
0
0
(5)
36.4
(11)
7.6
21.7
4
.02'l
0
0
0
5
.02')
15
0
0
(4)
35.7
(13)
-20.6
(21)
43.0
(29)
-29.7
(37)
50.9
(45)
-40.3
(54)
59.4
(60)
-42.()
.025
30
0
0
(5)
31.8
(11)
6
-15.8
(19)
34.7
(26)
-15.()
(34)
32.7
(41)
-12.8
(48)
30.Q
(56)
-10.7
(60)
13 .6
7
.025
Constant money
growth
(4)
30.5
(10)
-7.1
(17)
23 .1
(24)
1.8
(30)
16.1
(37)
6.9
(43)
12.7
(50)
9.1
(57)
11.3
.os
(5)
8
0
15
0
18.q
(16)
19.Q
(26)
25.3
(37)
23.6
(47)
24.7
(58)
24.0
?4 .o
9
.05
0
30
0
(5)
37. 5
(14)
15.4
(25)
~3. ()
(34)
20.4
(44)
21.4
(54)
21.0
(60)
21.1
10
.02'l
0
15
0
(5)
37.6
(14)
11.6
(24)
22.6
(34)
18.1
(43)
20.0
(53)
29.2
(60)
19.5
.0'.?5
0
30
0
(4)
35.1
(13)
11
1. 8
(21)
21.1
(30)
12.6
(39)
17.0
(48)
14.8
(56)
15.9
(60)
15.7
0
0
5
(14)
18.5
(26)
23.8
(48)
.025
(5)
37.6
(36)
12
22.3
72.7
(58)
22.6
13
.025
15
0
5
(4)
34 .o
(12)
-14.7
(19)
32.6
(27)
-12.2
(35)
30.7
(43)
-9.8
(51)
28.5
(58)
-7.3
~
1/
(22)
(37)
1().6
(43)
22.3
(57)
14.4
(31)
15.6
(40)
18.5
(49)
17.1
(60)
17.7
Asymptotically approaches 51.672
The number in parentheses above a value 1s the quarter in which that value occurs.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
(58)
17.8
(60)
(60)
10 .6
t-'
.J"I
- 17 -
row 7 of Table 1, the cycle is present even when money growth targets are
ad~ered to exactly.
Instead of targeting the levels of money, one might wish to consider
policy functions which target the rate of growth of money (the Y2 term in
equation 4).
Such a policy would resist undesired changes in the money
stock, hut once made they would be "forgiven" (if Yl = 0), and policy would
he directed at reasserting the desired growth rate of money.
11 show the results of Y2
= 15
and Y2
= 30.
Rows 8 through
These policies (with Yl = 0) fare
much better than the Yl policies which focus on the level of money balances.
Nevertheless, there is a distinct, although damped cycle.
Since we have heen implicitly presuMing that the final obJective
here is to smooth the path of income (at least in these cases where the rate
of inflation is fixed) it seems reasonable to ~uppose that a policy which
re~ponds directly to movement~ in income would be superior to one which chases
some other endogenous variable.
Accordingly, a more general policy might be
written as follows:
lnrt = lnrt-1 + Y1ln(M/M*) + Y2Mn(T>.f/M*) + Y3tilny;
(4)'
with Y3
> 0,
this policy would have the Federal Reserve raise interest
rates as the rate of growth of output increases.
~ow 12 of Table 5 reveals
that a pure "direct target" policy fyt, Y2 = 0) is indeed superior to all
other active policies considered so far.
cycle.
It virtually eliminates the short
Finally, the last row of Tahle 5 sho~s that adding a monest response
to monetary aggre~ates (Y1 = 15) to a direct target policy re-introduces
the cycle.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- lR -
VI.
SiMulations with Price Linkages in Effect
Tle have seen in the preceding SPction that the hypothetical monetary
policy is the source of the short cycle in our model, and that if policy
were to respond directly to an oucnut tarRet and not at all to money growth
targets, the path of output would be convergent and much smoother.
a policy Might have no protection from runaway inflation.
Rut such
These policies
must, therefore, be reconsidered in the full model inclusive of price linkages
Table 6 shows Lhe model responses un<ler different policies with
price linkages in effect.
The first row shows the results with R3 = .025
and our standard monetary targets (Y1
= 15)
policy.
~his is merely a
tabular presentation of the same data shown in Figure 1.
The undamped 18-
quarter cycle is readily apparent aq well as the presence of the longer
(expectations) cycle.
(y +5
= 5)
By contrast, a policy which responds to output only
shows no short cvcle but is explosive over the 60 quarters (row 3).
As shown in the final column of Table 6, the inflation rate at the end of
this simulation is 11.46 percent above the initial steady 7 percent rate.
The problem here is the divergence between the real and nominal rates of
interest.
simulation.
Real rates of interest are about -3 percent at the end of this
tf policy were to focus on real interest rates by compensat1nP,
for changes in inflation, the qystern should be expected to be damped in the
long run.
The policy function Might accordingly he modifie0 as follows:
(4")
where n* is the desired long-run rate of inflation.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
TARLE 6
Multipliers for a 1 Percent Sustained Increase in Aggregate Demand, Alternative Policies
A
-y
-y
1
1
' 2
-y
3
15
-y
4
0
0
Fixed money ~rowth
15., R3
=
.025
Inflation rate
mult1.pl1ers at
60 quarters'.!:._/
Values of output multipliers at turning points
')
0
=
(4)
-%.5
(14)
-40. l
(22)
32.'3
(31)
-63.4
(40)
-53.4
(48)
-53.4
(58)
7g_7
.68
(4)
30.7
(11)
-16.8
(17)
g_4
(25)
-1g.5
(32}
-4.2
(38)
-11.9
(60)
11.0
.()0
(60)
205.0
11.46
(60)
3.3
1~20
3
0
5
()
0
40.2
(12)
35 .B
4
0
5
l.'1
0
(5)
35.6
(14)
5.6
(23)
11.9
(36)
5.6
(5)
(14)
3.4
(22)
8.1
(60)
-5.6
(15)
.03
(21)
3.1
-9.5
(6)
1../
(40)
5.7
t-'
5
0
5
1.5
.02
35.5
6
0
5
1.5
.05
35.4
(5)
1/
2/
(SO)
-..0
.54
(60)
-7.8
The nuMber in parentheses above a value iq the quarter in which that value occurs.
Target rate of inflation assumeo to he 7 percent.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
.01
- 20 -
As can he seen in row 4 of Table 6, with Y3 = 1,
y1
= O,
~
= 1.5, and
the system is very well damped in both the short and the long run.
The main oh1ect1on to the re9ults of this nolicy is that the equilibrium
inflation rate is permitted to change in response to shocks.
In this partic-
ular simulation the inflation rate appears to be approaching a new equilibrium
level at above 8 percent; it is 1.2 percent above "target" at the end of 60
quarters.
Thi9 happens because policy (4'') with y1 = 0 resists changes 1n
the rate of inflation, but once the inflation rate changes, makes no effort
to return inflation to anv particular level.
The final two rows of Table 6
show that any significant response to the level of the inflation rate (y5
> O)
.
tends to increase the fluctuations in income, the effect increasing with the
size of YS·
0n the other hand, inflation is held reasonably close to the
targeten 7 percent rate for these runs.
=
The final targets policy with y1
.05 does about as well at holding inflation near its target as does the
fixed money growth policy and it does much better at smoothing income responses.
V.
Summary and Conclusions
The simulation properties of a simple dynamic IS-LM model have been
examined at some length.
It was found that the model responds cyclically to
9hocks under a fairly wide variety of monetary policies.
appear to be two basic cycles:
In fact, there
a shorter cycle associated with "money target-
ing" monetary policy responses and a longer price-expectations cycle (more
than 15 years).
The short cycle is especially noticeable with policies aimed
at ach1eving a nartLcular monetary a~gregate level.
This cycle hecomes ~uch
more pronounced as the interest elasticity of money rlemand is reduced or as
the interest ela9t1c1ty of the demand for goods is increased.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The strength
- 21 -
of policy response to deviations from monetary target levels
0r
from growth
rates priMarily affects the degree of damping in the system, leaving the
frequency virtually unchanged.
The la~ structure of the mo<lel, especially that of the demand for
goods, determines the frequency of the short cycle:
the shorter the mean
lag of interest rates in the IS curve, the shorter the duration of the cycle.
In all cases examined, though, this cycle is fairly long, ranging from about
three to four and one-half years.
It is not likely that for plausible sets
of parameter values this model will be useful for explaining the erratic
path of output for 1q7q Q3 through lq80 04.
Finally, it was found that a policy based on final targets of output
growth and inflation was under some circumstances superior to monetary aggregates policies in smoothing the system's output responses to shocks.
These results are, of course, strictly valid only for the specific
model consLdered.
The scope of experiments performed was broad enough to
give some reason to believe that these results would be qualitatively representative of a wide range of models of similar hut more elaborate structure
(for example, the MPS model).
The results would almost certainly not be
representative, however, of currehtly fashionable models in which expectations
variahles are specifjed to be identical with the model's outcomes for those
'
variables. In most ratLonal-expectations Models, final target policies will
ultimately do no better than monetary target policies.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The New Approach to Monetary nolicy A View fr~m the Foreign Exchange Trading Desk
At the'Federal Reserve Bank of New York
January
1981
Paper Written for a Federal Reserve
Staff Review of Monetary Control
Procedures
by
Margaret L. Greene
January 22, 1981
THE NEW APPROACH TO MONETARY POLICYA VIEW FROM THE FOREIGK EXCHANGE TRADING DESK
AT THE FEDERAL RESERVE BANK OF NEW YORK
Margaret L. Greene
The change in approach to monetary policy in October 1979 had
a strongly positive l.Illpact on the psychology of the foreign exchange market.
that time the foreign exchange markets had become thoroughly disappointed
with the previous monetary procedures.
To be sure, interest rates had
increased progressively since the summer of 1977, and by the end of September
1979 the discount rate was at 11 percent--some 5 1/2 percentage points
above its previous low.
But, with the rate of inflation accelerating
more rapidly and the growth in the money supply again far above target,
monetary policy appeared to have little bite.
In the view of most market
participants, the Federal Reserve's continuing difficulties in reaching
its targets reflected either an inability or an unwillingness to take
whatever actions were necessary to put up a credible fight against
inflation.
Against this background, the fact that the Federal Reserve would
take such a radical departure in the conduct of its open market operat1onf
underscored its seriousness about dealing with inflation.
In the foreign
exchange market, the new procedures, which put more emphasis on bank
reserves than on interest rate levels, were expected to be more
effective in containing the growth of the monetary aggregates.
If
there were fewer dollars sloshing around the international markets, the
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
By
2
argument went, then there would be less pressure against the dollar in
the exchange market.
Most market participants expected the new procedures
to entail quicker and, at times, larger interest rate adjustments than
before to changing economic or financial conditions.
They also
interpreted the change in approach as likely to lead to higher
interest rates in the United States than otherwise, in view of our
inflation record.
The central banks abroad also welcomed the change
in approach to monetary policy.
They, too, viewed it as indicating
seriousness of purpose on the part of the Federal Reserve and as
providing for potentially more effective monetary control.
For the Foreign Exchange Trading Desk, the change in approach
to monetary policy did not entail any new procedures for the conduct of
our foreign exchange operations.
We anticipated that, if interest rates
became more volatile as a result of the change in monetary procedures,
central banks as a group might need to be quicker to respond to short-run
disturbances in the foreign exchange market that threatened to render
foreign exchange trading disorderly.
But, for our part, we saw no need
to resist the impact ?f interest rate changes on the excr.ange rate for the
dollar.
In the event that the dollar came under pressure, resources to
finance U.S. exchange market intervention had been greatly enhanced
as part of the November 1, 1978, package.
Meanwhile, other central
banks would find it easier to cooperate with us to help contain
selling pressures against the dollar if any should develop.
For one
thing, they thought that, under the new procedures, they would be
less likely to be drawn into a sustained intervention effort that nught
pose complications for the control of liquidity in their own domestic
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
3
markets.
For another, they were satisfied that the U.S. author1.t1.es
had clearly taken action to restore balance to our domestic economy.
If
the dollar became well bid in response to the demand for money and credit
in the United States, the Desk could take advantage of the dollar's strength
to buy in the market and from correspondents the currencies needed to repay
the Federal Reserve's swap drawings,cover the Treasury's foreign currency
obligations, and rebuild currency balances for the U.S. authorities.
In
the process, our currency operations would help cushion some of the impact
of U.S. interest rate
volatility on the exchange rates and money markets
of our trading partners.
The experience with the new procedures to date has left the markets
with the impression that U.S. short-term interest rates were higher on
average during the past 14 months than they otherwise would have been.
In
any case, U.S. interest rates were more volatile on a day-to-day basiJ:./ and
subJect to sharper, if shorter,cyclical swings.
Also, expectations about
likely interest rate developments shifted far more abruptly and strongly
than before.
The greater volatility in interest rates in the United States has
imparted more volatility to the exchange markets as substantial changes
in the relative cost of funds or yield on assets among countries created
incentives for holders of the rapidly growing pool of internationally
mobile and interest-sensitive funds to shift from one currency to another.
On a number of occasions after October 6, 197~ the exchange markets became
disorderly.
Short-term movements in exchange rates, spurred freq~ently,
but not exclusively, by interest rate considerations, have been far greater
Jj
Dana B. Johnson, "The Impact of the Federal Reserve' s New Operating
Procedures on the Variability of Interest Rates" (BoJrd of Governors
of the Federal Reserve System, January 1981).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
4
than can be attributed to the more gradual shifts in the underlying
fundamentals.
(See Chart 1 and compare with Chart 2 for the period
November 1, 1978,to October 5, 1979.)
Day-to-day movements in the
pivotal dollar-German mark relationship have at times been wide and
erratic and rank among the greatest 24-hour changes in exchange rates
since the U.S. authorities resumed intervention in July 1973 after
the floating of the dollar earlier that year.
(See Chart 3 and compare
with Chart 4 for the eariier period._g_/) Also, spreads between bid and
asked rates widened,and foreign exchange dealers became reluctant
to take dollars into their positions even for short periods during a day.
On such occasions the Desk had to intervene, sometimes forcefully.
The most problematic period from the point of view of the Foreign
Exchange Trading Desk was April 8-July 16, 1980,when U.S. interest rates
tumbled rapidly, falling to levels below those generally prevailing abroad,
and when the two-phase elinunation of credit controls generated fears in
the foreign exchange markets that the Federal Reserve was actively moving
to an accommodative monetary stance.
This process began with an abrupt
turnaround in exchange market expectations that prompted a generalized
•
dumping of dollars around the world and pushed the dollar-mark spot rate
down 5-1/2 percent in Just 24 hours.
Subsequently, as traders recognized
gj
The data in these charts are derived from quotations in the U.S.
market including the close of one day and the range of quotations
during the entire succeeding trading session. Although they encompass
a 24-hour period, the data do not include actual exchange rate
quotations for the dollar-mark rate in Far Eastern or European markets.
When trading is active, rates in these other centers are more likely
to fall outside the range of rates quoted in the United States. Therefore,
these charts systematically understate the variability of the dollarmark rate at times when the exchange markets are most volatile.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
5
that the Federal Reserve's approa~h to monetary policy implied that some
easing in financial market conditions was appropriate as long as money
and credit demand weakened and evidence of recession mounted, the dollarmark rate fluctuated 1-1/2 to 2 percent on each of several days during the
remainder of the period.
In this environment, the Desk made no attempt to hold the
dollar rate.
The obJective was merely to contain the pressures against
the dollar so that they did not cumulate to the point of taking on a momentum
of their own.
' this obJective, the Desk maintained a flexible
To achieve
approach in the market in order to take advantage of any opportunity to
inject a sense of two-way risk to the exchange markets.
we had the close cooperation of the Bundesbank.
In this respect
In addition, we were in
frequent, often minute-to-minute, consultation with the Government Securities
Trading Desk.
Although that Desk did not basically have to alter its approach
to the conduct of monetary policy,-l/ we found several opportunities to
adjust the timing of operations to obtain maximum psychological
effect for our foreign exchange intervention.
In the end, the pressures
against the dollar during the April-July period were severe but did not become
as entrenched as they had been before.
Although we sold $200 million equivalent
of marks or more on five days during this period, we did not have as many
days of large intervention as we did in 1979.
Nor did we have to intervene
as frequently. (See Table 1 . )
Apart from these periods of pressure, the dollar has generally
fared well in the exchange markets during the period after October 6, 1979.
The market was more confident of the dollar's underlying strength
than before.
]./
In part this was because of the improvement in our
Fred J. Levin, "Implementing the New Operating Procedures: The View
from the Trading Desk" (Federal Reserve Bank of New York, January 1981).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
6
TABLE 1.
U.S. OFFICIAL SALES OF GERMAN MARKS IN THE MARKET,
NOVEMBER 1978-NOVEMBER 1980 1/
(in millions of U.S. dollar equivalent)
Time Periods
Total intervention
of that period
Number of days of
intervention/
total business
days in period
Number of days when ..
intervention exceeds
$200 million
equivalent
Before October 6 measures
November 1 December 29, 1978
$5,551.6
26/41 = 63%
11
June 15 July 26, 1979
$5,267.1
26/30 = 86.7%
11
August 30 October 6, 1979
$3,769.9
17/27 = 63%
10
After October 6 measures
December 3, 1979
February 7, 1980
$1,340.9
15/47 = 31.9%
2
April 8 July 16, 1980
$3,221. 7
27/72 = 37.5%
5
!/ These figures include sales by the Desk in the markets in New York and the Far East.
They exclude sales to correspondents. The sum of the intervention in the five
periods represents 95 percent of all mark intervention from November 1, 1978,to
July 16, 1980. The remaining $1,099.1 million equivalent was done intermittently
outside of the five periods of heavy dollar pressure indicated above.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
7
current account position and in part because other countries had, and were
more readily perceived ashaving, economic problems of their own.
Interest
rate differentials were strongly favorable to the United States at times
during the year, attracting inflows from abroad that served to buoy the
dollar in the exchanges.
Moreover, uncertainty under the new monetary
approach about the near-term course of U.S. interest rates added a new
dimension of risk for anyone about to sell the dollar short.
This more favorable attitude toward
the dollar permitted the nesk to
buy German marks and other currencies more often than it had sell and to
restore the foreign currency position of the U.S. authorities.
Because the
market was more willing to hold dollars under conditions in which we were in the
.
market to buy currency, we had more leeway for covering the Federal Reserve's
swap debt and the Treasury's foreign currency obligations.
In all cases we
operated to buy currency in a manner that would avoid leaving the impression
that we desired to put a cap on the dollar's rise.
But at times--particularly
in early April and again in the period October 9-November 28--we did intervene
quite forcefully to prevent the dollar's rise from becoming disorderly.
( See Table 2 . )
In the period October 7, 1979,to November 28 1980, the
Desk was a net buyer o~ currency.
On balance, we purchased
$1,393,7 nullion equivalent of German marks in market operations.
In
other operations (including net correspondent purchases and purchases
from the Bundesbank which do not reflect its intervention sales of dollars),
the Desk acquired an additional $5,873,6 million equivalent net of
German marks, bringing total net acquisitions of German marks after October 6
197~ to $7,267.3 million equivalent.
Of this a.mount,$3,750,3 million equivalent
was used to repay System swap debt and $2,493,3 million equivalent was put
into 'Ireasury balances to cover its "Carter-bond" liabilities.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The
8
TABLE 2.
U.S. OFFICIAL PURCHASES OF GERMAN MARKS IN THE MARKE'I
NOVEMBER 1978-NOVEMBER 1980 1/
(in millions of U.S. dollars equivalent)
Time pen.eds
Total operations in
market during that period
Number of days of
operations in market/
total business
days in period
Number of days
when operationsexceed $150 mil.
equivalent
Before October 6 measures
April 1 May 31, 1979
$2,647.1
31/43 = 72%
9
After October 6 measures
February 22 April 16, 1980
$2,543.6
28/38"" 74%
6
July 29 October 8, 1980
$
889.7
27/51 = 53%
0
October 9 November 28, 1980
$2,914.7
29/30 = 97%
6
1/ These figures include purchases by the Desk in the markets in New York, Frankfurt,
and the Far East, both spot and forward,and also purchases from the Bundesbank
which correspond to its intervention sales of dollars in the Frankfurt market.
They exclude purchases from the Bundesbank not directly representing its sales
of dollars in the market or purchases from other correspondents. The sum of tht
amounts in the periods above is 99 percent of all market purchases from April 1,
1979,to November 28, 1980. The remaining $96.2 million equivalent was done
intermittently outside of the four periods of dollar strength indicated above.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
9
remaining $1,023.7 million eqUJ.valent was also put into balances to establish,
in effect, a long mark position for the System to be used as needed to
finance future foreign exchange intervention.
In addition the Desk
purchased $507.9 million equivalent of other currencies, mostly Swiss francs
and French francs.
For the excqange markets in late 1979, volatility was not a new
phenomenon.
Ever since the dollar was floated, institutions participating
in these markets coped with considerable exchange rate variability
and had already proceeded to adJust their trading practices and operating
procedures accordingly.
Trading, credit,and position limits were
reviewed,and new procedures for monitoring the trading operations of each
bank or branch within separate guidelines were put in place.
Even so,
to take a "view" on a currency and hold a position for several months,
as had been common practice under the par value system of exchange rates,
had long beenconsidered risky.
Instead, the professionals in the market,
the dealers in the largest market-making banks, came to look for opportunities
where they could make a profit in a single day so as to avoid exposing their
position to the risk that exchange rates might move against them even in as
short a time as overnight.
This env1rorunent fostered daily "in-and-out"
transactions that ballooned daily turnover, generated a large increase in
spot trading at the expense of forward trading, and increasingly focused
exchange market attention on transitory rather than fundamental factors
affecting exchange rates.
This change already had 1mpl1cations for the way
the exchange market behaved. At times of economic or political uncertainty,
the exchange markets quickly lost resiliency.
In addition, the role ofrisk-
taking over periods of time shifted from the professionals in the interbank
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
10
exchange market back to their customers and ultimately to consumers, workers,
and investors whowereunwittingly being exposed to the effects on prices
'
and incomes of exchange rate variability.
After October 6, 1979,uncertainty over the implications of the new
approach to monetary policy for the near-term outlook for U.S. interest
rates added a new dimension of risk for the foreign exchange trading
community.
Indeed, the most active position-taking banks claim
they took
fewer and smaller short-dollar,long-mark positions for the period overall.
There is little evidence to suggest, however, that these banks changed
their position-taking activities in a systematic way.
They assumed
at times as large positions in German marks as before October 6, and the
foreign exchange managers of the largest position-taking banks confirm that
there was no effort to change the limits within which individual dealers or
particular trading operations traded.
Over the course of the year, some market participants sought ways
to respond to the new source of rate volatility.
At firs~ participants
in the interbank market stepped up their "in-and-out" spot trading as banks
either tried to take advantage of potential profit opportunities or acted
defensively simply to protect their o.m positions. But as each bank fought
to catch a turn in the exchange rate before its competitors, the
whole market found that volumes continued to soar, as documented by this
Bank's 1980
survey of foreign exchange turnov~r; but profits did not
necessarily grow.
Moreover, the back offices were encountering severe
strains in processing the work needed to effect payments, the risk of costly
payment errors was mounting, credit lines with counterparties were being
exhausted, and the sheer volume of trading was exposing banks to an ever-
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
11
increasing risk that settlement problems on a particular day could pose
a severe problem.
At the same time, exchange rates were increasingly
domnated by short-term and technical factors that were unpredictable,and
these erratic fluctuations merely accentuated the risk of carrying a
foreign exchange exposure.
Meanwhile,major money ~enter banks with a large comm.J.tment to the
U.S. domestic markets found that they could use their expertise in the foreign
exchange area to capitalize on the interest rate volatility while also
dealing with these operating problems in their foreign exchange back offices.
Their foreign exchange dealers had far more experience with volatile markets
r
than most of their domestic money and bond dealers.
The forward markets
in exchange can be used as an effective alternative to deposit-taking
activities for taking a view on interest rate~/ while leaving the bank's
currency position covered and not necessarily affecting the balance sheet
of the bank.2/
In addition, a bank can utilize some of its position-taking
a,uthori ty in currency to borrow where interest rates are relatively low
and thereby reduce the cost of funding for the bank.
of the more important trading banks began to shift
expertise from spot trading per se toward
Accordingly, a number
their trading
funding and arbitrage operations,
while others incorporated such a shif't in their plans for 1981.
To the
extent that banks were to move in this direction, they would have less
need to protect their spot positions.
Therefore, as long as no clear
':±.I
By doing a swap (that is simultaneously engaging in offsetting spot and
forward transactions in the same currency), a foreign exchange dealer can
either generate dollars to fund a loan or produce income in a manner
similar to placing a deposit.
2/
If the near side of a swap is rolled on a day-to-day basis, all of the
transactions are off-balance-sheet items.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
12
trend in exchange rates were to develop, the markets might prove to be more
resilient to short-term and potentiaJ.ly reversible movements in exchange
rates than before.
Not all market participants can make such a shift in their
approach to foreig~ exchange, however.
Those banks that may not be
connnitted to playing a major money .ro.arket role and may not have innnediate
access to money market developments in the United States say they find the
volatility of interest rates to be a major headache that serves to complicate
their efforts to fund their dollar operations.
If this volatility should
persist, we may find that it has different implications for different types
of institutions in the exchange market.
Non-dollar-based banks that find they
have difficulties with their dollar operations may have real incentives for
seeking to shift a larger part of their international business into
alternative currencies.
Market participants have expressed their concern
that investors, private or official, that have only an arm's-length
relationship to U.S. markets may seek to protect themselves from the
volatility of rates, lack of predictability over capital values, and
the possibility that yields can drop to quite low levels by stepping up
their efforts to diversify out of dollars.
In conclusion, the Foreign Exchange Trading Desk's experience
since October 6, 1979,has been generally favorable.
Although we have had
to intervene at times, the periods of selling pressure
have been less
prolonged than before,and on balance we have been able to build up foreign
currency reserves.
In part this more favorable environment for the dollar
reflected the market's assessment that the economic fundamentals for the
United States have improved both absolutely and relative to those for other
countries.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
But to the extent that the new monetary procedures resulted in higher
13
interest rates than would otherwise have prevailed or increased uncertainty
about the cost of funding a potential short dollar position, they may have
contributed indirectly to a generally more resilient dollar.
A longer-term assessment of the new monetary operating procedures,
as seen from the Foreign Exchange Trading Desk, is more difficult to draw.
So far,it is not clear whether the high degree of interest rate volatility
experienced so far is an essential feature of the new procedures rather
than an outgrowth of the imbalances in the economy at the time they were
adopted.
Moreover, the exchange market's adaptation to an environment of
dollar interest rate volatility has not yet been completed.
But, on the
basis o~what fragmentary evidence we have obtained to date,it appears
that, if increased interest rate variability becomes a perm.anent feature
for dollar-based money markets 9 we face the prospect of a substantial growth
in interest-arbitrage activities.
we live.,an increase in the volume
In the uncertain environment in which
0£
internationally based arbitrage activities
could expose the United States to even more interest rate volatility in response
to developments abroad.
If this volatility were to pose an increased risk
for the parties seeking outlets to commJ.t funds, it could~ in due course,
provide further impetus for shifting 1nternational lending, and investing
business out of dollars and into other currencies.
In the long run, such
a shift ~ould have far-reaching consequences for the dollar and for its role
in the internatior.al monetary system.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Chart 1
14
EXCHANGE RATES AND 3-MONTH EURO-RATE INTEREST DIFFERENTIALS
BETWEEN OM AND U.S. DOLLAR AFTER OCTOBER 6, 1979
OCTOBER 10, 1979
10
9i-------o----
-
NOVEMBER 26, 1980
INTEREST-RA TE
DIFFERENTIAL
Scale Left
ai------+----
71------+---
5---
41-------'-----_3 " - - - - - - 1 - - - - - -
----I
2.00
-------1
1. 95
1.90
2i;__-----1-----
£XCHANGE
RAT£S
- ~ 1.85
l"------1--
Scale Right
0
-----f
---------l
0
ND
1979
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
J
FM
AM
J
1980
J
AS
ON
1. 80
1. 75
Chart 2
15
EXCHANGE RATES AND 3-MONTH EURO-RATE INTEREST DIFFERENTIALS
BETWEEN OM AND U.S. DOLLAR BEFORE OCTOBER 6, 1979
NOVEMBER 8, 1978
-
OCTOBER 3, 1979
10
9i - - - - - + - - - - - - - - - - - - - - - - - - t
INTEREST-RA TE DIFFERENTIAL
8
Scale Left
----------f
1-----&------
- - - ~ 2.00
4- - - - - - - - - - - - -
3i-----+----------------11.95
- - - -........ 1.90
2
Scale Right
0! - - - - + - - - - - - - - - - - -
1.80
! - - - - - ~ - - - - - - - - - - - - , 1.75
.._.__.170
L...L..L..1...L..1-JU,..L.J.~~~,__..................._.___._........,_.............___,_....___._.............
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
ND
1978
J
FM
AM
J
1979
J
A
S 0
-....
c.Dc,:t·
___
==
--==="""'
...,
__...
----
.....
....
-==--
--=
....., .....
z
:u a::11 : : -
_...,=-==
-
B "-'I :::E ~
ll:::lllz:::!
ie-
,::::a
ID
.... :::E 11:111:a:""'
A. .......
~--.=..
---= -
q
=
....=
...==
-==
16
=
..,
=
:::3
=-
==....
=a::
=
=
~=
=~
=: ...
=-----.:--
=
c:::,
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-=t
...,
31=
~
Ci)
....
,._
..,
....
==
c:,~ -
~
-x
::c
c:::,
- -....
=
-
....,._
a:
.....
.....
0,
c:::,
--...=
..... c::::t -
c.:,
;a: ==
t:::I
...., ::ai:
C,jll)::E
.... a:11a:
-c
Cl!
(.)
.:.
~z:
,c -a:
,:a::a:...,
I.ii.I& a::
co
-....
c,)
z:
c:::,
::IE
D,_,IAolllCI
.....
>,c
c::a
=
c:::,
,_
c:::,
c:::,
....
c:::,
c:::,
..,
17
c:::,
c:::,
-
c:::,
c:::,
..,.
...
=
c::,
+
c.:,
-=
--=c:::,
=z:
-=
-
=:IE
=
==
=-
.....
...__
~-:::::;
~
==1
z:
0,
,_
-
==3
=l
~
-....,.
::::::! .....
:z::
c:::,
='.:1CI
~
=
--<
__,
=
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
INTEREST RATE VARIABILITY UNDER THE NEW
OPERATING PROCEDURES AND THE INITIAL RESPONSE
IN FINANCIAL MARKETS
January
1981
Paper Written for a Federal Reserve
Staff Review of Monetary Control
Procedures by
Dana Johnson
January 22, 1981
INTEREST RATE VARIABILITY UNDER THE NEW
OPERATING PROCEDURES AND THE INITIAL RESPONSE
IN FINANCIAL MARKETS
CONTENTS
I.
II.
Introduction and Summary
1
The Impacts of the Federal Reserve's New Operating
Procedures on the Variability of Interest Rates
6
A.
B.
C.
III.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
PAGE
The Implications of the New Procedures for the
Federal Funds Rate
Empirical Findings on the Variability of the
Federal Funds Rate
Variability of Longer-Term Interest Rates
Reactions in Financial Markets to the Increase in
Interest Rate Variability
A.
B.
C.
D.
E.
F.
Liquidity Premiums on Government Securities
Before and After October 1979
Treasury and Agency Security Markets
Underwriting Spreads on Corporate Bonds
Reactions in Mortgage Markets to Increased Rate
Volatility
Commercial Bank Response to Rate Volatility
Developments in Financial Futures Markets
6
11
20
36
36
45
53
63
72
79
INTEREST RA TE VARIABILITY UNDER THE NEW
OPERATING PROCEDURES AND THE INITIAL
RESPONSE IN FINANCIAL MARKETS*
I.
Introduction and Summary
In an effort to achieve better control over the growth of the
monetary aggregates, the Federal Reserve adopted reserves-oriented
-
operating procedures in October 1979.
-
It was widely expected:that the
increased emphasis i~ daily operations on the supply of reserves would
come at the cost of greater variability of the federal funds rate.
It
was unclear, however, to what degree the heightened variability of the
federal funds rate would be reflected in other interest rates, and thus,
it was uncertain whether there were likely to be significant repercussions
in financial markets.
Based on evidence from about one year's experience with the new
operating procedures, this study has found that the federal funds rate
has indeed been markedly more volatile than in other recent years and that
this increased variability has been transmitted to other interest rates to
a considerable extent.
Also, some evidence of reactions to the increase
in variability was found in key financial markets, though on the whole,
such adjustments have been relatively modest.
It should be emphasized that virtually all of the findings of
this study are highly tentative.
Given the turbulent economic environment
of the past year, it has been difficult to identify with assurance what,
~/ Dana Johnson of the Federal Reserve Board staff was primarily responsible for the preparation of this paper. Major contributors to this study
were Norman Mains, Jim O'Brien, Barbara Opper, Leigh Ribble, and David Seiders
of the Board staff, and Ed Stevens and Bill Gavin of the Cleveland Federal
Reserve Bank staff.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-2-
in fact, the reaction to the new procedures has been.
But even if this
study has been relatively accurate in describing the initial ~esponse,
it seems likely that the behavior of interest rates may continue to evolve
in light of further experience with the new procedures.
Likewise, the
\repercussions tn financial markets that might be expected as·a result of
increased interest rate variability may take considerably longer than a
year to emerge.
(1)
A summary of the principal findings are presented below.
Over the course of a day, from day to day, and from week
to week, the federal funds rate 'has shown a marked increase in volatility
subsequent to October 6, 1979.
This past year, the range over which the
weekly average funds rate moved was about comparable to the range of
movement in the 1973-75 cycle, but of course, the swings in the funds
rate recently have occurred much more quickly.
Even after the data are
smoothed by removing these cyclical swings, however, the funds rate on a
weekly average basis was about three times more variable over the past
year than over the previous 11 years.
(2)
Interest rates on Treasury securities across the maturity
spectrum likewise have shown a clear increase in volatility.
In the year
since October 6, 1979, the standard deviation of the weekly change in
rates for Treasury securities of various maturities has been three to four
times greater than in the previous 11 years.
Much of this increase
in variability reflects, of course, the unusually sharp cyclical swings
experienced this past year; but even apart from those swings, Treasury
rates have shown considerably more nonsystematic variability since
October 1979.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-3-
(3)
Despite the greater nonsystematic variability of the
federal funds rate, its level has continued to be very highly correlated
with the levels of rates on Treasury securities.
Moreover, expressed
either as a first difference or as a deviation from a moving average,
the federal funds rate has been more highly correlated with ouher rates
subsequent to October 1979.
This surprising result may reflect the
~
difficulty under the new procedures of distinguishing between cyclical
and random movements in the funds rate.
This finding also may be related
to the nature of the new procedure~which automatically respond to the
growth of the monetary aggregates.
Finally, the ability of banks to
speculate on the interweekly movements of the funds by altering their
behavior with respect to discount window borrowing and the carryover of
excess reserves also may help to explain the closer association of the
funds rate with other rates.
(4)
The greater variability of interest rates does not appear
to have significantly increased liquidity premiums either on Treasury
bills or on Treasury coupon issues.
The measurement of liquidity premiums
always has been difficult, however, and thus the inability to document
that an increase in premiums has occurred is not-surprising,particularly
for a period as short and as volatile as this past year.
(5)
Net positions in Treasury securities maintained by primary
dealers were about the same size on average this past year as in the
previous year.
Trading volume rose more rapidly than gross positions
this past yea½ indicating some increase in the turnover of dealer inventory.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Subsequent to October 6, 1979, bid-ask spreads and dealers' implicit
-4-
underwriting spreads appear to have widened for Treasury bills.
There
was no systematic evidence, however, of a comparable widening of bidask spreads on Treasury coupon issues.
(6)
The average underwriting spreads on publicly offered
corporate notes and bonds narrowed rather than widened this past year.
This overall result was due entirely to a relatively sharp narrowing
in the average underwriting spread for issues rated Baa and below.
Disaggregating the data further showed that average underwriting spreads
widened on issues that were awarded to syndicates via competitive bidding
in the period since October 6, 1979, but that spreads narrowed for issues
on which the offering terms were negotiated.
(7)
A number of adjustments were made by various institutions
to their mortgage connnitment and investment policies.
Mortgage originators
and investors have sought to shift some of their interest rate risk to
mortgage borrowers.
Originators have reduced their connnitment-period risks
by imposing larger nonrefundable commitment fees, by shortening periods over
which fixed-rate connnitments will be issued, or by writing floating-rate
connnitments.
Private investors have marketed more adJustable-rate mortgage
instruments, or instruments involving equity participations.
FNMA has
restricted the assumability of the conventional long-term fixed-rate mortgages that it purchases and holds.
Moreover, the availability of long-term
fixed-rate standby purchase commitments has contracted to some degree both
at FNMA and at the dealers in GNMA-guaranteed passthrough securities, and
the cost of these put options has increased.
The maJor GNMA securities
dealers also have used the futures markets more heavily to hedge their positions, margin requirements on firm-delivery forward contracts have become
more common, and bid-asked price spreads have widened somewhat.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-5(8)
Commercial banks appear to have made only limited adjustments
to the increase in interest rate volatility.
The overall share of assets
invested in rate-sensitive instruments has not changed much since mid-1979,
although the share of term loans carrying floating rates has increased.
At
the same time, most banks continued to experience an increase in the share
of their liabilities that are rate sensitive.
Thus, on balance, banks do
~
not appear to have reduced their exposure to interest rate risk over the
past year.
(9) Activity in financial futures markets continued to expand over
the past year.
Sales volume increased for the major contracts now being
offered, and the markets reportedly functioned smoothly despite the increased
volatility of rates in both cash and futures markets.
Open interest in
Treasury bill' contracts declined on balance over the past year but grew
enormously in Treasury bond contracts.
The limited data available on hedging
activity in futures market indicate that it continued to be substantial,
but there is no indication that hedging is accounting for a larger share of
overall activity.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-6-
II.
The Impacts of the Federal Reserve's New Operating Procedures
on the Variability of Interest Rates*
This section documents the extent to which the variability of
interest rates has increased as a result of the switch to reservestargeting procedures.
The first subsection develops a framework for
interpreting the empirical results.
That is followed in the second
subsection by a description of the empirical findings for the federal
funds rate.
The third subsection investigates the extent to which other
interest rates have reflected the increased variability of the federal
funds rate.
A.
•
The Implications of the New Procedures for the Federal Funds Rate
For an extended period prior to October 6, 1979, the Federal
Reserve's daily operating procedures were geared toward
confining the
'
federal funds rate to a narrow trading range.!/
Any tendency for the
federal funds rate to move away from the desired level was resisted by
providing more or fewer nonborrowed reserves.
The Desk essentially was
indifferent about the relative provision of nonborrowed and borrowed
reserves and was prepared to alter the supply of nonborrowed reserves to
keep the funds rate trading near the level desired by the FOMC.
Figure 1 characterizes the interaction of supply and demand
for nonborrowed reserves in the funds rate targeting regime.
The demand
curve indicates that for a given discount rate rd and a given level of
required reserves RR there is an inverse relationship between the demand
!/ For background on Desk operations see "The Role of Operating Guides
in U.S. Monetary Policy: A Historical Review," by Henry Wallich and
Peter Keir, Federal Reserve Bulletin, Vol. 65 (September 1979), pages 679-91.
*
The primary author of this section was Dana Johnson of the Board staff.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-7-
Figure 1
Interest
Rates
Demand
.;.,
·Supply
R*
RR
Nonborrowed
Reserves
Figure 2
Interest
Rates
r
Supply
Demand
r
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
R*
RR
Nonborrowed
Reserves
-8for'nonborrowed reserves and the federal funds rate.
Presumably, an
increase in the discount rate would shift this demand curve upward and
an increase in the level of required reserves would shift this curve to
. h t.1/
t h e rig
The Desk's willingness to alter the provision of reserves
* is represented by the
to hold the funds rate at a desired level rf
horizontal supply curve.
As shown in figure 1, the Desk would supply R*
*
nonborrowed reserves to hold the funds rate at rf.
In this circumstance,
borrowing at the discount window would total RR-R* ignoring excess
reserves and carryover provisions.
Under its new procedures, Desk operations are designed to achieve
a targeted level of nonborrowed reserves deemed consistent with the desired
2/
path for the monetary aggregates.-
There are, however, limits beyond
which the funds rate is not allowed to fluctuate without the Desk showing
resistance.
regime.
Figure 2 represents diagramatically the reserves-targeting
The demand curve is the same as in figure 1.
The supply curve
now is vertical at the targeted level of nonborrowed reserves R* between
the upper and lower bounds for the funds rate.
trading range, E. and
r
At the limits of the
in fl.gure 2, the supply curve becomes horizontal,
indicating the Desk's willingness to alter its provisions of reserves
when those limits are reached.
Given the target for nonborrowed reserves
* the equilibrium federal funds rate in figure 2 is rf,
* and the level
of R,
of borrowing at the discount window consistent with this outcome again is
*
RR-R.
1/ Empirical estimates of the determinants of borrowing at the discount
window confirm this assertion. See, for instance, Peter Keir, "Impact of
Discount Policy Procedures on the Effectiveness of Reserve Targeting," in
this compendium.
~/ For a detailed description of the new procedures see the staff paper,
"The New Federal Reserve Technical Procedures for Controlling Money" (Board
of Governors of the Federal Reserve System, January 31, 1980).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-9Figures 1 and 2 were drawn as if there were a stable relationship
between the demand for nonborrowed reserves and the federal funds rate.
Experience indicates, however, that the demand curve is relatively erratic,
sometimes shifting substantially from day to day and from week to week as
the distribution of reserves among banks changes and as expectations about
the very-near-term outlook for the funds rate are revised.!/
This observa-
tion has implications for the variability of the funds rate under the two
operating regimes.
Prior to last October, shifts in the demand curve from
period to period resulted in a changed provision of nonborrowed reserves but
had little impact on the funds rate.
October 6.
Just the opposite is the case after
Assuming that the trading range for the funds rate does not
become a binding constraint, a shift in the elastic port1on of the demand curve
under current procedures would result in a changed funds rate but would
leave the relative provision of borrowed and nonborrowed reserves unaffected.
Thus, it appears that in response to nonsystematic shifts in demand the
funds rate would tend to be more volatile in the very short run under the
.
2/
reserves-targeting
proce d ures.-
There also is reason to believe that the new procedures are likely
to result in frequent systematic changes in the funds rate.
When the mone-
tary aggregates deviate from the desired path, the new procedures are
designed to automatically generate pressures that push money growth back
toward the intended path.
For instance, growth of the aggregates that was
!/ For a discussion of the sources of variance in the relationship between
money market interest rates and the demand for reserves, see Keir, "Impact
of Discount Policy Procedures."
2/ This conclusion follows if it is assumed that the demand for nonborrowed
reserves is at least as erratic after October 6, 1979, as before and that
errors in the supply of reserves by the Desk both before and after October 6
are small relative to the variance of demand.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-10-
faster than targeted would lead to an increase in required reserves
relative to the provision of nonborrowed reserves.
In terms of figure 2,
there would be a rightward shift of the demand curve and
equilibrium level of the funds rate.
thus a higher
Moreover, if the faster growth of
the aggregates tended to persist, there would be a progressive rightward
shift of the demand curve, putting increasing upward pressure on the
funds rate.
Under funds rate targeting, in contrast, a progressive
rightward shift in the demand curve did not result automatically in
upward pressure on the funds rate, although the directive did allow
the Desk, within relatively narrow ll.mits, to adjust the funds rate
target in response to growth of the monetary aggregates.
Consequently,
adjustments in the funds rate tended to be made in small but easily
perceived discrete jumps.
In sum, there are two dimensions to the increase in the variability of the federal funds rate:
nonsystematic movements due to the new
procedures and systematic movements reflecting the interaction of policy
and the behavior of the economy.
The increase in nonsystematic variability
may turn out to be a lasting phenomenon.
The sharp systematic movements
in the funds rate, on the other hand, were associated with the unusually
turbulent economic environment encountered this past year, and presumably
were not primarily a consequence of the new procedures.
Accordingly, in
the empirical section that follows, attempts are made to disentangle these
two types of variability in order to focus on the additional nonsystematic
variability that is more clearly attributable to the switch to the reservestargeting procedures.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-11-
B.
Empirical Findings on the Variability of the Federal Funds Rate
It is easy to document that the federal funds rate has been more
variable over the past year than in other recent years.
example, the data in Table 1.
Consider, for
On a daily basis, the standard deviation of
the change in the funds rate in the year starting. October 1979 has been at
least 50 percent larger than that in any other year since 1973. Likewise, the
1
standard deviations of weekly changes over the past year have been consid-
erably greater than those in previous years.
At a 95 percent confidence level,
each of the standard deviations shown for 1980 in Table 1 is significally
greater than the comparable standard deviations for each of the previous
~even years.
Thus, these data clearly illustrate that the funds rate has
~een more variable over the past year; but,of course, these measures of
variability reflect movements in the funds rate due both to the new
procedures and to the interaction of policy with the turbulent economic
conditions experienced this past year.
The reserves-targeting procedures have been ~arried out ~nan
environment of extremely sharp cyclical swings.
On a weekly average basis,
the funds rate rose about 7-1/2 percentage points between early October
1979 and early February 1980 ~d then fell 10-3/4 percentage points by
late July.
Between July and late December 1980, the funds rate
rose again about 10 percencage points.
When compared to previous interest
rate cycles, the range of movements of the funds rate over the past year
• '-was not unprecedented.
Durfhg the''tast rate·cycle,from'1973 to-1975,- the
funds rate rose and then fell about 8-1/2 percentage points.
Those swings
in the mid-seventies occurred, however, over a considerably longer period
of time than was the case this past year.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Thus, the swings in rates this
-12-
Table 1
Standard Deviation of the Federal Funds Rate
Daily
Changes
Mon.
Weekly Changes.Y
Tues.
Wed.
Thur.
Fri.
1973
.58
.39
.99
1.33
.34
.34
1974
.59
.so
.62
1.31
.39
.37
1975
.60
.26
.81
.91
.33
.32
1976
.25
.17
.17
.62
.14
.15
1977
.17
.14
.13
.30
.13
.15
1978
.18
.12
.16
.43
.16
.13
1979
.38
.23
.37
.63
.18
.21
1973 to 1979 average
.39
.26"
.46
.79
.24
.24
1980
.91
·1.42
1.37
1.79
1.15
1.31
1. Data are for October of the previous year through September of the,year
indicated.
2. Each series represents the change from the same day a week earlier.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-13-
past year were not markedly wider than those that have occurred before, but
they certainly occurred much more quickly.
Accordingly, any measure of
variability which fails to abstract from this cyclical experience is
bound to show an increase in volatility compared to earlier years.
One simple way to\abstract from cyclical volatility is to focus
on very short periods of time.
Before last October, the funds rate target
1
typically would be held constant for a given week, and over the past year,
the nonborrowed reserves target usually was not revised as the week
1/
progressed except for technical reasons.-
Thus, measures of variability
of the funds rate over a day or within a statement week should reflect
primarily the System's operating procedures.
As a measure of movements of the funds rate within a single day,
Table 2 shows monthly and quarterly averages of the daily trading range
for the funds rate.
The trading range was measured simply by taking the
difference between the highest and lowest rate at which trades were
reported to have taken place.
Table 2 shows average trading ranges for all days
of the week, for Wednesdays, and for days other than Wednesday.
Looking
first at the data that include all business days, there was a sharp
increase 1n the trading range last October, but a peak was not reached until
April,when short-tenn rates were near their cyclical highs.
Over the
summe~when the funds rate moved below the discount rate and was constrained
by the lower bound of the FOMC's desired range, the average trading range
tended to narrow and by September it was only slightly greater than was typical
over the first three quarters of 1979.
This fall and early this winter, the
trading ranges again widened appreciably.
1/
See Fred Levin and Paul Meek, "Implementing the New Operating Procedures
The View from the Trading Desk," in this compendium, for a discussion of how
the nonborrowed-reserves-targeting procedures have been carried out over the
past year.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-14Table 2
Average Daily Traaing Range of the Federal Funds Rate
(Percentage Points)
Period
Wednesday
All Days
Quarterly
.
Non-Wednesday
.
1978:
QI
QII
QIII
QIV
',94
1.11
1. 22
1.63
1979:
QI
QII
QIII
QIV
1. 80
1. 74
1. 65
4.00
6.20
5.28
5.08
8.56
. 65
. 84
. 77
2. 77
1980:
QI
QII
QIII
QIV
3.05
4.15
2.10
4.73
8.20
9.75
5.31
10.25
1. 75
2.64
1. 28
3.12
1979:
Oct.
Nov.
Dec.
4.05
4.40
3.54
9.63
9.56
6.50
2.41
3.11
2.80
1980:
Jan.
Feb.
Mar.
2.60
3.63
2.93
8.86
9.44
6.31
. 76
2.32
2.14
Apr.
May
June
5.95
3.90
2.59
15.85
8.31
5.09
3.04
2.87
2.00
July
Aug.
Sept.
2.26
2.08
1. 96
5.50
4.81
5.62
1. 31
1.43
1.10
Oct.
Nov.
Dec.
4.21
3.51
6.47
8.53
8.31
13.90
2.94
2. 14
4.28
2.56
3.97
3.96
5.09
' . 51
. 3~
.48
. 67
Monthly
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-15-
It is interesting to note that the trading range on days other
than Wednesdays narrowed substantially but even this summer continued to
exceed the average of the first three quarters of 1979.
However, on
Wednesdays, the last day of the statement week, the decline in the trading
range was relatively more pronounced, averaging this summer about what it
did "before October 1979.
There'is some sense to this result.
Even before
last October, the Desk had only a limited ability to affect the funds rate
on settlement days, particularly after about 1:00 p.m.,when it becomes
increasingly impractical to conduct open market operations.
On Wednesday
afternoons, the supply of nonborrowed reserves becomes fixed, and the
federal funds rate must adjust to clear the market.
Thus, even before
October 1979, the Desk by necessity lost some control of the funds rate
at the end of the statement week.
One might expect, therefore, that the
new procedures would not result in as marked an increase in the volatility
of the funds rate on settlement days as on other days of the week; and
indeed, that appears to be the case.
Table 3 displays a measure of the dispersion of the funds rate
over a statement week.
the tradlng range data.
The data in Table 3 exhibit a pattern similar to that of
There is a sharp increase in variability immedia-
tely following October 6, a peak in volatility this past spring, some
tendency for volatility to diminish over the summer, and a resurgence of
variability late in the year.
While this measure of dispersion has been
consistently higher than it was before October 1979, it is interesting
to note that in 1973 and 1974, it was higher than in subsequent years,
suggesting that this measure may be affected by sharp cyclical movements
even though it focuses on dispersion over one statement week at a time.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-16Table 3
Average Absolute Deviation of Daily Effective Federal Funds Rate
from the Statement Week Average
(Basis Points)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Period
' '
Annual Averages
1973
1974
1975
1976
1977
1978
"
29
22
16
7
7
9
Quarterly Averages
1979:
QI
QII
QIII
QlV
12
10
12
45
1980:
QI
QII
QIII
QIV
42
71
32
64
Monthly Averages
1979:
Oct.
Nov.
Dec.
51
46
37
1980:
Jan.
Feb.
Mar.
29
58
40
Apr.
May
June
83
76
53
July
Aug.
Sept.
26
39
29
Oct.
Nov.
Dec.
37
64
90
-17-
A pattern apparent in these measures of intraday and intraweek
variability is that over the past year they have tended to be relatively
large at the times when the funds rate also was showing sharp cyclical
movements.
One possible interpretation is that even though those measures
are for very short periods of time, they are reflecting the policy-related
shifts in supply by the Desk.
In my judgment, however, another, perhaps
equally plausible,explanation is that the relationship between the demand
for reserves and the funds rate becomes more erratic at times of rapid
cyclical changes in rates.
Early in a statement week, the demand for
nonborrowed reserves, and thus also the level of the funds rate,depends
importantly on market participants' expectations abcut the very-near-term
outlook for the federal funds rate.
At times of rapid changes, there is a
greater potential for market participants to misperceive the level of the
funds rate consistent with new reserve conditions.
In such circumstances,
Desk actions to affect the availability of nonborrowed reserves sometimes do not
have the expected effect on the funds rate until late in the statement
week,when actual reserve availability has a more dominant influence on
the level at which funds trade.
An important likely implication is that
the additional volatility attributable to the change in operating procedures
will be greatest at times of sharp changes in reserve availability associated with significant deviations in the monetary aggregates from the
FOMC's desired path.
Various measures of the variability of the funds rate based on
statement-week averages are shown in Table 4.
Summary statistics are
shown for the intervals from January 1968 to September 1979, October 1979
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-18-
Table 4
Standard Deviation of the Weekl) Funds Rate
(Percentage Points
Item
Jan. 68
to
Sept. 79
Oct. 79
to
Sept. 80
Jan. 73
to
June 75
Jan. 69
to
Dec. 70
Levels
2.23
2.91
2.21
1.28
De trended Levels
1. 77
1.90
1.77
1.25
<llanges
.37
.97
.35
.40
Changes of Detrended Levell
. 23
. 62
.28
.37
Deviation from 3-week
centered moving average
.13
.37
.17
.21
Deviation from 5-week
centered moving average
.17
.55
.19
.27
Deviation from 7-week
centered moving average
.19
.73
.23
.29
Deviation from 9-week
centered moving average
.20
.92
.25
.29
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-19-
to September 1980, January 1969 t~ December 1970, and January 1973 to
June 1975.
The latter two intervals were chosen to include relatively
sharp cyclical movements in interest rates somewhat akin to the experience of the last year.
Throughout the period from January 1968 to
September 1979, the Federal Reserve operated under a funds rate targeting
procedure.
•
The first two lines of Table '4 show the standard deviation of
the level c£ the funds rate using unadjusted and detrended data~
In
both cases, the greatest variability is apparent over the past year.
The
detrended data show, however, a less decided increase in variability this
past year.
The third and fourth lines of Table 4 show the standard devia-
tion for the change in the statement-week average of the funds rate based on
unadjusted and detrended levels.
Again, these statistics indicate that the
variability of the funds rate increased subsequent to October 6.
As discussed above, the increased interweekly variability of the
funds rate in level or first-difference form rs a reflection partly of the
sharp cyclical swings in the funds rate that have occurred since last
October.
In an attempt to focus on the variability due to the new proce-
dures, deviations from centered moving averages were calculated.
The
moving averages were intended to represent the cyclical movements in rates.
The deviations from the moving averages are meant to represent unsystematic
movements in the funds rate.
As shown in the bottom 4 lines of Table 4,
these deviations are,.substantiaH.-y,.more,.variable ,subsequent to ·October 6, ..... ..,,.. "•·•,
even compared to those in the other periods of relatively rapid cyclical
movements in interest rates.
For instance, the data based on three-week
moving averages appear to be two to three times more variable this past
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-20-
year than in any of the other periods shown in Table 4.1/
These data
strongly suggest that there has been a marked increase in interweek
volatility due to the new operating procedures.
C.
Variability of Longer-Term Interest Rates
While conventional theories of the term structure of interest
ratesemphasize to varying degrees the importance of factors such as
.
liquidity premiums and market segmentation, they all have as a common
element the proposition that expectations about the future course of veryshort-term interest rates are of fundamental importance.in the determinations of longe=-term rates.:?:./
Thus, there is general agreement that
expectations about the future course of the federal funds rate are a
central feature of interest rate determination in the United States.
For
instance, theories based on expectations would assert that the rate on a
three-month Treasury bill embodies, among other things, the market expectations about the likely course of the federal funds rate over the
three-month maturity of the bill.
An important implication of these theories is that the greater
volatility of the federal funds rate docunented above has the potential
for making longer-term rates more volatile.
The degree to which greater
variability of the funds rate is reflected in longer-term rates should
depend importantly on how sensitive expectations are to unexpected move~ents
in the funds rate.
Rational expectations, for instance, imply that market
·"!/'The standard deviations·of the sbattstics based on moving~averages·were
significantly greater this past year than in the other periods shown in Table
4 at a 99 percent confidence level.
:?:.I
For a review of the literature on the term structure of interest rates,
see J.C. Dodds and J.L. Ford, Expectations. Uncertainty and the Term Structure
of Interest Rates (Barnes and Noble, 1974).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
,,_,
-21participants would efficiently use all information that is economically
available.
Thus, a wide variety of information in addition to the actual
behavior of the funds rate would tend to be used in formulating expectations about the future course of the funds rate.
In particular, it would
be expected that the change in operating procedures last year would be
taken into account by market participants and that their response to the
behavior of the funds rate might appear to be somewhat altered.
But even
if it were widely agreed that expectations are rational, it would remain
-unclear to what extent movements in the funds rate would be discounted
and therefore whether interest rates would appear to be more or less
responsive to the funds rate.
The nature of expectations is inherently
an empirical issue.
The evidence presented in Tables5 through 9 based on weekly average
data suggests that the increased variability of the funds rate has had an impact
•
on other rates.
Interest rates on Treasury securities across the maturity
r
spectrum have shown greater variability subsequent to October 6.
Levels,
changes, and deviations from moving averages all have been more variable
over the past year than in any of the other periods shown in the tables.
As predicted by expectation theories of interest rates, the variability of
rates as measured in percentage points diminishes as maturity lengthens.
(See Table 9 in particular.)
In relative terms, however, there was a
marked increase in the volatility of rates on Treasury securities throughout
the range of available maturities.
It is conceivable that rates on Treasury securities were more
volatile this past year for reasons unrelated to the variability of the
funds rate.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Therefore, in assessing the influence of the funds rate, it is
-22-
Table 5
Standard Deviation of Three-month Treasury Bill ~ate
(Weekly data, percentage points)
Jan. 68
to
Sept. 79
Item
Oct. 79
to
Sept. 80
Jan. 73
to
June 75
I
Jan. 69
to
Dec. 7p
Levels
1.57
2.39
1.16
.10
Detrended Levels
1.17
1.53
.93
• 68
Changes
.20
. 62
.33
.15
Deviation from 3-week
centered moving average
. 08
.22
.13
.07
Deviation from 5-week
centered moving average
.13
.:n
.21
.10
Deviation from 7~week
centered moving average
.16
.43
.28
.12
Table 6
Standard Deviation of 52-week treasury Bill Rate
(Weekly data, percentage points)
Item
Jan. 68
to
Sept. 79
Oct. 79
to
Sept. 80
..'.Jan. 73
Jan. 69
to
to
June 75
Dec. 70
Levels
1.39
1.92
.97
.68
De trended Levels
1.00
1.24
. 78
. 66
Changes
.16
.48
.23
• 15
Deviation from 3-we.ek
centered moving average
.06
.15
.09
.06
Deviation from 5-week
centered moving average
.10
.22
.14
.09
Deviation from 7-week
centered moving average
.13
.32
.18
.11
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-23-
Table 7
Standard Deviation of Five-year Treasury Note Rate
(Weekly data, percentage points)
Jan. 68
to
Sept. 79
Item
Jan. 73
to
June 75
Oct. 79
to
Sept. 80
Jan. 69
to
Dec. 70
1.03
1.26
.62
.65
De trended Levels
. 65
.81
.42
.56
Changes
.12
.41
.14
.13
Deviation from 3-week
centered moving average
.05
.14
.05
.05
Deviation from 5-week
centered moving average
.07
.22
.08
.08
Deviation from 7-week
centered moving average
.10
,30
.11
.10
Levels
•
Table 8
Standard Deviation of 20-year Treasury Bond Rate
(Weekly data, percentage points)
Item
Levels
Jan. 68
to
Sept. 79
Oct. 79
to
Sept. 80
Jan, 73
to
June 75
Jan. 69
to
Dec . 70
1.08
. 84
.56
.40
De trended Levels
.42
.48
.29
.28
Changes
. 09
.29
.11
.11
Deviation from 3-week
centered moving average
,03
.11
.04
.05
Deviation from 5-week
centered moving average
.05
.15
.07
.07
Deviation from 7-week
centered moving average
07
.21
.08
09
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-24Table 9
Standard Deviation of Changes in Rates on Various Treasury Securities
(Weekly data, percentage points)
Jan. 68
to
Sept. 79
Oct. 79
to
Sept . 80
Jan. 73
to
June 75
1-month Bill
.21
. 85
.33
.17
3-month Bill
.20
.62
.33
.15
6-month Bill
.18
.56
.27
.15
52-week Bill
.16
.48
.23
.15
3-year Note
.14
.46
.16
.15
5-year Note
.12
.41
.14
.14
10-year Note
.09
.32
. 08
.13
20-year Bond
.99
.29
.11
.11
Type of security
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Jan. 69
to
Dec. 70
-25revealing to examine the correlation of the funds rate with other rates.
Correlations between the three-month bill rate and the funds rate using
wee1tly a~nragP. data are shoiro in Table 10.
Similar correlations of the
funds rate with several longer-maturity Treasury securities are reported
in Tables 11, 12, and 13.
Focusing on the results in Table 10 for the funds and three-month
bill rates, the correlation of the levels of rates shown in line 1 was
essentially the same before and after last October.
This finding is of
major importance, as it suggests that the change in procedures has not
markedly altered the'relationship between the level of the funds rate and
the levels of other rates.
down with the funds rate.
Longer-term rates still tend to move up and
This linkage is, of course, a critical aspect
of the Federal Reserve's monetary control mechanism under either the
funds rate or the reserves operating procedures.
If this relationship
had been loosened markedly by the change in procedure, it could have
impaired the Federal Reserve's ability to influence the monetary aggregates.
Interestingly enough, the correlation between changes in rates
appears to have increased over the past year as is shown in line 2 of
Table 10.
Presumably, this result reflects pr1.marily the response of the
three-month bill rate to the sharp cyclical movements of the funds rate over
the past year.
This interpretation is supported by the observation that
the correlation between changes also was relatively high in 1973-75,
another period characterized by sharp cyclical swings in the funds rate.
Nevertheless, the high correlation of changes since October is somewhat
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-26'Tab'le 10
Correlation of the Funds Rate with the Three-Month Treasury Bill Rate
·(Weekly data)
Item
Jan. 68
to
Sept. 79
Oct. 79
to
Sept. 80
Jan. 73
to
June 75
Jan. 69
to
'Dec. 70
Levels
.93
.92
.89
.82
Changes
.11
• 72
.62
.09
Deviations from 3-week
centered moving average
-.02
.56
-.06
-.02
Deviations from 5-week
centered moving average
.01
.62
-.07
-.07
Deviations from 7-week
centered moving average
.04
.64
-.03
.06
Deviations from 9-week
centered moving average
.08
. 72
-.02
.05
Table 11
Correlation of the Funds Rate with the 52-Week Treasury Bill Rate
(Weekly data)
Jan. 68
Oct. 79
Jan. 73
Jan. 69
Item
to
to
Sept. 79
Sept. 80
June 75
Dec. 70
Levels
.90
.89
.91
.82
Changes
.22
.65
.28
.07
Deviations from 3-week
centered moving average
.02
.43
.04
-.05
Deviations from 5-week
centered moving average
.10
.53
.19
.01
Deviations from 7-week
centered moving average
.17
.61
.28
.03
Deviations from 9-week
centered moving average
.23
.65
.33
.08
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
to
to
-27-
Table 12
Correlation of the Funds Rate with the Five-Year Treasury Coupon Rate
(Weekly data)
Item
Jan. 68
to
Sept. 79
Levels
Oct. 79
Sept. 80
Jan. 73
to
June 75
Jan. 69
to
Dec. 70
.68
. 74
.48
.55
Changes
.19
.48
.22
.16
Deviations from 3-week
centered moving average
.01
.21
-.05
.03
Deviations from 5-week
centered moving average
.10
.41
.15
.05
Deviations from 7-week
centered moving average
.22
.Lis
.27
.20
Deviations from 9-week
centered moving average
.29
.46
.33
.27
to
Table 13
Correlation of the Funds Rate with the 20-year Treasury Bond Rate
(Weekly data)
Jan. 68
to
Sept. 79
Oct. 79
to
Sept . 80
Jan. 73
June 75
Jan. 69
to
Dec. 70
Levels
•49
.53
.29
.oa
Changes
.09
.39
.11
.07
Deviations from 3-week
centered moving average
-.01
.18
-.04
.03
Deviations from 5-week
centered moving average
.01
.31
.04
.01
Deviations from 7-week
centered moving average
.08
.28
.10
.11
Deviations from 9-week
centered moving average
.10
.26
.09
.13
Item
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
to
-28surprising.
Given the greater short-term variability in the funds rate
created by the new operating procedures, it might have been expected that
market participants would have reacted more cautiously to changes in the
funds rate.
In any case, the relatively high correlations between changes
in rates are another indication that the new Federal Reserve procedures
did not weaken the tie between the funds rate and longer-term rates.
The results presented in the bottom four lines of Table 10
were even more surprising.
The data indicate that deviations from moving
averages for the funds rate and the three-month bill rate were essentially
uncorrelated previous to last October but quite highly correlated
afterward.
As was indicated above, the moving averages we~e used to
approximate the cyclical movements of interest rates.
The deviations from
the moving averages were intended to represent unsystematic movements in
interest rates.
My initial expectation was that if this technique were
successful in isolating the cyclical movements in rates, the deviations
for the funds rate would not show much correlation with other rates,
particularly since last October.
The results shown in Table 10 appear to
indicate, however, that abstracting from cyclical movements, interest
rates on three-month Treasury securities have been more, not less, correlated
with the funds rate over the past year.
Moreover, as is summarized in
Table 14, this result also holds for Treasury securities throughout the
maturity spectrum.
Even the rates on instruments with maturities as long
as 20 years appear to have been moving closely with the federal funds
rate subsequent to October 1979.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-29-
Table 14
Correlation of the Funds Rate with the Rates on Various Treasury Securities
(Weekly data, deviations from three-week centered moving averages)
Jan~68
to
Sept. 79
Oct. 79
to
Sept. 80
1-month Bill
.03
.24
.03
.04
3.'.:month Bill
-.02
.56
-.06
-.02
6-month Bill
-.02
.56
-.01
-.10
52-week Bill
.01
.43
.04
-.05
3-year Note
.01
.30
-.05
-.001
5-year Note
.01
.21
-.05
.03
10-year Note
-.03
.21
-.04
.04
20--year Bond
-.01
.18
-.04
.03
Type of security
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Jan. 73
to
June 75
Jan. 69
to
Dec. 70
-30-
As can be seen in Table 15, the relatively high correlations of
these deviations from moving averages that a~ apparent for various maturity
Treasury securities do not merely reflect changed expectations for the very
near term.
To remove the impact of changes in relatively near-term exp~c-
tations, forward rates implicit in the structure of rates on various
maturity Treasury bills were calculated.
The implied forward rates for a
, three-month Treasury bill purchased in three months and for a six-month·Trea~
sury bill purchased in six months' both exhibited essentially th.e same pattern of
correlation with the funds rate as did yields in the cash markets.
In
particular, the deviations from moving averages for the two forward rates
both exhibited a higher correlation with comparable deviations for the
funds rate after October 6, 1979.
Thus, it appears that over the past
year expectations about the future course of the funds rate beginning
three months or six months into the future have been sensitive to what appears,
at least in retrospect, to have been nonsystematic movements in the funds
rate.
The correlation coefficients presented in the tables indicate
the degree to which various interest rate series have tended to move
together but do not indicate the relative size of such movements.
To
assess this latter characteristic, simple regressions were run with the
rate on Treasury securities of various maturities as the dependent variable
and the federal funds rate as the independent variable.
Using weekly
average data and expressing interest rates as deviations- from moving
averages, it was found that over the past year a change of 100 basis points
in the funds rate has been associated on average with a change of 34 basis
points in the three-month bill rate.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The comparable responses for the 52-week
-31-
Table l5
Correlation of the Funds Rate with Implied Forward Rates
(Weekly Data)
Item
Jan. 68
to
Sept. 79
Oct. 79
to
Sept. 80
Jan. 73
to
June 75
Jan. 69
to t-l
Dec. 70
3-month Rate for Delivery
in 3 months
Levels
. 91
.90
. 92
.80
Changes
.17
.66
.25
.oo
-.001
.43
.04
-.14
Levels
.85
.84
.84
. 79
Changes
.20
.49
.29
.07
Deviations from 3-week
centered moving averages
.05
.20
.07
.003
Deviation from 3-week
centered moving averages
6-month Rate for Delivery
in 6 months
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-32-
bill, the 5-year note, and the 2O-year bond were 17, 9, and 6 basis points
respectively.
Thus, as predicted by expectations theories of the term
structure of interest rates, relatively short-term rates showed the largest
reaction to deviations in the funds rate from centered moving averages.
Comparable regressions run with data for the pre-October period consistently
indicate that there was no statistically significant response of Treasury
rates to deviations in the funds rate as would be expected given the finding
reported above that such deviations were essentially uncorrelated.
Thus,
evidence from regressions as well as from simple correlations suggests that
subsequent to October 1979, rates on Treasury securities tended to be more
closely related to nonsystematic movements in the funds rate.
Several
complementary explanations for this unexpected finding are given below.
The first explanation focuses on the difficulty of distinguishing
between cyclical and unsystematic movements in the funds rate under the
new operating procedures.
Under the old funds rate targeting procedures.
market participants usually knew with considerable precision the System's
current funds rate target.
By observing the Desk's intervention points,
knowledgeable observers could easily see when the funds rate objective was _changing.
The average funds rate showed some variance from week
to week, but usually it was easy to distinguish unsystematic movements from
policy-related changes in the funds rate.
The situation is now, of course, quite different.
It is often
difficult for market participants to identify when,and particularly the
degree to which,the interaction of Federal Reserve policy with the behavior
of the economy is likely to have a lasting effect on the funds rate.
Thinking back to figure 2 presented earlier in the paper, the public
typically does not know now whether an observed change in the funds rate
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
,
-33-
is reflecting shifts in the supply or the demand curve or both.
In this
continuing state of uncertainty, it is understandable that risk-averse
market participants typically would show some reaction to movements in the
funds rate, even though in retrospect they often would realize that the
cyclical pressures on the funds rate had been misperceived to some extent.
Of course, when it does become apparent that rates had over- or underreacted to a change in reserve availability, the funds rate and other rates
would tend to move together to their appropriate levels.
Thus, after the
fact, when the cyclical pattern of interest rates became more apparent, there
would appear to be parallel deviations from the cyclical trend for both
the funds rate and other market interest rates.
A second explanation for the higher correlation of changes in
market rates with changes in the funds rate is based on the more immediate
response to unexpected changes in the monetary aggregates that typically
occurs under the reserves-targeting procedures.
For instance, growth in
the monetary aggregates fast\ than desired would result in an increase in
required reserves relative to the Desk's provision of nonborrowed reserves.
Again returning to figure 2, the increase in required reserves would cause
a rightward shift in the demand curve and thus a higher equilibrium federal
funds rate.
Presumably, other interest rates also would increase in
response to this rise in the funds rate.
Under the funds rate targeting regime, in contrast, there was no
automatic mechanism that assured a prompt response in the funds rate target
to unexpected growth in the monetary aggregates.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Market interest rates
-34-
often responded to the monetary aggregates data in anticipation of policy
responses, but the System sometimes failed to validate market expectations
of a change in the funds rate target for significant lengths of time.
Thus,
uncertainty about the precise timing or size of policy changes tended to
reduce the correlation between the funds rate and other rates under the old
operating procedures, when rates are expressed as first differences or
deviations from moving averages.
As has been noted above, however, the
correlations based on levels of rates were essentially the same before and
after October 6, suggesting that on average the response of longer-term
rates to the funds rate has been comparable even though the timing of
movements may have been less close in the earlier period.
A third explanation for the higher correlation of rates hinges
on the behavior of member bank borrowing at the discount window.
Because
banks are limited as to how often they may borrow_at the discount window,
banks must make some judgement about when it will be most advantageous to
borrow.l/
Thus, if banks come to expect that the funds rate will be rising
relative to the discount rate in the near future, they might be somewhat
more reluctant to borrow in the current period in hopes of borrowing when
there is a greater rate advantage.
This increase in the reluctance to borrow
for a given spread between the funds rate and the discount rate would be
represented in figures 1 and 2 as an upward shift in the demand for nonborrowed reserves.
Under the funds rate targeting regime, such pressures
were offset by changing the provision of nonborrowed reserves,and thus the
change in expectations would not have much effect on the funds rate.
In
the new reserves-targeting regime, in contrast, such pressures on the funds
rate would not be offset and could result in an increase in the funds rate.
!/ Frequency guidelines in most Federal Rese~ve districts are in terms of
hnT,T nf'f-on .<a h.<anlr m.<au
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
hn't",.nT.T
in
.<a
t-hirt-i:>.P.n-week interval.
-35-
Carryover provisions also give banks s,ome ability to take
advantage of anticipated interweekly movements in the funds rate.
For
instance, if the funds rate generally were expected to decline in the
upcoming week, banks would have an incentive to carry over the maximum
reserves deficiencies permitted.
In terms of figure 2, the demand curve
would shift down, putting immediate downward pressure on the funds rate.
Such pressures, of course, would have been resisted by the Desk if they
had emerged in the context of the funds rate targeting regime but are
not resisted under current procedures.
Thus, if banks do tend to alter
their borrowing at the discount window and their carryover of excess
reserves in response to anticipated movements in money market rates, a
change in interest rate expectations in the context of reserve targeting
could affect the funds rate along with other short-term rates, tending
to make them somewhat more highly correlated.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-36III.
Reactions in Financial Markets to the Increase in Interest Rate Variability
The increase in the variability of interest rates associated with
the Federal Reserve's new operating procedures exposes many participants in
financial markets to greater risks.
It is natural to expect that financial
practices will evolve as market participants attempt to shield themselves
from the adverse consequences of rate variability.
In this section, various
aspects of major financial markets are examined in hopes of identifying
responses to the heightened variability of interest rates.
begins with an examination of liquidity premiums.
is a report on the Treasury securities market.
This section
The second subsection
This is followed by a related
piece on underwriting spreads in the corporate bond market.
The fourth sub-
section summarizes recent developments on the supply side of the mortgage
market.
Subsection E discusses the adjustments made by commercial banks
to shield themselves from volatility.
The final piece relates the growth
of futures markets to recent developments.
On the whole, the evidence pre-
sented below indicates that the adjustments to the new policy regime were
relatively limited over the past year.
A.
Liquidity Premiums on Government Securities before and after October 197'
Theories of the term structure of interest rates often assert that
longer-term interest rates embody risk premiums that compensate investors for
the capital risk they are exposed to when they make multiperiod commitments
In an environment of increased interest rate volatility, it might be expected
that risk premiums would increase, since uncertainty about the future course
*The primary author of this section was Jim O'Brien of the Board staff.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-37of rates is greater.
Thus, there is reason to believe that the switch to
reserves-targeting procedures, which is widely perceived to have made interest
rates more volatile, might have led to an increase in risk premiums.
As
is reported below, however, there is little evidence at this point that
such an increase has occurred.
Liquidity premiums are associated with interest rates on forward
loans--commitments to lend at a given interest rate at a certain time in
the future.
There are, of coursej markets in which forward commitments
are made and traded.
However, spot rates where the term to maturity
extends for more than one period also contain implicit forward rates.
A two-period spot rate,
for example, may be thought of as a one-period
spot rate followed by a one-period forward rate with a "delivery term" of
one period.
The liquidity premium embedded in a forward rate for a given
security typically is defined as the difference between the forward rate
and the interest rate expected to prevail on such a security at the
appropriate time in the future.
That is, if Ln
denotes the liquidity
t,m
premium, then
(1)
where Fn
is the forward rate at time t with a delivery term of m and a
t,m
n
maturity of n, and E(Rt+m) is the market expectation of the interest rate
to prevail at time tfm.
The liquidity premium, as defined in (1) may vary
for different maturity and delivery terms (n and m) as well as with the
time at which the forward rate is quoted (t).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Typically, however, the
-38-
simplifying assumption is made that the premium for a given maturity and
delivery tent is constant through time.
Liguidity_,Premiums for Treasury bills. To estimate premiums
implicit in the spot market quotations for Treasury bills, one-month (28 days)
returns were calculatedforbills with maturities of one month (28 days) to
thirteen months (364 days).
1
For example, the one-month return for a six-month
bill is simply the actual return that could have been obtained by purchasing
a six-month bill and then selling it one month late.
Liquidity premiums
then were estimated by calculating over various sample periods the average
epread between the one-month return on n-month bills (n~l) and the known
return that could have been obtained by purchasing a one-month bill and
holding it to maturity.
Defining this premium as L1
the formula for
n-1,
estimation is
(n~2, ... , 13)
(2)
n
is the one~month return on then-month
t, 1
where Tis the sample size, R
bill,and R! is the return on the one-month bill.
It can be shown that this estimate of the liquidity premium is
equal to the true liquidity premium plus the average error in market
forecasts of the one-month returns on multimonth bills.
Thus, on the
"1
assumption that market forecasts are unbiased, L
is an unbiased
n-1
estimate of the true liquidity premium, and this estimate will have a
standard error of (n-1)/T times the standard deviation of the market
forecast errors.
1. The maturity structure of outstanding bills dictated the use of maturities
with 28-day intervals. The returns were calculated using bid-side prices on
Treasury bills. The monthly returns were calculated assuming continous compounding and all returns, and hence liquidity premiums are annualized.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-39Using month-end quotations for various Treasury bills, liquidity
premiums were estimated for three sample periods:
January 1973-June 1975
(32 monthly observations), July 1975-September 1979 (55 observations),
and October 1979-0ctober 1980 (14 observations).
The results are presented
in Table 16 with the standard error of each estimate in parentheses.
As
can be seen in the table~ the premiums estimated for the most recent period
often are positive but have large standard errors and are noc significantly
differen~ from zero at usual significance levels.
The large standard errors
imply that market forecasts were very inaccurate during the past year. 1
Thus, the premiums cannot be estimated very precisely.
For the two earlier
periods, the market forecasts were subJect to smaller absolute errors~
giving more precision to the estimated premiums.
Still, many of the
estimated premiums for the two earlier periods also are not significantly
different from zero.
Finally, it might be noted that the premiums do not
exhibit eny clear tendency to increase for Treasury bills of longer
maturity.
In Table 17 the differences between the estimated premiums
for the post-October 1979 period and each of the two earlier periods are
presented.
The estimates of premiums for the past year tend to be larger
than those of the two earlier periods for the shorter maturities but
srlklller for the longer maturities.
For all of the maturities, however,
the differences between the recent premiums and those of the two earlier
periods are not significantly different f~om zero even at low significance
levels.
Thus, these results provide little evidence that liquidity
1. The implied market forecast errors are obtained by multiplying the
~eported standard errors (in parentheses) by T/n-1.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-40Table 16
Liquidity Premiums on Treasury 1-Month Forward Rates
with delivery terms of 1 to 12 Months 1/
(Percentage points)
Delivery term of
21
Forward Loan (Months)-
Jan. 1973
-June 1975
July 1975
mSept. 1979
October 1979
-October 1980
1
0.22
(0.20)
0.10
(0.05)
1.17
(0.68)
2
0.30
(0.36)
0.24
(0.10)
1.04
(1.26)
3
(0.56)
(0.45)
0.48
(0.12)
1.14
(1.59)
4
0.68
(0 .64)
0.68
(0.20)
0.92
(2 .24)
5
(0.50)
(0.75)
a.so
(0.25)
0.75
(2. 70)
6
-0.24
(0.84)
0.06
(0.30)
-0.54
(3.00)
7
0.56
(1.05)
0.70
(0.35)
0.56
(3.64)
8
0.40
(1.12)
0.56
(0.40)
-0.32
(3. 76)
9
0.45
(1.26)
o.s4
(0.45)
(4.50)
10
0.20
(1.50)
o.so
(0.50)
-0.40
(5.10)
11
0.22
(1.54)
(0.55)
(0.55)
0.11
(5.39)
-0.24
-.12
(0.60)
(1.68)
1/ Standard errors of the sample means are in parentheses.
"f_l One month equals 28 days.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
12
o. 72
-0.36
(5.76)
-41Table 17
Differences in Liq_uidity Premiums
l/
on Forward Rates with Delivery Terms of 1 to 12 Months(Percentage points)
Delivery Term of
2/
Forward Loan (Months)-
Oct. 1979-0ct. 1980
minus July 1975-Sept. 1979
1
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
0.94
1.07
(0.68)
(O. 71)
2
0.78
(1.26)
0.78
(1.32)
3
0.66
(1.59)
0.57
(1.65)
4
0.24
(2 .28)
0.49
(2.32)
5
0.25
(2. 70)
0.25
(2.80)
6
-0.60
(3.00)
-0.30
(3.12)
7
-0.14
(3.64)
(3. 78)
8
-0.88
(3.76)
-0. 72
(3. 92)
9
0.18
(4.59)
0.27
(4.68)
10
-0.80
(5.10)
-0.50
(5.30)
11
-0.55
(5.39)
-0.11
(5.61)
12
-0.24
(5.88)
-0.36
(6 .OO)
1/ Standard errors in parentheses.
One month equals 28 days.
"'%../
Oct. 1979-0ct. 1980
minus Jan. 1973-June 1975
o.oo
-42premiums have increased over the past year.
Moreover, the same conclusion
emerged when liquidity premiums were estimated for the future delivery of
Treasury bills with maturities of two to four months.
This conclusion
is, of course, highly tentative given the difficulty of estimating
premiums over a period as short and as volatile as this past year.
Liquidity Premiums for Treasury Coupon Issueso
Estimating
JS
more
difficult for premiums associated with rates on Treasury coupon issueb
than for Treasury bills.
One aspect of coupon issues that creates con-
siderable complications is the semi-annual payment of interest.
Each
payment presumably should be discounted at a rate equal to that on a
discount loan whose maturity matches the duration of time elapsed before
the coupon is paid.
Consequently, the yield to maturity on a coupon
security should be viewed as reflecting a type of "averaging" of discount
rates for loan maturities appropriate to the security in question.
Thus,
each of the discount rates embedded in the rate on the coupon security
will have its own liquidity premium.
Nevertheless, the liquidity premium
on a coupon security still can be defined as the excess of the issue's
yield to maturity over the average expected yield on investments in
one-period securities for the time during which the coupon is outstanding.
In principle, the liquidity premium on the coupon issue co~ld
be estimated by estimating the premiums in each of the discount ratei
embedded in the coupon security's rate in a manner analogous to that
used for Treasury bills.
implement.
This process, however, would be difficult to
For this work, liquidity premiums were estimated as the
average spread between the return for a one-month holding period on certain
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-43Treasury coupon issues and the return on the one-month bill.
That
is, for each of the three sample periods, we estimated
-n -1
•~-R,
(5)
where
Hi
.,.:,Q
is the sample mean of the one-month returns on Treasury coupons
-1
with an n-period maturity and R is the average return on the one-month
bill for the same period.
The estimated premiums for coupon maturities of 5, 10, and 15
years are presented in Table 18.
As indicated in the table, the premiums
are very small, mostly negative,and quite insignificant.
1
Moreover, there,
is no indication of the premiums tending to increase in the post-October 1979
period.
These results are in general agreement with the results regarding
the bill premiums in that there is little evidence of a significant increase
in li~uidity premiums in the post-October 1979 period.
Of course, the
difficulty of obtaining precise estimates in a period of large forecast
errors again should be noted.
There is
another difficulty in estimating premiums on coupon
issues that might account for lhe tendency of the estimates of the premiums
to be less than zero.
Some of the income obtained from coupon issues
purchased at prices below par is likely to be taxed as capital gains rather
than as ordinary income.
As securities selling below par mature, their
prices have a tendency to increase toward their par value.
These price
1. The coupon maturities actually varied from month to month for a given
category (for example, for the five-year category) because the same maturity was not
always available on a monthly basis. The average of the maturities for each
of the three categories was about 4-3/4, 9-1/2, and 14 years, respectively,
with respective ranges of 0-6 months, 0-9 months, and 0-2 years.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-44Table 18
Liquidity Premiums on Treasury Coupon Issues
(Percentage points)
Years to
Maturity
Jan. 1973June 1975
July 1975Sept. 1979
Oct. 19790ct. 1989
5
-.04
(.06)
-.07
(. 06)
-.29
(. 37)
10
-.02
(.03)
-.01
(.07)
-.02
(. 36)
15
-.05
(.07)
.01
(. 07)
-.21
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
(. 38)
-45appreciations are subject to the capital gains tax advantages.
The price
appreciation on Treasury bills, in contrast, is taxed as ordinary income.
Since investors presumably are sensitive to after-tax yields, coupon
issues selling at a discount tend to show a before-tax yield that, ceteris
~aribus, is less than that on a Treasury bill. 1
Therefore, the estimated
premiums for coupons in Table 18, which are spreads between before-tax
coupon and bill yields, may tend to have a negative bias.
Whether the
average size of this bias might have changed this past year is uncertain.
Taken as a whole, the evidence presented in this section gives
little indication that liquidity premiums have widened subsequent to
October 6, 1979.
As was noted several times, the relatively short period
of time since the new procedures were introduced makes empirical estimates
of risk premiums particulary imprecise.
Even if considerably more data
wet~ available, however, it is not clear that the empirical results would
be ,~y more conclusive.
The nature and stability of liquidity premiums
have long been issues of considerable dispute, both on conceptual and on
empirical grounds.
B.
2
Treasury and Agency Security Markets*
The secondary market for Treasury and Agency securities is an
over-the-counter market maintained by a variety of dealers.
The largest
and most active of these dealers, designated primary dealers, regularly
1. A detailed analysis of the effects of the capital gains tax provisions
on ½efore-tax yields on coupon securities of various maturities is presented in A.A. Robichek and W.D. Niebuhr, "Tax-Induced Bias in Reporting
Treasury Yields," Journal of Finance, Vol. 25 (December 1970), pp. 1081-90.
2. For discussions of this dispute see Dodds and Ford, Expectations,
Uncertainty and the Tenn Structure of Interest Rates, chap. S, and
H.A.J. Green, "Uncertainty and the 'Expectations Hypothesis'," Review of
Economic Studies, Vol. 34 (October 1967), pp. 387-98.
* The primary authors of this section were Ed Stevens and Bill Gavin of
the Cleveland Federal Reserve Bank staff.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-46report a variety of information to the New York Federal Reserve Bank.
Based
on this as well as some other sources of data, this section examines several
aspects of dealer behavior that might be sensitive to an increase in interest
rate volatility.
Currently, there are 34 primary dealers reporting to the
Federal Reserve Bank of New York.
and 1 began.
During 1980, 4 firms stopped reporting
Over the previous 10 years, 23 £inns began reporting and
6 firms discountinued reporting.
The relatively large number dropping out
of the market in 1980 may be an indication that success in the dealer
business has become more uncertain.
Des?ite the net loss of three reporting dealers, the volume of
dealer transactions ln the market for Treasury securities was 4? percent
larger in the year after October 6, 1979, than in the year before (Table 19).
At the same time, dealers' gross positions in Treasury securiti~s were
29 percent larger on average this past year so that average daily turnover
rose from 68 percent to 75 percent of gross positions.
Turnover of gross
positions in Agency securities increased more substantially, from 18 percent
to 35 percent, reflecting in larger part a 31 percent reduction in gross
positions.
These large changes in transactions and gross positions did
I
not result in any substantial change in dealers risk exposure as mea~~red
by net cash positions.
Net cash positions in both Treasury and Agency
securities have been unchanged on average.
About 80 percent of the
expansion in gross positions in Treasury securities represented an equal
increase in long and short positions.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-47-
Table 19
Treasury and Agency Securities
Dealer Transactions, Gross and Net Positions
(Daily averages, billions of dollars)
~Yp~ ~f security
10/6/78 - 10/5/79
10/6/79 - 9/3/80
$11.83
17.61
3.06
17.01
22.73
4.19
0.68
0.75
0.17
0.18
2.44
13.88
1.18
3.30
9.51
1.07
0.18
0.35
0.08
0.11
Treasuries
Transac;t,ons
Gross positions
Net position
Turnover: Transactions .
gross P'Osition
Exposure:
Net posit1on t
gross position
Agencies
Transactions
Gross positions
Net position
TL.rnover: Transactions .
gross position
Exposure:
Source:
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Net position.gross position
Federal Reserve Bank of New York.
-48-
More volatile rates might be expected to lead to wider bid-ask
spreads.
It has been argued, however,~that in periods of-uncertainty market
participants gravitate to the more actively traded issues, making the deep
1
markets deeper and the thin markets thinner.
Thus,the spread in thin
markets would be more likely to widen, while spreads in actively traded
markets might be less likely to widen if the redistribution of traders from
thin markets to active markets offset the-overall tendency toward.wider
spreads.
Figures 3 and 4 indicate some widening of bid-ask spreads on
the three-month and 52-week Treasury bills in October 1979.
I
The Treasury bill
spreads are the Friday quotes of MerrillLynch published we~kly in the
Money Manager.
These data are list prices and are used as a proxy for
transaction prices,which are not available.
Many transactions take place
at other than the list price, notably for regular customers and on large
transactions.
In the 1973-75 period, the bid-ask spread on the three-month
Treasury bill was 10 basis points in all but six observations.
The spread
generally stayed at that level until the week of June 24, 1977,when it
fell to 4 basis points.
In the fourth quarter of 1977 the spread varied
between 4 and 20 basis points.
remained at 4 basis points.
From January to October 1978 it generally
After some variation at the end of 1978,
the spread returned to 4 basis points and generally remained there until
the tntroduction of the new procedure.
Then in the week immediately
1. See K. Garbade and W. Silber, "Price Despersion in the Government
Securities Market," Journal of Political Economy, Vol. 84 (August 1976),
pp. 721-41.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
I
0'I
I
~
,-1
,-1
....
,:q
l>-.
1-1
::I
en
co
1-1
(I)
E--1
.c: -
....i::
+J en
i:: +J
0
P-1
;:.::
I
C"")
C"")
en
0
1-1
(I)
1-1
Ill
....
::I
00
4-1
co
0 ....
fa«
-0 ,:q
ti) (I)
1-1
i:l.
LI)
......
.
.-f
en
N
......
C
• cu
::,
'J
0
+J
M
......
r-4
en
.
......
'J
>,
en
,::,
~
en
1
-0
•r-1
,:q
0
C"l
an
,
C"")
0
N
an
0
N
I,
~
'0
•= ~.
an
,-i
,-i
an
0
. . . . . . . .
...::t'
'
7/4/75
6/6/75
1/3/75
12/6/74
-11/1/74
.-f
.
1,0
N
s..
cu
.0
Cl.
...,m
•1-'
I
~
10/3/80
9/5/80
4/4/80
3/7/80
J/1/80
1/4/80
• 12/7/79
• 11/2/79
9/7/79
10/5/79
...,
en
8/3/79
QJ
- _ ,_ 6/7/74
co
......
1,0
en
0
-: _.. 5/3/74
7/6/79
3/1/74
'""4/5/74
~
s..
a,
0
.0
u
+J
5/4/79
4/6/79
10/6/78
12/1/78
11/3/78
3/2/79
- 10/5/73
- 9/7/73
'- 8/3/73
7/6/73
,-i
- 12/7/73
0
6/1/79
.
-
.-f
-7/5/74
: 1...8/2/74
-9/6/74
10/4/74
en
• 8/1/80
0
5/2/75
0
. 7/3/80
0
4/4/75
11"1
,-i
• 6/5/80
0
N
3/7/75
N
5/2/80
0
M
2/7 /75
0
co
M
. an. . an. . . . an.
2/2/79
-1/5/79
0
~
-11/2/73
- 1/4/74
: : - 2/1/74
0
,,,
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
I
I
0
I/'\
-.:I"
Q)
~
;::I
co
~
""
.-I
.-I
r:Q
""
~
>,
;:I
{I)
......
I/'\
0\
,-f
.
ca
Q)
~
......
N
E-1
- ;:I
Q)
,1-1
i::
.=
:s
,-f
>,
.
......
.
,-f
C'I
......
C")
.w
0
{I)
.c:0
=
,1-1
{I)
ca
""
{I)
P-4
=f "8
!'.:I
~
0
4-1
ca
"Ce
Q)
~
i:i.
Cl)
{I)
~
'1
"C
r:Q
""
0
•
C")
I/'\
C")
N
I/'\
0
N
--
~~
I/'\
0
r-4
-
~
~
r-4
0
- -- -
--
r. ~-
I.fl
0
. . . . . . . .
-.;t
-- -
J
0
7/4/75
6/6/75
5/2/75
4/4/75
3/7/75
2/7/75
1/3/75
12/6/74
11/1/74
10/4/74
9/6/74
• 8/2/74
7/5/74
6/7/74
5/3/74
• 4/5/74
.
3/1/74
2/1/74
1/4/74
12/7/73
11/2/73
10/5/73
9/7 /73
8/3/73
. 7/6/73
0
en
co
..-4
.
U)
N
s..
a,
ia,
a,
a.
+>
V)
0
+>
00
"'
en
.
..-4
'°s..
C1J
...,
.0
0
u
0
0
.
~
- - -- -
M
0
11"1
N
0
N
. . . .
C")
I.fl
:
-
.
r-4
I.fl
-~--~--
-
~~
- - --
r-4
0
-
0
I/'\
~
-
-
-
-
-
-
-
0
.
10/3/80
9/5/80
8/1/80
7/3/80
6/5/80
5/2/80
4/4/80
3/7/80
2/1/80
1/4/80
12/7/79
11/2/79
10/5/79
9/7/79
8/3/79
7/6/79
6/1/79
S/4/79
4/6/79
3/2/79
2/2/79
l/S/19
12/1/78
11/8/78
10/6/78
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-51-
following the introduction of the reserves~targeti~g procedures, the
bid-ask spread rose to 10 basis points,around which it has continued to
vary ever since.
The bid-ask spread on the 52-week bill, and presumably
on other bills as well, followed essentially the same pattern, as that for
the three-month bill, widening in the early part of October 1979.
A similar response to the introduction of the new procedure was
not apparent on Treasury coupon issues.
For example, Figure 5 shows for a
five-year note the Wednesday quotations of Discount Corporation, which are
published in the Money Manager.
at 4/32nds.
In the earlier period the spread stood
By 1978 it had risen to 8/32nds.
Two months before the
introduction of the new procedure it rose to 32/32nds and generally has
remained there over the past year~
at face value.
Perhaps this result should not be taken
As was mentioned above, it is well known that list-price
quotations do not always accurately reflect the actual prices at which transactions occur.
Also, ,there were this past year a fair number-of qualitative
reports that bid-ask spreads did widen in the coupon market after the intro~uction,of the new OP.erat1ng
procedures.
il ~
,..tt~~{:!-1 ➔-I,,-..,-.
~
_.,.
1
... _ ~
C:.
li,-',.1
l
<--l ..... ... -
>-~-r'T'
,<,1.t,.,~J
.tl>-'11't
-,,
It
may be that the
widening
of bidr~
~~
~:!" ..., "
~-- :,
.J
""
....
,,..
.,.
..
1:1
....
~l-"'
...
ask spreads just before October 1979 would have been a temporary phenomenon,
but that the heightened variability of rates that has emerged sustained the
increase in spreads.
More volatile rates also have been associated with an apparent
increase in what may be called the "implicit" underwriting premium that the
market extracts from the Treasury for the distribution of a new issue.
Such
a concession may be said to exist if the auction price of a Treasury security
is lower than the prevailing market price of comparable securities.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
This
I
N
I
L/"'j
U"'I
Q)
~
::s
Q)
.µ
~
--,
'~
c'
.-,,
-
~
~
I>,
::,
- U')
-,Ol
"
"N
-~
.
11)
'Q)°
~,...
E-4 .µ
s::
tll 0
~-,-4
r:,
C:
QJ
tll
'-
~
('f')
~
"Ol
.
:::,
- r:,
-:
-
-
-'°>, '
-
: 0
-
:::,
11-1
~~
I
L/"'j
0,
~.o
'l-l'i:o
OD
fz-4
-,-4
cw,
~....,
"0
"0 i::
tUN
Q)
~
ti)
~
l:IJ
1
-,-4
't,
i::1:1
r
-
i
0
..;t
i
I
ID
C""'I
N
I"")
N
CX)
N
..;t
,-j
'°
,-j
N
- - - -
- - - -- - --
0
N
....
~
i,,i,.1,,
~
..::I"
.,1,:1.-
CX)
-1->
'
7/4/75
6/6/75
-~ '...,__
.
N
00
..;t
0
10/3/80
9/5/80
6l?/80
-- 5/2/80
u
,-j
-
·'
12/1/78
'
1{5/79
3/2/79
2/2/79'
4/6/79
5/!+/79
6/1/79
7/6/79
- 8/3/79
-
9/_7 /79
10/5/79
-
11/2/79
12/7 /79
1/4/80
2/1/80
3/7/80
2/7 /75
Cl'
.
r-4
N
'°
QJ
S-
E
QJ
.0
C.
~
QJ
Vl
0
~
00
~
"Ol
.
'°
QJ
S-
.0
0
~
0
N
11/3/78
-::
4/4/80
3/7/15
-
8/1/80
..-4
ID
5/2/75
,,
N
0
1/3/80-
..;t
4/4/75
,;
0
OJ,
N
00
-
1/3/75
'
12/6/74
'
11/1/74
10/4/74
,,
9/6/74
8/2/74
""
7/5/74
6/7/74
5/3/74
- 4/5/74
,.. 2/1/74
: ,- • ~/1/74
-
-
N
C"l
. 8/3/}3
C"l
10/6/78
-
f 1/2}73
~
- U4/74
• 12/7 /72
~--
O"' ID
'!
:.
..;t '
7/6/73
- .10/5{73
.- 9/7 /73
0
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-53spread would be expected to increase when interest rates become more volatile,
as dealers seek compensation for the greater risk they assume in distributing
Treasury issues.
.
The average implicit underwriting premium for the three-month Treasury
bill before and after October 6, 1979, is shown in Figure 6 for both the bid
and ask side of the market.
Measurement of an implicit premium hinges on the
choice of a comparable security whose price can be compared to the Monday
auction price of the three-month Treasury bill.
Three comparisons were made.
One
calculation used the just-issued three-month bill compared with its price observed
at the market close on the Thursday prior to a Monday auction.
The presumption
is that this Thursday price is determined before mark~t participants have begun
to be affected by the impending auction but close enough to that auction to
assure comparability of other determinants of price.
A second comparison used
the clo\ing Monday price of the outstanding bill that has a maturity identical
to that of the three-month bill being auctioned.
The presumption is that.these
are identical aecurities except that one has yet to be distributed and the
other is already lodged in investors' portfolios.
A third comparison was
based on the Tuesday market price of the "when issued" three-month bill.
The pre-
sumption is that, with distribution under way, some of the tmderwriting premium
will have been removed from price.
In each of the three cases, the average
implicit underwriting premium was higher after October 6, 1979, than before,
measured on e-li.ther the bid or the ask side of the market.
C.
Underwriting Spreads on Corporate Bonds*
A considerable body of empirical evidence has shown that underwriting
spreads on both corporate and municipal securities are positively related to
*
The primary author of this section was Norman Mains of the Board staff.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-54Figure 6
'
IMPLICIT UNDERWRITING PREMIUMS IN 3-MONTH Bill AUCTIONS
Mean Basis Point Spread from Auction Average Price 60 Weeks Before and After October 6, 1979 1
•
Basis Points
30
25
After
20
+
15
10
+
After
5
Before
Before
0
•
•
-5
-10
-15
-20
-25
Thursday
Prior to Auct1ona
Mondayb
Auction
Tuesday
Following Auct1onc
(+) Mean spread ask price minus auction average price Each mean spread "after" Is s1gnif1cantly d 1fferent
from "before" (95%)
(•) Mean spread bid price minus auction average price Each mean spread "after" Is s1gmflcantly different
from "before" (95%) except for Monday auction
1
a
b
c
Introduction of FOMC New Operating Procedure
Based on closing quotations for 3-month bill auctioned 3 days earlier
Based on closing quotations for 6-month bill with same maturity as 3-month bill auctioned on that day
Based on closing quotations for 3-month bill auctioned the previous day
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-55the "riskiness" of the securities.
For example, studies have demonstrated
that the underwriting spread for a longer-term corporate or municipal security
offering is positively correlated with the rating awarded the issue by the
recognized rating agencies such as Moody's or Standard & Poor's.
Other studies
have attempted to explain differences in underwriting spreads by postulating
additional variables associated with the underwriting community's perception
of risk at the time of the offering.
1
In this tradition, this section examines
whether the increased volatility of interest ratesthis past year has been
associated with a ~idening of underwriting spreads on corporate bond issues.
Data on underwriting spreads for all publicly offered, nonconvertible
corporate rtotes and bonds were assembled for the 18-month period beginning
Aprill, 1979, and ending September 30, 1980,
The bond ratings and initial
maturities also were tabulated. 2 Over the entire 18-month period, data on
underwriting spread, rating, and original maturity were available for 411 issues.
On average, these public offerings of notes and bonds carried an underwriting
sp~~ad of $10.79 per $~~~00.~o~d_(tabl~ 20).
As expected, the underwriting
spread tended to be larger on lower-rated issues.
The average underwriting spread for the 117 issues publicly offered
between April 1, 1979, and September 30, 1979, was $11.47 per issue.
The
average spread for the 294 issues publicly offered between October 1, 1979, and
September 30, 1980,was $10.53 per issue, or $0.94
~
than the average under-
writing spread in the preceding six-months, A disaggregation of the issues by
1. For example, L.H. Ederington, '*Uncertainty, Competition, and Costs in
Corporate Bond Underwriting," Journal of Financial Economics, Vol. 2 (March
1975), pp. 71-94, presented evidence showing an inverse relationship between
the number of bids and the underwriting spreads on competitively bid issues.
Ederington concluded that the ntnnber of bids was a measure of corporate bond
underwriters' undertainty about the near-term outlook for interest rates.
2. Issues were classified according to the ratings assigned by Moody's
Investor Services. For those issues not rated by Moody's, the bond rating
of Standard & Poor's was used.
-56-
Table 20
Underwriting Spreads,
by Bond Rating
Bond rating£_7
Aaa & Aa
No.
Av. spread
Baa & helow
Av. spread
Period
All issues
No.
Av. spread
4/1/79 to 9/30/79
117
$11.47
47
$7.11
38
$8.61
32
$21.26
10/1/7g to 9/30/80
294
$10.53
109
$7.38
126
$8.61
59
$20.44
Hl.J79 ,to 9/3q/8o
411
$10. 79 _
.. 156
$7.30
164
$8.61
91
$20.72
'J:./
Issues classified by Moody's bond ratings.
issues not rated by Moody's.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
No.
A
Av. spread
No.
Standard & Poor's ratings used for those
-57ratings shows, however, that those rated Aaa and Aa had an average underwriting
spread of $7.38 per issue over the more recent 12-month period versus $7.11
per issue for the earlier period, issues rated A averaged $8.61 per issue for
both periods, while the issues rated Baa and below averaged $20.44 per issue
in the more recent 12-month period versus $21.26 for the issues offered in the
earlier 6-month period.
These data show that, in the aggregate, underwriting
spreads have not widened since the Desk implemented the new procedures.
Instead,
the data indicate that underwriting spreads narrowed somewhat, with this result
due entirely to a narrowing of spread for issues rated Baa and below.
State public utility commissions typically require regulated utilities
to sell their debt obligations via competitive biddings, while most other businesses, such as industrial and financial concerns, usually sell their debt
securities through investment banking organizations.
It is possible that the
underwriting process may have had an influence on the recent movements in
underwriting spreads.
Disaggregating the data into four maJor industry cate-
gories shows that underwriting spreads have widened in the more recent period
for corporations that tend to conduct their longer-term debt offerings via
competitive bidding (Table 21).
The average underwriting spreads both for
electric and gas utilities and tor telephone companies have widened in recent
months.
Moreover, this result remains for each subcategory when the issues
are grouped by their respective ratings.
In the case of negotiated offerings, however, the average underwriting spreads for both industrial and financial corporations declined in
the more recent 12-month period.
One possible explanation for this narrowing
of underwriting spreads may be related to the unprecedented decline in longer-
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Table 21
Underwriting Spreads, by
Bond Rating and Industry
Industry group
& ~riod
E&G
Utilities
Telephone cos.
Industrials
All issues
No.
Av. seread
Aaa & Aa
No.
Av. SEread
Bond rating=.
A
~o.
Av. SEread
No.
Baa & below
Av. s:eread
4/1/79 to 9/30/79
10/1/79 to 9/30/80
35
93
11
28
$6.51
$7.60
14
Li.3
$ 7.76
$ 8.84
$ 8.13
1()
22
$ 9.88
$11.81
4/1/79 to 9/30/79
10/1/79 to 9/30/80
7
33
$ 7.08
$10.17
4
16
$6.24
$7.88
3
16
S 8.20
$12.55
1
$ 8.75
$ 7.97
4/1/79 to 9/30/79
38
10/1/79 to 9/30/80
$16.58
11
107
$13. 20
27
$7.87
$7.81
11
51
$ 9.32
$ 8.09
16
29
$27.22
4/1/79 to 9/30/79
10/1/79 to 9/30/80
37
El
$10.35
$ 8.58
21
38
$7.20
$6.70
10
16
$ Q.15
$ 7.5R
6
7
$23.38
$21.10
4/1/79 to 9/30/79
10/1/79 to 9/30/80
117
294
$11.47
$10.53
47
109
$7.11
$7.38
38
126
$ 8.61
$ 8.61
32
59
$21.26
$20.44
$27.57·
I
lJl
00
I
Financial tos.
All issues
1/
Issues classified by Moody's bond ratings.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Standard & Poor's ratings used for those issues not rated hy Moody's.
-59term yields that occurred in the second quarter of 1980.
This sharp decline
elicited a record amount of note and bond financing by industrial corporations
and a sharp increase in financings by financial'concerns.
Corporate bond
underwriters may have been willing to narrow their spreads if they·perceived
that their risks were reduced in this market environment.
the underwriting spread data supports this hypothesis.
A disaggregation of
For all issues, the
average underwriting spread was $9.15 per issue in the second quarter of 1980
versus $11.55 per issue in the other three quarters subsequent to October 6,
1979 (Table 22).
However, the average underwriting spread for both industrial
and financial issues for the 12-month period excluding 1980Q2 is still less
than the average underwriting spreads in the 6-month period prior to the adoption of the nonborrowed-reserves strategy.
One other possible explanation for the recent narrowing of underwriting spreads for negotiated issues is a shift in the original maturities
of the financings.
Corporate bond underwriters traditionally charge their
clients a smaller underwriting fee for short-term financings.
Thus, the
narrowing in average underwriting spreads for industrial and financial issues
since October 1979 may reflect a reduction in the average maturity of offerings
during the period.
This hypothesis was tested by a further disaggregation of
the underwriting spreads by initial maturity for both the industrial and
financial issues (Tables 23 and 24).
It was generally found that the disaggre-
gation on the basis of maturity still showed a narrowing of spreads this past
year, particularly on note offerings.
It is interesting to note that for
'
issues with original maturities of 20 years or more the underwriting spreads
'
were about the same before and after October 6, 1979.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
In th~ case of note
Table 22
Underwriting Spreads, by
Bond Rating and Industry
Industry group
All issues
Av. s:eread
No.
& J2!riod
E&G utilities
Telephone cos.
Industrials
Financial cos.
All issues
!!
4/1/79 to 9/30/79
10/1/79 to 9/30/80:
Excluding 1980 Q2
19801 Q2
4/1/79 to 9/30/79
10/1/79 to 9/30/80:
Excluding 1980 Q2
1980 Q.2
Bond ratinP.I7
A
No.
Av. spread
No.
Baa & below
Av. soread
35
$ 7.97
11
$6.51
14
$ 7.76
10
$ 9.88
71
22
$ 8.94
$ 8.51
22
6
$7.56
$7.76
31
12
$ 7.91
$ 8.70
18
4
$12.42
$ 9.07
7
$ 7.08
4
$6.24
3
$ 8.20
21
12
$11.07
$ 8.60
10
$7.58
$8.38
11
$14.24
$ 8.84
1
$ 8.75
6
5
4/1/79 to 9/30/79
10/1/79 to 9/30/80:
Excluding 1980 Q2
1980 Q,2
38
$16.58
11
$7.87
11
$ 9.32
16
$27.57
51
56
$16.58
$10.12 ,
13
14
$7.51
$8.09
21
30
$ 7.98
$ R.17
17
12
$31.54
$17.38
4/1/79 to 9/30/79
10/1/79 to 9/30/80:
Excluding 1980 Q2
1980 Q2
37
$10.35
21
$7.20
10
$ 9.15
6
$23.38
26
35
$ 9.09
$ 8.20
14
24
$6.55
$6.79
9
7
$ 7.87
$ 7.21
3
4
$24.7'3
$18.38
117
$11.47
47
$7.11
38
$ 8.H
32
$21.26
169
125
$11.55
$ 9.15
59
50
$7.31
$7.46
72
54
$ 8.90
$ 8. 2'3
38
21
$22.92
$15.95
4/1/79 to 9/30/79
10/1/79 to 9/30/80:
Excluding 1980' Q2
1980' Q2
Issues classified by Moody's bond ratin~s.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Aaa & Aa
No.
Av. s:eread
Standard & Poor's ratings used for those issues not rated by Moody's.
I
0
°'
I
Table 23
Underwriting Spreads for Industrial Bonds,
by Maturity and Rating
Maturity
period
&
20 years or more
4/1/79 to 9130/JCJ
10/1/79 to 9/30/80
Memo
10/1/79 to 9/30/80
Excluding 1980 Q2
1980 Q2
Less than 20 years
4/1/79 to 9/30/79
10/1/79 to 9/30/80
Memo
10/1/79 to 9/30/80
Excluding 1980 Q2
1980 Q2
All issues
No.
Av. spread
Aaa & Aa
No.
Av. spread
No.
Baa & below
Av. spread
$9.97
$9.28
12
25
$26.52
$26.46
16
$0,03
$9.45
17
8
$2Q,34
$20.35
$8.57
$7.16
3
24
$7.60
$6.38
4
7
$30.74
$18.29
$7.60
$6.82
10
$5.93
$6.70
3
14
4
$22.84
$14.88
26
63
$17.32
$16.00
11
$8.71
$8.75
34
29
$19.lli
$12.34
6
5
$8.75
$8.75
11
12
47
$15.72
$ 8.42
5
16
20
27
$ 9.05
$ 7.95
7
9
8
27
I
1/ Issues classified by Moody's bond ratings.
by Moody's.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
6
Bond rating::,:
A
Av. Spread
No.
Standard
&
Poor's ratings used for those issues not rated
°'I-'I
Table 24
Underwriting Spreads fo~ Financial Bonds,
by Maturity and Rating
Bond Rating=:c
Maturity &
Period
20 years or more:
4/1/79 to 9/30/79
10/1/79 to 9/30/80
Memo:
10/1/79 to 9/30/80
Excluding 1980 Q2
1980 Q2
Less than 20 years:
4/1/79 to 9/30/79
10/1/79 to 9/30/80
Memo·
10/1/79 to 9/30/80
Excluding 1980 ~2
1980 Q2
All issues
No.
Av. Spread
15
23
Aaa & Aa
No.
Av. Spreacl
A
Av. Spread
No.
Baa & below
Av. Spread
$10.80
6
11
$8.75
$8.59
5
6
$8.50
$8.75
4
$12.40
6
$16.75
$23.04
4
7
$8.12
$8.86
4
$8.75
$8.75
3
$34.91
12
$15.92
$ 9 .42
3
$11.17
22
38
$ a.n
$ 6.27
15
27
$6.58
$5.92
10
$9.130
$6.88
2
1
$24.13
$ 9.50
16
22
$ 6.51
$ 6.09
10
17
$5.89
$5.94
5
5
$7.16
$6.60
1
$ 9.50
11
1/ Issues classified by Moody's bond ratine;s.
by Moody's.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
No.
2
5
Standard & Poor's ratings used for those issues not rated
I
°'
N
I
-63-
,_
offerings--issues with original maturity of less than 20 years--the data indicate a narrowing of spreads even when the issues are cross-classified by rating.
D.
Reactions in Mortgage Markets to Increased Rate Volatility*
Announcement effects.
Immediately following the Federal Reserve
policy announcements in October 1979, a number of mortgage lenders shut down
or severely reduced their fonzard commitment activity because of heightened
uncertainty about the course of market interest rates and flows of loanable
funds.
A special survey of large savings and loan associations conducted by
the Federal Home Loan Bank Board, for example, indicated that a fourth of the
institutions ceased making any new commitments between October 6 and month-end,
and that an additional 50 percent ceased making mortgage connnitments for certain
classes of loans or borrowers.
During the next few months, these associations
gradually reopened their commitment windows or broadened the types of commitments being made, despite highly variable interest rates and a marked slowing
of peposit flows; by mid-January 1980, only 7 percent of surveyed associations
indicated that they were making no new connnitments at all.
Although market
conditions have settled down considerably, various participants have made
adjustments to their commitment and investment policies since late last year,
particularly in the case of long-term or "permanent" mortgage financing.
Mortgage originators/bankers.
Among mortgage market participants,
mortgage bankers are susceptible to the greatest damage, relative to capital
positions, from increased short-term variability in market interest rates.
Indeed, the increased rate variability during the past year apparently has
prompted some mergers of mortgage companies and acquisitions of smaller ones
*
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The primary author of this section was David Seiders of the Board staff.
-64-
by larger ones.
It might also be expected that the remaining institutions
would attempt co shift more of their interest rate risks--associated with
the issuance of forward commitments to borrowers and the carrying of mortgage
inventories--forward to the borrowers, down the line to investors or dealers
in passthrough securities, or to speculators in futures markets who seek to
profit by bearing risk but do not intend to acquire the mortgage instruments.
Mortgage bankers could minimize the size of their mortgage inventories or secure investor purchase commitments to cover larger portions
of inventories held and outstanding connnitments made to borrowers.
However,
neither type of adJustment has been striking for the industry as a whole.
The size of mortgage banker inventories has fallen substantially since September 1979, but the ratio of inventory to current origination activity has
declined only moderately (Table 25).
Moreover, the coverage ratio for mortgage
bankers--outstanding long-term commitments received from investors (including
FNMA and the GNMA securities dealers) divided by the sum of mortgage inventory
and outstanding long-term commitments issued to borrowers--has changed little,
on balance, since the fall of 1979.
Nor does it appear that mortgage bankers have substantially increased
their use of the GNMA futures market in order to hedge interest rate risks
incurred during the mortgage origination/marketing process.
Mortgage banker
activity in the GNMA futures market has been quite limited since the creation
of the market in 1975, for a number of reasons.
For one thing, the delivery
provisions of the maJor GNMA futures contract are rather cumbersome, calling
for delivery of a Collateralized Depository Receipt or
11
due bill 11 that provides
a claim on a pool of GNMA securities rather than the securities themselves.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-65-
TABLE 25
Mortgage Company Inventory and Commitment Positions
Long-term mortgage investor
Ratio to current
originations
Commitment coverage ratio 1
{percent)
Home
Total
lllortgages
mortgages
Period
Amount
($millions)
1979-Jan.
Feb.
Mar.
8,750
9,500
8,640
1.99
71
2.36
2.57
66
67
85
79
79
Apr.
May
June
7,880
8,820
2.29
2.71
2.92
67
64
67
80
79
77
July
Aug.
Sept.
9,620
10,970
11,380
2.81
2.95
2.99
56
53
,57
85
65
68
Oct.
Nov.
Dec.
10,620
2.84
9,920
8,960
2.88
2.95
60
61
63
73
79
80
1980-Jan.
Feb.
Mar.
8,500
8,080
6,310
3.00
2.79
2.06
59
64
75
74
80
89
Apr.
nay
June
5,800
5,760
5,680
2.49
2.35
2.11
74
5g
61
92
80
R3
July
Aug.
Sept. p
9,260
7,560
8,440
3.14
2.so
2. 78
63
58
73
RO
84
8,820
68
1. Outstanding long-term commitments received from investors divided
by the sum of mortgage inventory and long-term commitments issued to
borrowers.
Source: Mortgage Bankers Association of America.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-66Moreover, the futures market requires posting of an initial margin and then
maintenance of a specified margin.
It appears that mortgage bankers, and some other originators, have
attempted to shift some of the increased risk incurred during the mortgage
commitment period primarily to mortgage borrowers, rather than to mortgage
investors or to speculators.
While comprehensive data are not available,
field reports conducted regularly by FNMA and HUD have noted increased reluctance by mortgage originators to issue the standard long-term fixed-rate
optional-takedown commitment to prospective borrowers.
Adjustments generally
have involved imposition of larger nonrefundable commitment fees to discourage
cancellations of commitments by borrowers when market rates fall, shortening
of periods over which fixed interest rates on conventional loans or fixed
nlllllbers of points on FHA/VA loans will be guaranteed, or use of floating
rates and discount points tied to some visible market indicators--such as the
average yield determined in FHLMC's auctions of immediate-delivery purchase
commitments.
Of course, the extent to which mortgage originators can shift
commitment-period interest rate risk to borrowers, or the cost of such risk
shifting, will depend upon the state of competition in local mortgage markets.
HUD surveys have noted significant declines in mortgage volume at some mortgage bankers>who have attempted to discontinue issuance of fixed-rate commitments.
It does not appear that mortgage originators have sought to shorten
the total length of the commitment period (in contrast to the fixed-rate
portion), probably because this period is determined largely by the tl.Ille lags
inherent in the real estate construction and sales process.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Available data
-67for thrift institutions and life insurance companies (Table 26) indicate
that the maturity distributions of mortgage commitments outstanding at these
institutions have not changed appreciably since the fall of 1979 despite
substantial declines in new commitment activity.
GNMA securities dealers. FNMA. and FHLMC.
Dealers who have made
forward markets in GNMAs have found it difficult to contend with the increase
in rate volatility.
A number of dealers have left this market largely
because of the increased capital risks, and the roster of prominent market
makers has dwindled to a handful.
Margin requirements have become more common
at the large dealers that now dominate the market, as these dealers have sought
to minimize their credit risks in the wake of some reneges on the standard
"firm" or "mandatory" delivery forward contracts.
Moreover, the dealers
apparently have been hedging more of their interest rate risks, associated with
inventory positions and uncovered commitments, in the GNMA futures markets.
Bid-asked price spreads also have widened somewhat at the major dealers,
apparently because of the increase in risk and possibly some erosion of competition in the dealer market for GNMAs.
The bid-asked spread on new and
recent issues of GNMAs generally has widened from 1/8 to 1/4 percentage point
since the fall of 1979.
Heightened interest rate uncertainty has made both the GNMA dealers
and FNMA reluctant to issue long-term fixed-rate standby commitments, compounding the problem faced by mortgage originators/sellers.
Such commitments have
become unavailable or quite expensive at the GNMA dealers during the past year,
reflecting the reluctance of private investors to enter into standby arrangements and the unwillingness of dealers to absorb credit risks on these contracts.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Around the end of 1980, moreover, FNMA made substantial changes to
TABLE 26
Maturity Distribution of Outstanding Mortgage Connni tmen ts
at Thrift Institutions and Life Insurance Companies
(Percent scheduled to be taken down within the number of months indicated)
Period
1
Insured savings and loans
3 months
6 months
12 months
2
NY State mutual savings banks
3 months
6 months 9 months
Life insurance
companies 3
6 months
12 months
1979
Jan.
Feb.
Mar.
61.0
62.1
64.0
79.5
79.7
80.6
93.8
93.8
94 .2
38.5
40.0
42.4
61.2
60.8
63.8
74.0
73.0
76.3
33.1
34.0
33.7
59.3
59.4
58.4
Apr.
May
June
63.9
64.2
62.9
79.9
80.8
80.1
93.8,
94.3
93.8
41.1
40.2
41.4
61.5
60.0
65.2
74.7
76.2
78.4
34.9
37 .4
37.1
61.3
61.2
63.3
July
Aug.
Sept.
62.2
62.1
62.2
79.8
78.8
78.4
93.6
93.1
92 .9
41.4
,40.6
46.6
65.2
63.9
68.2
78.4
79.0
80.2
37 .9
35.7
37.9
64.0
63.0
62.1
Oct.
Nov.
Dec.
59.6
54.8
52.2
76.2
73.5
72.1
92.3
91.3
89.6
44.2
43.5
44.7
67.0
66.3
63.S
78.0
78.7
76.1
39.5
38.0
34.0
61.7
61.6
57.9
1980
Jan.
Feb.
Mar.
52.4
54.1
52.4
71.7
72.3
70.6
89.9
90.2
89.2
44.0
44.6
46.5
64.3
68.6
67.8
75.4
79.6
80.4
37.4
36.6
37.2
62.4
62.5
64.2
Apr.
May
June
48.7
48.3
53.6
68.2
68.4
72.0
89.9
90.4
90.9
45.0
44.1
41.5
64.9
62.6
64.2
77.7
76.8
73.4
35.4
36.7
35.8
60.4
60.6
60.1
July
Aug.
Sept.
59.1
60.7
61.2
76.2
77.2
78.0
92 .1
91.8
92 .2
43.9
47.9
53.8
65.5
66.7
67.4
76.9
77 .6
79.4
36.2
37.4
60.4
60.6
Oct.
58.6
76.6
92.0
f
I
1. Insured S&Ls account for about 98 percent of the total assets of all operating S&Ls.
2. NY State mutuals account
. for about 55 percent of the total assets of all operating mutual
savings banks.
3. Reporting life companies account for about three-fourths of the total assets of the life
insurance
ipdustryr
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
a-
00
I
-69its connnitment programs in order to reduce its exposure to rate variations .
.
The four-month optional-delivery purchase connnitments issued by FNMA through
its biweekly Free Market System auctions now guarantee a mortgage rate for
two months rather than four.
In addition, the fixed-rate twelve-month standby
program has been terminated, and longer-term convertible standby connnitments
issued by FNMA are now subject to rate adjustment every two months.
FNMA also increased the fees charged successful bidders in the biweekly auctions of optional-delivery purchase commitments, from 5/8 to 3/4 of
1 percent of the connnitment amount (effective January 19, 1981), and FNMA
apparently raised its yield requirements to some degree during the past year.
FNMA can exert a limited degree of control over the average yield on accepted
bids by varying the proportion that are accepted.
Th.is proportion was unusually
low during the last quarter of 1979 and the first half of 1980 but returned
1
to its historical average in the latter half of the year.
The lower proportions accepted in the three previous quarters reflected efforts by FNMA to
boost sagging net earnings and apparently had little to do with increased
short-run rate variability, per se.
Mortgage investors.
Traditional mortgage investors, including FNMA,
have shown increased interest in shorter-term mortgage contracts, or contracts
providing for equity participations, during the past year.
Thu~ increased
rate volatility has reinforced the desires of investors to move away from
assumable fixed-rate, long-term mortgage assets--desires that had developed as
1. At FHLMC (which quickly resells most loans that it purchases), the
proportion of bids accepted in its commitment auctions has not changed markedly
during the past year, and commitment fees have not been raised.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-70-
it became evident that expectations of short-term rate movements embodied in
long-term mortgage rates agreed to in past years had been biased downward.
Rate volatility has stimulated regulatory actions to permit thrift
institutions and connnercial banks to offer variable-rate and renegotiablerate home loans as well as contracts that permit lenders to share in the price
appreciation of mortgaged homes. 1 Thrift institutions have been aggressively
marketing adjustable-rate mortgages in some areas of the country.
A special
survey conducted in September 1980 by the U.S. League of Savings Associations
indicated that nearly one-third of all S&Ls were offering renegotiable-r~te
loans.
Moreover, many major life insurance companies recently have been
requiring some form of equity participation when granting long-term incomeproperty mortgages.
This practice has reemerged despite a rather checkered
experience by life companies with equity participations during the late 1960s
and early 1970s.
FNMA has been considering programs for the purchase of adjustablerate mortgages, but has not as yet instituted such programs.
FNMA has taken
steps, however, to limit the assumability of the fixed-rate conventional loans
it purchases in order to reduce its term-structure risks.
For both FHA/VA and
conventional loans sold to FNMA under commitments issued after November 10,
1980, assumptions will be permitted only after the rates on the loans have
been adjusted to current market levels--except in 17 states where due-on-sale
1. In April 1980, federal S&Ls were empowered to acquire renegotiable-rate
mortgages, and in September the FHLBB issued proposed regulations concerning
graduated-payment adjustable-rate mortgages as well as shared-appreciation
mortgages. Also in September, the Comptroller of the Currency issued proposed
regulations that would authorize all national banks to offer adjustable-rate
home mortgages (state law had posed restrictions in a number of areas).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-71provisions are prohibited.
In those states where FNMA's new assumption policy
cannot be fully implemented, FNMA will reserve the option to call conventional
loans due and payable-in-full seven years from the date of origination.
The value of prepayment privileges provided to mortgage borrowers
naturally increases when mortgage rates become more volatile, since the likelihood of profitable refinancing is greater.
Even so, most lenders apparently
have not attempted to impose larger prepayment charges in order to discourage
borrowers from exercising this option.
Prepayment penalties may not be charged
on F'HA/VA loans, prepayment charges on conventional loans made by federal S&Ls
are fixed by regulation, 1 and several states have outlawed penalties or set
maximums that apply to all types of lenders.
The changing environment in mortgage markets may have had some
effects on the level of mortgage interest rates and on the relationship between
mortgage and other long-term yields.
Aggressive marketing of adjustable:rate
loans has placed some downward pressure on yields for this type of instrument
and upward pressure on yields for the traditional fixed-rate contract.
Mort-
gage borrowers apparently prefer long-term fixed-rate contracts when the
interest rate outlook is highly uncertain, even if the relationship between
the long-term mortgage rate and the initial rate on adjustable-rate contracts
is in line with the prevailing term structure.
Correspondingly, field reports
and trade sources indicate that S&Ls generally have been offering renegotiablerate loans at interest rates ordinarily 1/2 to 1 percentage point below going
rates on fixed-rate loans even though the yield curve has been flat or inverted
in the relevant range.
1. Up to 20 percent of the original loan balance may be prepaid without penalty
during any 12-month period, and the penalty for prepayments in excess of 20 percent cannot exceed 6 months' interest (at the original contract interest rate)
on the excess amount.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-72E.
Commercial Bank
Response to Rate Volatility*
There are· several courses commercial banks might have followed in
response to the October 1979 change in operating procedures, depending upon
both their assessment of its contribution to interest rate risk and their
risk preferences.
Dn the one hand, if banks believed that the change in
operating procedureswould increase very-short-run interest rate volatility
but would not materially affect cyclical or secular interest rate movements,
their response might be slight, since profits from rate movements on one day
would be cancelled by losses on another day.
On the other hand, if banks
believed that the new operating proceduresalso would amplify cyclical or
secular interest rate movements, they might alter their fundamental balance
sheet positions.
Banks seeking to avoid additional risk and believing their
long-run portfolio-funding strategy before October 1979 provided the optimum
amount of interest risk exposure would move to keep that exposure constant,
for example by closer matching of rate-sensitive assets and liabilities; more
frequent adjustments in their administered interest rates; and, to the extent
those actions did not bring them down to their desired interest risk exposure,
the use of futures contracts to hedge the remaining undesired risk.
Other
banks might try to profit from the larger expected cyclical and secular rate
movements by altering their portfolio-funding strategies--and perhaps increasing their trading activity--according to their interest rate forecasts.
*
The principal author of this section was Barbara Opper of the Board staff.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-73Conunercial bank activity was analyzed in an attempt to detect which
of these possible courses banks have taken.
Unambiguous conclusions did not
'
emerge, in part because of data limitations and in part because
of difficulties in interpreting the changes that were measurable.
Some large
banks, and virtually all small banks, were not completely in control of their
balance sheet positioning since October 1979 because of the nature of the
market conditions that have prevailed since that time.
For most of that
period, rate-sensitive claims were the only liabilities banks could market;
asset turnover, particularly at consumer-oriented banks, was inadequate to
fund acquisitions of enough rate-sensitive assets to balance the rapid shift
in their liabilities.
Rate-sensitive portfolio assets.
Large banks typically have allo-
cated a large share of their total assets to short-term assets and to longerterm instruments having floating rates.
Smaller banks in the past only had a
small portion of such rate-sensitive assets,,which was consistent with the
composition of their liabilities (Figure 7).
Since mid-1979, there appears to
have been little change in the allocation of large banks' assets invested in
rate-sensitive instrmnents.
When small banks' liabilities began to become
significantly more interest rate-sensitive after the mid-1978 introduction of
the six-month money market certificate, they began to increase their portfolio
allocations of interest-sensitive assets.
One method they used was to increase
the incidence of floating-rate provisions on longer-term business loan acquisitions (Table 27).
Another was to increase, in proportion to asset growth,
their sales of federal funds and their acquisitions of other short-term loans
and investments.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Figure 7
Percentage of Assets Funded by Rate-Sensitive Liabilities
and Invested in Rate-Sensitive Instruments/
Insured Commercial Banks, by Asset Sizel
I
,-
I
30
I
I
.,. -'
I
10
0
1976
1978
30
30
INVESTED
✓
I
----------
I
I
20
I
INVESTED
FUNDED
$100 MILLION-$! BILLION
$25-100 MILLION
UNDER $25 MILLION
1980
..
,, I
INVESTED
______
FUNDED
_._
✓
✓
FUNDED
"'
10
----------------1976
1978
20
20
19-80
10
0
-1976
- - - - - - -1978
- - - - - - -1980
--- 0
OTHER OVER $1 BILLION
MONEY CENTER
I
70
70
60
60
INVESTED
__________ .,,,.
FUNDED
50
50
INVESTED
40
✓
,I'
... ________ ., .,. ti'
FUNDED
30
1976
1.
,I'
30
1978
1980
1976
1980
1978
Rate sensitive assets are interest-bearing deposits, federal funds sold, reverse RPs, loans and government debt maturing in one year or less, and other loans with floating rates. Small banks do not report
the loan detail, so their holdings of loans to financial institutions, construction loans, and purpose
loans are included. Rate-sensitive liabilities are large time deposits and foreign office deposits due
in one
. year or less, federal funds, RPs, MMCs,and other short-term borrowings. Last plot: September 1980.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
~
.
-.J
~
I
-75-
Table 27
Percentage of Gross Business Loan Exte~sions Having
Floating Rate Provisions!/
.!lli.
- 1980
Q2
Q3
19-77
19.78
1979
Q4
Ql
48 large banks
73.0
64.4
79.2
86.5
80.3
85.0
80.2
86.8
Other
39.0
45.1
38.5
46.9
44.7
56.0
48.2
48.2
48 large banks
65.8
63.5
62.8
69.7
55.1
53.9
31.6
54.4
Other
39.2
48.4
37.4
35.6
44.7
35.8
41.7
43.9
Item
g4
Lon,g-term
Short-term
Short-term, excluding below-prime loans
48 large
65.2
68.7
68.1
78.4
73.7
68.0
67.8
61.6
Other
37.0
52.6
44.1
46.0
55.1
33.3
42.1
47.1
1.
Annual data are averages of survey weeks. The Survey of the Terms of Bank
Lending is taken as of the first full week of the middle month of each
quarter.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-76Asset pricing.
Large banks since mid-1979 have moved toward more
aggressive manipulation of their prime lending rate.
They have taken more
control over fluctuations in the prime rate, in effect achieving better
management of the interest sensitivity of their prime loan portfolios.
adjustment is depicted in Figure 8.
That
The upper panel of the chart shows the
interest rate spread to a large bank that issues a 30-day CD to acquire a
prime-rate loan.
The bottom panel of the chart depicts the prime-rate dif-
ferential over the effective interest rate on commercial paper.
The long
period of adherence to a relatively stable pricing formula is clearly indicated by the stability of these differentials from late 1975 through late
1979; the departure from that formula also is clear, beginning with the small
but abrupt increase in the differentials in late 1979 and extending to the
unprecedented peak reached in early 1980.
That change in pricing should not
be attributed entirely to attempts to manage loan portfolio interest risk
exposure, however.
The sharp early 1980 decline in domestic office business
loan growth at commercial banks--middle panel of the chart--was clearly consistent with the growth constraints prescribed by the Special Credit Restraint
Program and with the incentives for large business borrowers to use commercial
paper as a far cheaper alternative to counnercial bank loans priced at the
prime.
Rate-sensitive liabilities.
Large banks typically have relied
relatively heavily on rate-sensitive liabilities for their funding (dashed
lines in Figure 7).
further.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Since mid-1979 that reliance has increased somewhat
Thus, on balance, large banks shifted toward a position that
Figure 8
Business Loan Growth at Domestic Offices of Commercial Banks
and Prime Rate Spreads,
'
PERCENTAGE 'FOINTS
I
I.
PRJME MINUS RESERVE-ADJUSTED 30-DAY
.c:--:----::--..-.
m
~ 1\,
i'-v"'II ::,~~,~~:_--""'~~,:'v-j\/_....-,::1
.---~
=======1.=========j::1:_1_70" I
:1
l
~
RATES
•
~-.,._____
-
8
.d
0
4
PERCENT, SAAR
I
80
60
cx:MIBRCIAL AND INDUSTRIAL IDAN GROWIH
I
I
~/:~M~1·I
I
j'
. , _ _ _ . _ _ _ . . . _ " - . _ _ ~ . . . _ _ _ . . _ _ ~ ' - - ' - - ' ' - - - - ' - - - - ' ' - - - J _ _ _ j- - - - - - - - - - - - - -
--
20
20
PERCENTAGE 'FOINTS
'
I
I
PRIME MINUS CDM-1ERCIAL PAPER. RATES
Latest Plot:
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
8
0
1960
* Serves break
1964
1968
(January 1973).
November 1980, excent loan arowth (October).
1972
1976
1980
I
--.J
--.J
I
0
'
V
"'J
40
-78-
implied that they began to fund some longer-term or otherwise rate-insensitive
1
assets with rate-sensitive liabilities.
While large banks' interest risk
exposure appears to have increased as a result of the shift toward more ratesensitive liabilities, the change has been far more dramatic at smaller banks.
Although acquisitions of rate-sensitive assets of small banks have increased
substantially since mid-1979, their issuance of rate-sensitive liabilities was
far more pronounced.
Smaller banks now are financing a sharply increased frac-
tion of longer-term, rate-insensitive assets with liabilities carrying short
maturities and interest rates tied closely to the short-term markets.
This apparent shift toward increased interest risk exposure seems
to be largely a result of conditions that are not easily controlled by many of
these institutions, particularly the smaller banks.
Despite their sharp swings,
market yields have remained comparatively high since October 1979.
With the
opportunity costs to depositors of holding demand deposits or fixed-ceiling
savings and small time deposits consistently high, banks efperienced attrition
in those deposits.
These outflows were, offset primarily by issuing money
market liabilities and floating-ceiling, small time deposits.
Use of the futures market by banks.
tn principle, banks could hedge
an increase in interest risk exposure in the financial futures markets.
Although a few large banks use futures contracts in this way, most do not.
An
1. Large banks did not issue a stepped-up volume of floating-rate intermediateterm liabilities. In 1978, public issues amounting to $20 million or more
totalled $200 million; in 1979, they totalled $2 billion, all of which was
issued in the first half of the year; and in 1980 through Novembe~ they issued
$250 million.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-79important~impediment to widespread use of financial futures contracts to
hedge portfolio positions js the present accounting treatment required by
bank regulators.
Futures contracts must be carried at the lower of cost or
market value; any difference in the carrying value from one accounting period
to the next must flow through the income account.
Consequently, an asymmetry
exists because the portfolio assets and liabilities that the contract is
designed to hedge are not carried 'at market value; gains or losses stemming
from the portfolio-funding positions are not reported until they are realized.
This accounting treatment does not affect the final reported profit or loss
resulting when the portfolio position/hedging combination reaches its maturity.
It potentially does introduce, however, more volatility to interim reported
earnings--volatility that is directly a function of the market interest rate
movements that the futures contracts are intended to offset.
F.
Developments in financial Futures Markets*
The greater variability of interest rates might be expected to
attract greater trading interest in financial futures markets from both
speculators and hedgers.
From the point of view of speculators, an increase
in the volatilit~ of interest rates provides more opportunity for profit
(and loss) for t~ose willing to bet on the future course of interest rates.
At the same time,, for those holding positions in the cash market, more volatile
interest rates create greater risk and, therefore, more reason to hedge those
cash positions in the futures markets, 1 Financial futures markets have,
1. As a simple example of hedging, a government securities dealer with a
large inventory of Treasury bills could protect itself against a rise in
interest rates by taking a short position in the Treasury bill futures markets.
If interest rates did rise, the resulting profit on the dealer's short position in the futures market would at least partially offset the loss on the
dealer's long bill position in the cash market.
* The principal author of this section was Leigh Ribble of the Board staff.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-80indeed, continued to prosper and grow over the past year.
It is by no means
clear, however, to what extent this growth may be attributed to the greater
volatility of interest rates, especially given that these markets have
enjoyed strong growth each year since their inception.
As may be seen in Table 28, sales volume has increased over the
past year for each of the three major financial futures contracts now being
offered, with trading activity in one of them--the CBOT's Treasury bond futures
contract--more than tripling over this period.
Thes~ sales figures are con-
sistent with reports from market participants that financial futures markets
have been functioning quite well.
According to participants, it usually has
been possible to put large transactions through in the major financial futures
markets without greatly affecting price, even at times when it was not possible
to do so in the cash market for the corresponding instruments.
Open interest--the total number of contracts outstanding that have
not been offset by opposite futures transactions nor fulfilled by delivery-has grown at an extraordinary pace in the Treasury bond contract since October
1979,and has increased somewhat in the GNMA-CDR contract as well.
In the
Treasury bill contract, however, open interest declined over most of the
same period, Market professionals are uncertain about,_ th~, reasons for the
1
decline.
The decline began in mid-1979, before the recent large fluctuations
in interest rates, and continued until this past September, reaching a low of
about 22,000 contracts.
By early December, however, open interest had rebounded
to 45,000 contracts.
1. According to some, the contraction in open interest since mid-1979 reflected
the exit of speculators who entered the market aggressively earlier in 1979
trying to profit from what they expected to be a cyclical peak in interest rates.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-81-
Table 28
Sales Volume and Open Interest in Financial Futures
(Monthly averages in billions of dollars)
Period
3-month
Treasury
bill
(IMM)
1979-Ql
4.3
Q2
Sales Volume
GNMACDR
delivery
(CBOT)
Open Interest
Treasury
bond
(CBOT)
3-month
Treasury
bill
(IMM)
.4
5
54.4
6.3
4.3
7.3
.5
.7
59.6
6.7
5.2
Q3
8 6
.5
.9
46.1
7 .4
6.3
~4
10.6
.8
1. 2
39.2
7.8
7.3
1980-Ql
12 3
.8
1 6
29 7
6 2
6.9
Q2
14 6
1.0
2 5
27.0
5 8
7 7
Q3
10 6
7
2.5
22.6
6.7
11 6
Oct.
11. 9
9
3 0
23 5
7 0
13. 7
N'1V,
16.0
1.2
4.0
32.4
9.3
18 2
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
GNMA-
CDR
delivery
(CBOT)
Treasury
bond
(CBOT)
-82It is particularly difficult to gain a reliable impression of the
growth in hedging activity in financial futures markets over the past year.
According to both the available data and discussions with market participants,
hedging activity has continued to grow since October 1979, although perhaps no
more rapidly than speculative and arbitraging activity in these markets.
The
primary source of data in this area is the CFTC's Commitments• of Traders
report, which shows the open interest of large reporting traders in futures
markets on the last day_ of each month.
1
In this report, large traders classify
their positions as either speculative or hedging. 2
Table 29 shows the hedging
counnitments of large traders in both absolute terms and relative to total open
interest in the three major financial futures markets.
3
According to these data, hedging commitments of large traders in the
Treasury bond futures contracts have increased enormously in absolute terms
over the past year.
Total open interest in the bond contracts has grown at an
even faster pace over this period than large-trader hedging commitments in the
contract.
The absolute increase in hedging appears consistent with market
counnents that the Treasury bond contract increasingly is being used to hedge
1. Traders are required by the CFTC to report if they hold or control large
positions in any one futures contract. In the case of financial .futures, a
"large" position is now defined to be 25 contracts.
2. As an additional source of information, the CFTC made surveys of futures
market participants on November 30, 1974 and March 20, 1979. The first survey
classified traders as either speculators or hedgers. However, the second
survey did not attempt to make this classification, because of what was felt
to be the difficulty of doing so.
3. Total open interest is the total amount of either the short contracts or
the long contracts, while commitments of large traders are the sum of both the
short and long positions of those traders. As a result of this double,counting the proportions calculated in the last two columns of Table 2 could theoretically be as large as 200 percent.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Table 2 9
Hedging Corranitments of Large Traders in Selected Financial Futures Markets
(Averages of end-of-month figures)
l:ombine.d
Hedging Commitments
-Large Traders
Total Open Interest
-All Traders
(in
Quarter;
1978
$ billions)
1980
1979
.
(in
1978
Combined Hedging
Commitments as a
% of Total Open
Interest3_/
\in percent)
1980
1978
1979
--
$ billions)
1980
1979
3-month Treasury bill contract (IHM)
I
Ql
Q2
Q3
Q4
32.0
53.2
27.9
26 0
22.5
53.5
60.2
42.3
38 0
I
5.7
10 5
u.s
16 5
21. 7
16 7
15 3
11. 5
7 9
7 9
18
20
31
36
40
40
41
30
35
64
43
53
51
44
33
39
42
39
58
66
76
72
65
66
74
54
52
Treasury bond contract (CBOT)
I
Ql
Q2
Q3
Q4
1.3
3.5
4.4
5.4
6.5
7.7
6.5
8 9
11. 7
.8
1.5
2.4
2.8
2.8
2.5
2 5
3.8
4 6
GNMA -CDR contract (CBOT)
Ql
Q2
Q3
Q4
4.2
5.9
6 3
6 8
7.4
7.9
-
5.6
6 3
6.6
2.4
3.9
4.8
4.9
4.8
5.2
4 1
3 4
3 4
I
1 Sum of short and long commitments.
2. Figures could be as large as 200 percent because of the double counting of hedging commitments.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
I
CX)
I.,.)
I
-84cash positions in Treasury coupons as well as corporate bonds and other longterm instruments.
Cross hedging in the Treasury bond futures markets reportedly
is being used, both by those wishing to protect current holdings of corporate
and other bonds and by firms that anticipate borrowing in the bond market
some time in the future. 1
In the GNMA contract,large-trader hedging commitments declined substantially, both in absolute terms and relative to total open interest.
the three-month bill contract, hedging activity also declined.
In
Compared to the
falloff in total open interest in that market, however, large-trader hedging
commitments declined only marginally in relative terms.
Yet another source of information on the use of futures markets··vy
hedgers is data collected since February 1980 by the Federal Reserve Bank of
New York on the positions taken in futures and forwards by primary government
securities dealers.
Although these data are not available for a long enough
time to compare developments before and after October 1979, they make clear
that the government securities dealers were making active use of the bill
.
futures market in 1980.
As may be seen in Table 30, the dealers' bill-futures_
position (column 2) fell to a deep net short in April, rose steadily to a sub-stantial net long in the summer, and returned to a deep net short in late
November.
These movements were much greater than those in the dealers' cash
positions in bills (column 1), suggesting that dealers have been using the
futures markets as a vehicle
for speculating as well as for hedging.
1. A firm planning to borrow in the corporate bond market several months hence
could protect itself against a rise in bond yields by taking a short position
in the Treasury bond futures market. If interest rates did rise, the firm's
profit in the futures market would at least partially offset its loss in the
cash market (that is, its higher borrowing costs).
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
-85Table 30
Dealer Positions in Treasury Bills
(Par value in billions of dollars)
19_80
Cash
Futures
Forwards
Net
Feb. 29
1.8
-4.5
-.3
-3.1
Mar. 12
3.0
-3.7
-.3
-1.0
31
4.4
-7.4
-.3
-3.3
9
8.5
-8.6
-.2
-
30
8.0
-5.1
-.1
2.7
14
4.1
-1.6
-.3
2 2
30
3.9
.6
-.2
3.1
June 11
4.1
-
.3
-.7
3.1
30
1.4
1. 2
-.1
2.5
9
3.8
2.4
-.2
6.1
31
5.5
2.5
*
7.9
Aug. 13
5.5
4.3
-.1
:} . 7
29
3.1
3.1
i':
6. 2
Sept.IO
4.8
2.0
-.1
6.7
30
3.8
.4
*
4.2
8
2.4
-1. 2
*
1.1
31
3.1
-1. 6
*
1.5
Nov. 12
3.7
-3.0
.7
1.4
28
3.2
-7.9
.8
-3.9
Apr.
May
July
Oct.
*Less
than $50 million.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
.
.3
I
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
IMPACT OF DISCOUNT POLICY PROCEDURES
ON THE EFFECTIVENESS OF RESERVE TARGETING
January
1981
Paper Written for a Federal Reserve
Staff Review of Monetary Control
Procedures
by
Peter Keir
CONTENTS
I.
II.
INTRODUCTION
A.
Role of Window in Reserve Targeting ..............
2
B.
Questions Addressed by Paper ....................
4
SUMMARY OF FINDINGS
A.
B.
III.
~ffects on ~~isting Dis_count_ Policy on
Reserve-Targeting Strategy ..... .
6
Alternative Approaches to Managing
Discount Policy.. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
8
IMPACT OF EXISTING DISCOUNT POLICY ON RESERVE
TARGETING EXPERIENCE SINCE OCTOBER, 1979
A.
Relat1on of Actual to Expected Levels of
Borrowing
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1
Econometric test of borrowing relatLonship ... 17
B.
Relation of Borrowing to !Rate Spreads............. . . . 20
C.
Sources of Variability in the Borrowing Relationship. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
IV.
D.
How Well Has the Desk Coped with Deviations of
Actual Borrowing from Expected Levels? .......... 26
E.
1980 Experience with a Discount Rate Surcharge ... 30
ALTERNATIVE APPROACHES TO THE MANAGEMENT OF DISCOUNT
POLICY
A.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Maintaining the Discount Rate Consistently at
a Penalty. . . . . . . . .
........................
39
1.
The case for a penalty discount rate ........
41
2.
The case against a penalty discount rate ...
42
i
CONTENTS
Page
B.
C.
Adop~1on of a Tied Discount Rate ..
45
1.
The case for a tied rate. . . . . . . . . . . . . . . . . . . . . 48
2.
The case against a tied rate . . . . . . . . . . . . . . . . . 49
Establishment of a Graduated Structure of
1.
Specifics of~ a- graduated. discount rate p'lan .. 1•53
2.
The case for a graduated d1scoul}t rate. ·.·-·. . . 58
3.
The case against a graduated disc'ou:wt,,.,r,ate ... 60
Appendix A.
_Aggregat·e B0rr6w1pg and Money 'MaTr.ket ltnterest'
Rates
11
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
••r••··· •••..•.
Introduction
Because the FOMC's shift to a reserve-targeting strategy
places a higher premium than the earlier federal funds rate
strategy on estimates of expected levels of borrowing at the discount window, the question has been raised whether traditional
a~proaches to establishment of the discount rate and to administration of the discount window are sufficiently supportive of the
reserve-targeting process.
Although discount policy procedures
were modified somewhat by the recent revision of Regulation A,
Reserve Bank discount officers still rely pri~arily on broad
administrative guidelines to control borrowing by individual institutions.
As a result, there is still considerable looseness
in the relationship of total borrowing to changes in the spread
of the federal funds rate over the discount rate.
Moreover, since
the discount rate continues to be adJusted to changes in market
rates on a Judgemental basis and typically with an appreciable
lag, its spread against other rates tends to vary substantially
as market rates fluctuate
This paper attempts first to sort out how these features of discount policy have affected the reserve-targeting
process over the past year.
Note
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Then it considers whether alternative
While responsibility for drafting this paper was assumed
by Mr Keir, the paper draws heavily on, and quotes
liberally from, a set of papers on this subJect prepared
for the Board's Discount Policy Group by Messrs Paul
Boltz and James O'Brien, Dana Johnson and John Spitzer,
Perry Quick, and Ms Karen Johnson.
These other authors
should not be held responsible, however, for the summary
conclusions reached.
- 2 approaches to management of the discount rate and the discount
window might be more supportive of the process.
Role of Window in
Reserve Targeting
Under the old FOMC strategy of targeting day-to-day
transactions on the federal funds rate, erratic fluctuations in
the volume of borrowing at the discount window posed no significant problem for the System Account Manager.
The magnitude of
fluctuation in borrowing was generally moderate, and there was
no operating need for precise forecasts of expected levels of
borrowing.
Desk operations sought simply to provide or absorb
whatever nonborrowed reserves were needed to maintain the
federal funds rate within the Committee~s desired range.
The
level of borrowing was viewed essentially as an incidential
source of economic intelligence.
UnQer the new reserve-targeting strategy, however, the
role of borrowing has become more important.
Implementation of the
strategy starts with a forecast of expected borrowing derived 1n part
from recent discount window experience.
The staff then establishes
the nonborr0wed reserve path, which (when added to this expected
level of borrowing) will provide the total reserves thought,to
be consistent with the Committee's money growth targets.
As the period between FOMC meetings progresses, Desk
operations seek to keep nonborrowed reserves on this specified
path.
Any tendency for total reserves to deviate from their
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 3 desired path is then reflected in borrowed reserves.
The obJec-
tive of forcing more or less borrowi~g pn the banking system is
to press market interest rates in the direction needed to nudge
growth of money and total reserves back toward their desired
target paths in subsequent weeks.
If deviations in demands for
total and borrowed reserves appear to be large, compensating
changes may be made in the path for nonborrowed reserves in
order to accelerate the process of adJustment in interest rates
and the monetary aggregates!/
The reserve-targeting process will tend to be more
complicated if tne linkage between changes in borrowed reserves
and market i11terest rates proves to be erratic or unreliable.
Since the discount window is administered on a decentralized
basis by 12 different discount officers using rather general
operating guidelines, it would not be surprising if different
users of the discount window developed somewhat different
attitudes regarding the potential costs of additional borrowing
at the Federal Reserve.
To the extent actual borrowings reflect
such differences of view, there is some room for looseness in
the relationship between changes in borrowing and money market
interest rates.
!/
This reserve-targeting process is described in more detail
in pages 6-11 of the companion paper of this series by
Paul Meek and Fred Levin, 11 Implementinq the New Operating Procedures:
The View from the Trading Desk. 11
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 3 desired path is then reflected in borrowed reserves.
The obJec-
tive of forcing more or less borrowing on the,banking system is
to press market interest rates in the direction needed to nudge
growth of money and total reserves back toward their desired
target paths in subsequent weeks
If deviations in demands for
total and borrowed reserves appear to be large, compensating
changes may be made in the path for nonborrowed reserves in
order to accelerate the process of adJustment in interest rates
and the monetary aggregates!/
The reserve-targeting process will tend to be more
complicated if t11e linkage between changes in borrowed reserves
and market iriterest rates proves to be erratic or unreliable.
Since the discount window is administered on a decentralized
basis by 12 different discount officers using rather general
operating guidelines, it would not be surprising if different
users of the discount window developed somewhat different
attitudes regarding the potential costs of additional borrowing
at the Federal Reserve.
To the extent actual borrowings reflect
such differences of view, there 1s some -room for looseness in
the relationship between changes 1n borrowing and money market
interest rates.
1/
This reserve-targeting process is described in more detail
in pages 6-11 of the companion paper of this series by
Paul Meek and Fred Levin, 11 Impl ementin<1 the New Operating Procedures:
The View from the Trading Desk. 11
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 5 last spring, and most recently--in which the basic discount rate
was supplemented by a discount rate surcharge.
The second section of the paper then assesses the pros
and cons of alternative discount policy procedures that might be
substituted for the traditional approach.
are considered
Three general options
The first would establish a single discount
rate and maintain it consistently at a penalty above the federal
funds rate
The second would establish a single basic discount
rate automatically by
tyina
it in a fixed spread relationship
to a key market rate (most likely slightly below the federal
funds rate)
The third alternative would establish a graduated
structure of discount rates, with a base rate below the federal
funds rate, one or more steps above the base rate and a peak
rate set at a penalty above the funds rate
Quantitative limits
would be established on the amount of borrowing permitted at
each step below the federal funds rate, but there would be no
limit (other than prudent lending standards) on the volume of
borrowing at the top (or penalty) step.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Summary of Findings
Effects of Existing Discount Policy
on the Reserve-Targeting Strategy
1.
Even before the introduction of reserv~ targeting, the
volume of borrowed reserves--at given spreads of the federal
funds rate over the discount rate--s'howed considerable
variability.
Since the shift to reserve targeting, the variability
of this relationship has been accentuated, and the
average propensity to borrow (at given rate spreads)
seems to have increased somewhat.
These changes in borrowing patterns appear to have been
strongly influenced by the heightened interest rate
volatility that has emerged in money and securities
markets since the inauguration of reserve targeting.
2.
Despite the frequently wider spreads of market rates over
the discount rate since October 1979, the operating guidelines that discount officers use t'o administer access to
the discount window seem to have continued working reasonably
well.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
This has been particularly true for the largest banks
whose use of the window, although often sizable in
absolute dollar terms, is largely restricted to situations late in the day and at the end of a statement
week when misses on estimates of their reserve needs
and a temporary drying up of market sources of federal
funds sometimes force them into the window.
For midsized
regional banks with deposits ranging
around $1 billion there may have been some modest
erosion in the effectiveness of the guidelines as the
spread of market rates over the discount rate has
widened
But where active use of the window by such
banks has developed, this seems to have been partly
attributable to accommodative procedures developed
prior to adoption of the reserve-targeting strategy
when discount officers were especially sensitive to
the fact that banks in this size category were
weighing the pros and cons of abandoning membership
in the System.
- 7 3.
When actual borrowed reserves have deviated significantly
from expected borrowing, the FOMC Desk has managed to adJust
to these changes reasonably well, through counterbalancing
changes in nonborrowed reserves.
Since the Desk staff monitor the factors accounting for
such deviations regularly, there is leeway to adJust
the target for nonborrowed reserves on a timely basis
if such changes appear to be needed.
For various policy reasons, however, weekly respecifications of reserve targets to allow for latest
developments do not always seek to correct completely
fon deviations in total reserves.
4.
When actual growth patterns in money and total reserves have
drifted off the FOMC's desired target paths and borrowed
reserves have adJusted to compensate for the change, the
response of market interest rates has sometimes been quite
loose.
Once the shift to a different average level of borrowed
reserves has become clear, market interest rates have
responded in the direction needed to initiate adJustments in growth of the aggregates.
But the process of clarification has sometimes taken
several weeks, and the magnitude of the rate responses
to given changes in the level of borrowing have been
stronger in some periods than others,l/
5.
Changes in the Federal Reserve discount rate are, of course,
used to help reinforce the responsiveness of market interest
rates to adJustments in reserve-targeting policy.
If pursued too aggressively, however, such discount rate actions
risk inducing an overcorrection in market interest rates.
A discount rate surcharge was introduced on two occasions
1n 1980 in an effort to provide a clearer signal on policy
intent, while at the same time inducing a more moderate
response in market rates than might have developed from
an across-the-board increase in the basic discount rate
of the same size as the surcharge.
1/
The additional question whether the responses in interest
rates that'did occur were sufficient to induce timely adJustments in bank lending policies is a broader issue not
directly addressed by this paper.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 8 These obJectives of the surcharge approach appear to
have been realized, although it is difficult to isolate
the impact of each surcharge on market rates, since
other factors affecting financial markets were so
different in the two surcharge periods.
Since the surcharge was applicable only to large banks,
smaller commercial banks with minimal access to money
market sources of funds were still able to borrow at
the very favorable basic discount rate.
In the future, to keep the effective cost of discount
window credit in somewhat closer alignment for different
types of instit~tions that are active users of the
window, any use of the- surcharge may need t~ apply to
smaller, as well as to large, institutions. Since
smaller institutions would have no effective access to
money market sources of funds, how~ver, criteria for
the application of any such surcharge would have to be
more liberal for them than for large institutions.
6.
While the evident looseness in the linkage between changes
in borrowed reserves and changes in interest rates has not
negated the effectiveness of the reserve-targeting strategy,
it has posed the question whether there are alternative
means for managing discount policy that might be more
supportive of the strategy.
Alternative Approaches to the
Management of Discount Policy
1.
Maintaining the discount rate consistently at a penalty.
Advantages
This approach, by reducing borrowing to negligible
levels, would make nonborrowed reserves virtually the
same as total reserves and thus improve the Desk's
ability to hit its target for t9tal reserves.
Since a penalty rate approach would reduce the role
of the discount window as a buffer for adJusting the
reserve positions of individual banks, any change in
bank demands for reserves relative to the volume of
nonborrowed reserves being supplied through Desk open
market operations would exert a significantly sharper
reaction on market interest rates than is now the case.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 9 This would tend to induce quicker offsetting reactions
in bank lending strategies and lead to more rapid
general responses of deposit growth throughout the
banking system.
The penalty rate would also be an advantage over the
current system if it were thought that it was desirable
to reduce the subsidy to banks implicit in lending
them reserves at below market rates.
Disadvantages
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Under a penalty rate approach, if the volume of nonborrowed reserves being supplied by the Federal Reserve
were to fall significantly short of the volume of tota+
reserves being demanded by the banking system--as might
sometimes be the case--the federal funds rate would rise
especially sharply.
In fact, under the present system of
lagged reserve accounting, if nonborrowed reserves proved
to be insufficient to cover banks' required reserves, this
discrepancy could be covered only at the discount window
In such circumstances, the federal funds rate could be
expected to rise to a level above the discount rate.
Even temporary needs for reserves--arising from such
things as Desk misestimates of other factors, computer
breakdowns, or unanticipated demands for reserves against
nonmonetary liabilities--could thus be strongly reflected
in money market rates, even though they were unrelated
to changes in basic money growth rates. Sharp interest
rate responses to these strictly temporary needs for
reserves would tend to be counterproductive.
To the extent swings in market interest rates were
exacerbated under a penalty rate system, this could
&ake it more difficult for outsiders to Judge the
obJectives of monetary policy.
While resort to a penalty rate would tend to minimize
borrowed reserves and hence the Committee's need to
forecast expected levels of borrowing, changes 1n
- ,10 -
borrowed reserves could still be sizable at times.
Moreover, new wider margins of error would very likely
be introduced into staff estimates of excess reserves,
since banks when forced to rely more completely on the
federal funds market for back-up liquidity would be
more likely to adJust their cushion of excess reserves
as their concerns about possible near-term changes in
market rates waxed and waned.
2.
Adoption of a tied discount rate.
Advantages
A discount rate tied in a fixed relationship to some
market rate (most likely at a spread slightly below
the federal funds rate) would help to insulate the
volume of borrowed reserves against changes in market
interest rates.
Borrowed reserves would thus become less accommodative
of changes in the public's demand for money
While (under the current system of lagged reserve
accounting) there would be a technical problem if the
discount rate were tied to an immediate measure of the
federal funds rate--since this could produce an explosive interaction, with changes in the two rates feeding
on each other--this potential difficulty could be
modified by selecting a moderately backward-looking
rate average as the tJe.
The tied rate approach--by minimizing the spread of
market rates over the discount rate--would also reduce
the implicit subsidy that bank users of the discount
window obtain when market rates rise relative to an
inflexible discount rate
Disadvantages
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
As under a penalty rate scheme, the key disadvantage
of a discount rate closely tied to a market rate is
that borrowed reserves would be less responsive--not
Just to interest rate pressures arising from basic
shifts in the public's demand for money (which ought
not to be accommodated through the window), but also
to rate pressures arising from strictly temporary
deviations in reserve factors (which should be so
accommodated).
-
11 -
Where temporary reserve pressures of this latter type
did arise, a tied discount rate procedure would, thus,
accentuate changes in money market rates and tend to
lead to destabilizing pressures on the monetary aggregates.
Use of a tie that averages the rate series selected for
this purpose over a month or so would help to deemphasize
the effect of short-run stochastic deviations in reserve
factors and focus on more fundamental changes in underlying money demands. But this compromise would at the
same time continue to tolerate significant changes in the
spread of market rates over the discount rate, and unless
adJusted Judgementally would tend to place the discount
rate in a lagging penalty-rate position when market rates
were declining.
The obJective of achieving more timely adJustments in
the discount rate, which is the essential purpose
of a tied-rate system, might be better achieved, with
fewer technical complications and more balanced analysis,
if Judgemental actions that emphasized a discount rate
surcharge were used to supplement and reinforce changes
in the basic rate. This approach would retain the key
element of discretion, always so important in Judging
the full range of policy considerations that are involved in any discount rate action.
If a discount rate tied to a yield series on market securities were employed, in order to dampen the rate effects
of temporary stochastic influences, the rate series used
as the tie would have to be evaluated regularly to make
sure that no temporary influences were distorting its
relationship to the overall structure of market rates.
Where central banks in other countries have introduced
tied-rate procedures for setting their discount rates,
they have typically had to face up to the issue whether
the obJective of tying
should take precedence over
other policy considerations. Often, to accommodate
other policy requirements, the rules for implementing
the tied rates have had to be set aside.
After a period
of mixed results, these other ties have, therefore,
generally been abandoned.
3.
Establishment of a graduated structure of discount rates
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
12 _b.clv ,1,n tages
A graduated discount rate policy--wh.icb for example,,
mJ,ght establish a structure of di.scount rates, witbl a:
given quantity of funds availal:He to eligi.b'le b'orrowers
at each rate--would help to imprdve' the' ~ccura;ey of
Federal Reserve forecasts of borrowed reserves.
Since this approach could eliminate' aidm:un.s,traitJ v<:5
guidelines- on appropriatene·ss and frequiency, each
eligible borrower would be encoura:g,ed! to ptl.ls,fo its use'
of the window to the point whe·re· tlme mairg,1ma1] e0's,t 01:fr'
its permitted quota of borrowJ!..ng eql!l!a:1'.]ed! the prevall]lJ.'.n'g,
federal funds rate--i. e. ,, the· cost 0 1f ai]ternaitl!Ve' funds,.
Also, because the structure of rates, wo-u]d al])p·J'.y to, a:m
1.nstitut1.on' s borrowing llE! each staitememt wee·k ,, there·
would be no reason for 1. t to ]ook t01 llts· piais,t borrow.inig,
record or its expectatllons about fl!ltl!l!re sp1reads betweem
the discount and market rates. wbe·n 'ct1ee·:itcl.l"l!l!llg, whertl:ler t©l
borrow.
-·
«::
"-4!.,f"
,,
.,..
As a result, the Federal Reserve col!l!]dl "be· more· ccin;:l:'fd1en,t
that any basic change in the v01lume· od[ borr0wedi reserves;
would p,rodu.ce an associated! res,po,ns·e• iz:i1 mairke·t :i!mteres,t
rates, which would! then feed! back a])Jp,r0,p1riaiteTy 0n mone)~
demands.
Disadvantages.
ln l..9'2'3, fo,]Jlowing protes.ts, fr0m the• lD•ankl!]!llg, l!1!10_l-1ms,tr)3/
abolll!.t the 011.gh marg1.na] d!J!s,e0,unt rat,es, tb,at, we1reJ l'.mp,0,s,e,c:il
by some• Federal Rese•rv;e, BaI.J.,ks. cl!ur1mg the• llmme,d!J!aite, ]),0.s,t,World!. Wiar 1 period! o,f l!.F.l!f1at1.0Jn ,, the 11:eng,r,ess, riesc;imd'edl
language l!.n the F'edera] Ries,e•rv;,e, Ac,t,, whilleh h,a:<11.1 art!l,tl:J.©>rJ.!z,e,d.1
gradua t1..on of the d!iscoi"t!.l!nt rate on the; 0as,1s, OJ:f t,he.:
amm.1.n.t o,f crech t extended!.
Flo)r thlls, re.:a:S,0Jll>, t,here is_,
considerable doubt whether t,b,e Fled!e,ra:J! Re,se•rVte; n0,w, has.,
the authority to graduate the d'isc<D.il!ln>t. ra:t,e; on t;he
bas1..s of q_uant1..ty borrowed!.
Under the graduated d!l!. s,C©itl!nt rate pr0,ip,0,s,a:Jl ,, t:h.e-1 un,ll_=
verse of regular users, o)f the chs,c,e>,l!l!nt w1:l!lil_d!_0,w, w€1ullcll
jump to perhaps 2,0 ~ 00,0 ins;t}!tut 1.ons, ,, as, c,Qm~a:re,<i]!_ Wlll..t:b
no more than 400 day-to-d!a,y 1!l!S_,e,rs, €1:tf the, w11a~cl~i\~ 1W:t:1idle..:r·
present arrangements,.
Op,eratl!.ng s,t,a:f:f n.ee..Cile..d! t:€1 ~..:r·:r•@.._ti~e.:
loans,, administer co,llatera] ,, a:ndl m.Q.Ul!_t;Q,r fq,:r s,~ll-we.:a~G:¥'
could!, therefore~ be expected t:0, :i_n.c,rea6.e.: s;1.Ji"li~;ta.xrt'.!!.a.lll.~-
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 13 Because the mechanics of the proposed system are rather
complicated and the number of potential borrowers would
be greatly expanded, the transition to the new system
would probably require considerable experimentation.
As
a result, for some indeterminate time, actual borrowing
might well deviate from expected levels by more than it
has under existing arrangements, and there would always
be some erratic margin of error in the system because
the expanded universe of borrowers could not be expected
to effectively maximize its opportunities at all times.
Moreover, within the current statement week banks would
still need to make Judgements about the likely course of
the federal funds rate, and their efforts to allow for
such changes could continue to lead to some variance in
the relationship between borrowing and money market rates.
If the discount rates on the initial tranches of credit
provided in the graduated structure were set consistently
at levels below the federal funds rate and there were
no administrative guidelines on borrowing, users might
tend to view such credit as a low-cost basic borrowing
privilege that could be used to finance longer-term
uses than are permitted under existing discount window
guidelines.
To the extent this proved to be the case,
although the amount ,of credit released to any particular
borrower would not be large, the arrangement would
entail an element of subsidy that might' not be warranted.
- 14 Impact of Existing Discount Policy on ReserveTargeting Experience Since October 1979
To help Judge how existing discount policy procedures
have affected the reserve-targeting process since its inception
in October
1979, this part of the paper addresses the following
questions.
To what extent have borrowed reserves differed from
their targeted levels?
Where deviations have been apparent, what
factors seem to have accounted for them?
How has the Desk
adJusted its operations to cope with such deviations?
And
finally, how successful have these Desk adJustments been in
keeping deviations in borrowed reserves from hindering effective
implementation of the reserve-targeting process?
Relation of Actual to Expected
Levels of Borrowing
Table 1 compares actual with expected levels of daily
average borrowing for the
sub periods
between FOMC meetings on
which the Trading Desk focused its intermeeting reserve targeting
operations
As would be expected, actual borrowing tended to
exceed anticipated levels by the widest. margins when growth in
the monetary aggregates appeared to be pushing well above targeted
rates and market interest rates were under significant upward
pressures
The most dramatic overshoot in actual borrowing
occurred in the first half of
of possible new administration
March, when widespread discussion
programs to control inflation
triggered large anticipatory drawings on the discount window
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 15 Table 1
Actual Minus Path Levels of Borrowed Reserves
For Target Periods Between FOMC Meetings
From Early October 1979, to Mid-November 1980
(Millions of dollars) l/
Time
period
(1)
Actual
(2)
Targeted
(3)
Deviation
1979
4 weeks ended
October 31
2,119
1,500
+
619
3 weeks ended
November 21
1,885
1,500
+
385
4 weeks ended
December 19
1,607
1,700
3
3 weeks ended
January 9
1,128
1,550
422
4 weeks ended
February 6
1,251
1,000
+
251
3 weeks ended
February 27
1,830
1,317
+
513
3 weeks ended
March 19
2,982
1,650
+l,332
weeks ended
April 23
2,427
2,600
173
696
1,525
830
4 weeks ended
June 18
142
111
3 weeks ended
July 9
68
73
1980
5
-------4 weeks ended
May 21
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
+
31
5
- 16 Table 1 (continued)
Actual Minus Path Levels of Borrowed Reserves
For Target Periods Between FOMC Meetings
From Early October 1979, to Mid-November 1980
(Millions of dollars) l/
Time
periods
(1)
Actual
(2)
(3)
Targeted
Deviation
1980
5 weeks ended
August 13
239
75
+
5 weeks ended
September 17
697
225
+
472
"
164
5-weeks ended
October 22
1,300
950
+
350
4-weeks ended
November 1.9
1,841
1,450
+
391
!/
Computed from final data and adJusted paths.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 17 In the periods when growth of the aggregates appeared to be falling
short of expectations--such as late 1979 and the spring and early
summer of 1980--borrowed reserves ran below expected levels.
While both the expected levels of borrowed reserves and
the deviations from these levels (shown in Table 1) are generally
consistent with patterns that would have been anticipated from
a reserve-targeting strategy; the size and variability of some of
these deviations raise a question whether the level of borrowed
reserves is as closely linked to the spread of market rates over
the discount rate as the logic of the reserve targeting strategy
implies.
Econometric test of borrowing relationship.!/
focus more rigorously
To
on the recent relationship of borrowed
reserves to this interest rate spread, the weekly experience for
the period since October 6, 1979, was compared through econometric analysis with the relationship that prevailed throughout
the 1970s.
This analysis suggests that for a given spread
between the federal funds and discount rates, borrowing at the
discount window has become considerably more erratic since
October 6, 1979.
It also suggests that borrowing at given
spreads may have become somewhat
than before October
larger in the recent period
1979.
1/ This and the following two sections of the paper are taken largely from
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Dana Johnson and John Spitzer, Recent Behavior of Member Bank Borrowing
at the Discount Window, paper presented at the Financial Analyst Conference
at the Federal Reserve Bank of Minneapolis, June 12 and 13, 1980.
11
11
- 18 -
A variety of econometric relationships were estimated
relating member bank borrowing to the spread between the federal
funds rate and the discount rate and to an assortment of other
variables
that
measure various aspects of conditions in the
money market.!/
Weekly data running from the beginning of 1970
through September 1979 were used in the analysis.2/
The results
strongly suggest that the level of borrowing is significantly
correlated with the spread between the two interest rates
but
not with the absolute level of those rates.
When the level of required reserves is added to the
analysis, the results indicate that there is also some tendency,
albeit quite weak, for member bank borrowing to increase along
with the level of reserves.
Also, borrowing appears to be
greater in the initial week in which there is an increase in
required reserves than in subsequent weeks.
This suggests that
the discount window is serving one of its desired purposes,
namely allowing1banks to make an orderly adJustment to a change
in required reserves by borrowing initially and then adJusting
their portfolios.
!/
2/
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The appendix explains the specifics of this econometric
analysis.
The emergency borrowing by Franklin National Bank in 1974
was treated as if it were nonborrowed reserves and excluded
from the data. For the recent period emergencv borrowing
by the First Pennsylvania Bank was handled in the same way.
- 19 A few variables tested
that showed little relation to
levels of borrowing also are important to note.
Several different
measures of the degree of administrative pressure for all member
banks were added to the test models but were never close to being
statistically significant.
For example, such variables as the
number of banks of all sizes that have borrowed relatively frequently in the preceding 13 weeks, and the dollar amounts of
borrowing by such banks, do not help to explain the level of
member bank borrowing.
However, when the analysis is restricted
to borrowing banks with deposits of $500 million or more, there
is evidence that the past frequency of borrowing has had an
influence on the spread
The conclusion reached from the econometric analysis
is that since October 9, 1979, the link between borrowing and
the spread of the federal funds rate over the discount rate
has become considerably more variable than it was before that
date.
In addition, it appears that at given spreads between
these two rates, there may have been a greater willingness to
borrow in the recent period. Specifically,
this admittedly still
tentative finding suggests that an increase of 10 basis points
in the spread of the funds rate over the discount rate has been
associated on average with an increase in borrowing at the window
of $110 million, whereas for the period tested before October 6
a similar change in the spread was associated with an increase
in borrowing of only $40 million.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 20 -
Relation of Borrowing
to Rate Spreads
Because borrowed reserves have expanded in tandem with
the spread of the federal funds rate over the discount rate, some
analysts believe that such changes in spreads have operated as
an incentive for banks to expand their use of Federal Reserve
credit and that this has contributed in the process to an
excessive growth of the monetary aggregates.
However, the same
evidence on the relationship of borrowing to rate spreads can
also be interpreted as support for th~ traditional view that
greater bank reliance on borrowed reserves helps to dampen the
expansion of money and bank credit
As noted earlier, when incoming data indicate that
growth in money and total reserves is exceeding desired rates,
the Desk holds to its target for nonborrowed reserves and forces
'
banks to increase their borrowing at the discount window.
If
the overshoot in growth rates seems very large, the target for
nonborrowed reserves may also be adjusted downward, forcing a
still larger volume of borrowing on the banking system.
Since
individual banks are expected to utilize alternative sources
of funds before coming to the discount window,they turn first to
these other sources, in the process, they bid the funds rate
and other short-term market rates up relative to the discount
rate.
Because the volume of nonborrowed reserves available to
the banking system is being constrained by the Desk, the supply
of federal funds actually available to accommodate bank demands
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 21 is more constrained than before.
Consequently, some banks are
forced to expand their borrowing at the discourit window and are
brought under increased administrative pressure.
Looking back
on any such shift in the mix between nonborrowed and borrowed
reserves, it will generally be evident that the increase in
borrowing was significantly correlated with a widening of the
spread of the federal funds rate over the discount rate
But
the causation can be viewed as coming primarily from the Desk,
since it forced more borrowing on a reluctant banking system by
reducing the relative availability of nonborrowed reserves.
Bank decisions on when to borrow at the discount
window are, of course, influenced by prevailing and expected
spreads between the discount rate and market rates.
The step-
up of borrowing that developed in the weeks of 1980 when increases
in the discount rates were widely anticipated is one obvious
example of such behavior.
But the econometric analysis suggests
that erosion in the effectiveness of the traditional guidelines
for administration of the discount window since the shift to
reserve targeting has actually been substantially less than might
have been expected.
If any sizable number of banks were now
persistently abusing their use of the window in an effort to
maxmize returns from favorable interest rate spreads, one would
expect to see much less upward pressure on the federal funds
rate relative to the discount rate than has actually developed.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 22 -
When one considers the uses to which Federal Reserve
credit is expected to be limited, it is not so surprising
that wider spreads between the discount rate and rates on market
sources of funds have failed to produce larger changes in the
borrowing relationship.
Use of the window to expand investment
portfolios or to promote a program of concerted expansion in
lending is strictly prohibited, as well as borrowing to sell
federal funds, or to reduce regular patterns of reliance on sales
of large CDs
and Eurodollars.
On the relatiyely rare occasions
when maJor banks with ready access to domestic and international
money markets do turn to the discount window, they are generally
expected to repay these borrowings on the next business day and
to limit such uses to no more than a limited number of weeks in
any quarter.
While medium-sized and smaller banks that lack
such ready access to money markets are granted leeway to turn to
the Federal Reserve more often and for somewhat lonqer periods, their use
of the discount window is also substantially constrained.!/
Sources of Variability
in the Borrowing Relationship
The econometric analysis did, nevertheless, identify
considerable variability in the relationship of borrowed reserves
!/
Some analysts have asked why--if increased use of the discount
window is supposed to dampen banks' enthusiasm for expanding
credit--there was not more evidence in the latter part of
1980 that bank lending activity was being so constrained.
One answer to t~is question is that even though use of the
window was relatively large in certain periods of 1980, for
large banks funds obtained through the window still represented a small margin of their total fund availability. For
smaller bank8 that used the window, the need to repay borrowing at the discount unndow did exercise a more evident
constraint, presumably because t~ey have few alterhative
sources.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 23 to the spread of the funds rate over the discount rate.
While
there was variability of this type even before October
1979,
it appears to have increased significantly since then.
This
raises two questions
first, what accounted for the variability
in the relationship before October
1979, and second, what
accounts for the evident increase since then?
The primary explanation for the normal variability in
the aggregate borrowing relationship is that banks' attitudes
about borrowing are quite diverse.
Some banks choose to avoid
.
borrowing at the discount window under virtually all circumstances,
others choose to borrow somewhat more freely but considerably
less often and for smaller amounts than the numerical guidelines
permit, while still others apparently attempt to borrow as much
and as often as the rules permit if it 1s profitable to do so.
In
such circumstances, aggregate borrowing behavior in a given week
reflects the composition of the group of banks that need to
obtain additional reserves.
For example, the federal funds rate
may tend to be a bit lower for any given provision of nonborrowed reserves in the weeks in which banks with high propensities to borrow at the discount window happen to have
relatively large reserve shortages.
Another possible source of variability in the intensity
of borrowing has been the fact that banks have been subJect to
administrative standards
that have varied slightly in their
relative stringency from Reserve District to Reserve District.
Numerical guidelines used to oversee the legitimacy of borrowing
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 24 requests have not been wholly uniform.
And the interpretation of
what constitutes "appropriate" borrowing has sometJ.mes differed a
bit from one discount officer to another.
As a result, some bor-
rowers may have had more leeway than others to take occasional indirect advantaf;e of the arbitrage opportunity that exists when there
is a large spread between the federal funds rate and the discount rate.
If in ~ny given circumstance, a disproportionate number of banks
that are subJect to somewhat more stringent borrowing guidelines
need to obtain additional reserves, relatively more pressure
will tend to develop in the funds market.
Both of these explanations for the normal variability
of the borrowing relationship stem from aggregating banks with
different propensities to borrow.
The variability among banks
in their attitudes about borrowing and the slight difference
in adl.111nistrative practices from Reser1S.Te District
District
to Reserve
may have created a situation in which the aggregate
level of borrowing has fluctuated as the banks that are choosing
between borrowing reserves in the funds market or borrowing
directly from the discount window have changed from week to week.
Neither of these explanations, however, would appear to explain
adequately the increase in the variability of borrowing since
October
1979.
Even though there is evidence that banks in the
aggregate have been slightly less reluctant to borrow since
October 6, there is little reason to suspect that banks' attitudes
towards borrowing have become dramatically more disparate.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 25 Moreover, there is no evidence to suggest that adm1n1strat1ve
practices have become more variable from District to District
since October 6.
In fact, with the recent revision of Regulation
A and the introduction of uniform operating guidelines,
any
lingering variation in discount window administration across
District
lines should be effectively minimized.
The most plausible explanation of the additional vari-
ability in the borrowing relationship is the substantially increased volatility of interest rates that has emerged over most
of the period since October
1979.
Limits on the frequency and
amount of discount window borrowing, whether self imposed or not,
imply that a bank when deciding whether to borrow today may consider the likelihood of wanting to borrow in the near future
Hence the expected future behavior of the spread between the
discount rate and the federal funds rate, in addition to the
current spread, may influence its borrowing decision.
The increased volatility of the funds rate reflects
the fact that demands for reserves are not accommodated under
the new reserve-targeting procedures.
For this reason the near-
term outlook for the funds rate is now far less certain than it
was before October
1979, when even week~o-week changes in the
funds rate usually were relatively minor and a bank could often
safely assume that the spread was likely to continue to be
about the same for the foreseeable future.
A bank, therefore,
is now much more uncertain about the relative advantage of
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 26 borrowing at any particular time.
Under the new procedures,
banks must try to assess the demand for money and whether such
demands are likely to lead to increased or decreased pressures
in the funds market.
For example, if it were generally expected
that money demand would start exceeding the Federal Reserve's
near-term targets, banks might become somewhat more reluctant
to borrow, anticipating that the funds rate would soon rise
relative to the discount rate and make future borrowing more
attractive than current borrowing.
How Well Has the Desk Coped
with Deviations of Actual
Borrowing from Expected Levels?
The fact that actual borrowed reserves can deviate
significantly from expected levels obviously complicates the
task of implementing the reserve-targeting strategy.
However,
as described in the previously cited paper by Messrs. Meek and
Levin, the Desk has exhibited sufficient operating flexibility
to hit its targets for nonborrowed reserves quite effectively.
By making selective adJustments in nonborrowed reserve targets
from time to time, it has also managed to offset a sizable part
of weekly deviations in borrowed reserves from their expected
levels.
This has not kept total reserves from tnissing their
target paths at critical points.
But these misses have been
partly a reflection of policy Judgements that corrections for
such deviations should be accomplished over more extended periods
than the immediately succeeding statement weeks.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Thus, the new
- Z7 target levels for nonborrowed reserves ' that were specified
following receipt of new weekly data on depos1ts~and-resertes
often did not seek to correct completely for recently observed
deviations in borrowed and total reserves
For this reason,
, it is not vaild to assume that observed overshoots or undershoots
of 1ntermeeting targets for total reserves are attributabl~ in
any significant degree to the.staff's 1nabil~ty to correctly
forecast borrowed reserves~
In fact, devia~ions- resultin~ from
~
misspec1ficat1ons of expec~ed eorrowing do not appear, by them-
•
selves, t~ have imposed a maJor obs~acle to reasonable at\ainment- ef the desired reserve targets.
Interest rate response.
As- might have been ex1>-ec-ted.,-
the ~ink between polfcy-induced changes
in-
the level of boFrowed
reserves and the interest Fate responses needed to encourage
banks to adJust their creation o~ deposits and credit has
proved to be more tenuous.
Like the FOMC Desk, market parti-
cipants cannot be sure whether any observed change in the current
level of borrowed reserves reflects strictly temporary influences
'
likely
to be soon reversed or something more fundamental.
As
a result, immediate changes in borrowed reserves are not generally viewed by the market as reliable signals of adJustments
1n the FOMC's policy posture.
Thus, over short time periods
.,.
changes in borrowed reserves and money market rates are not very
tightly linked, as
can ~e seen in Chart I
In fact, directions
of change 10 the two are occasionally divergent
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Cha1;tt
Borrowings
(Billions)
2.0
i
Changes in Adjustment Credit Bor~ow!~gs
CQmp~red with th~
Difference pf the Fede~al Funds Rate
M~nus the O!sc~unt:~~t~
(Weekly averag~li)
Rate
Differential
(Pevcentage
points)
---"1"------------"1""----,.,--------........------...-----..--------------- 3, o
...
Solid Curve - left scale
1.6
2.4
Dashed Curve - right scale
I
~
-
'
"
11
1.2
,.
,1
I 1
I I
11
I 1
- • l
I I
.8
I
1
I
I
I
I
I
I
t,
I
I -l
,4
I
0
1.2
''
I 1
I
I
J
I
I l
+-- Dtscount
~t,a-.,,;.;..,.+
a~ Penalty
I I
• J 1 I
I I \ l
I
.6
t,.:)
CXl
,,
0
. 1
\
,4
1.8
II
.6
\
'I
'
I
.8
I
I
l
1.2
1 6
1.2
l
'l'•.
'-------'----'--------'----".,___ __.,____.,_r-_,.,...-,....,............._....,...,___ _..,_____....,..,______..._..,___. 2,4
Jul.
Aug. 'S~pt. Oc~.
Nov.
nee.
Oct.
Nov.
Jan.
Mar. Apr.
Dec.
Feb.
1979
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
1980
- 29 -
Nevertheless, any persistent tendency for borrowed
reserves to expand or contract over several statement weeks
is usually reflected (as the chart also suggests) in changes
of money market rates that move in the same direction, even
though the degree of rate response to given changes in borrowing
has been varied.l/
In the late summer and early fall of 1980, for example,
a significant pickup
in the level of borrowing was not associ-
ated with as large a rise in money market rates as the Desk
was anticipating
Undoubtedly this sluggish response reflected
the fact that the banking system had been virtually out of debt
at the Federal Reserve over most of the summer.
Since few banks
were even close to coming under administrative restraint at the
discount window, the sizable dollar increase in borrowing
apparently exerted less than normal pressure on money market
rates
Later in the fall, however, after many banks had begun
to come under administrative pressure at the discount window,
the imposition of a discount rate surcharge on top of an increase
in the basic rate produced a somewhat sharper general reaction in
market interest rates than was initially expected.
The uncertain dimensions of interest rate responses
to changes in the level of borrowed reserves 1s not so surprising 1n a money market where participants I expectations are
l/
These
relationships are also discussed in Levin and Meek,
11
Implementing the New Procedures. 11
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 30 affected by such a wide range of influences, particularly since
the volume of borrowed reserves is often not a maJor element in
this total complex.
Nevertheless, in view of the general
importance of interest rate responses to the logic of the
reserve-targeting process, the evident looseness in the relationship between borrowing and interest rates is a cause for some'
concern.
Such doubts lead logically to the question whether
there is an alternative means of managing discount policy that
might make it easier to predict both the likely volume of bor-,
.
rowed reserves and the relationship of changes in that borrowing
to changes in money market rates
Before turning to a considera-
tion of alternative discount policies, however, the 1980 experience
with a discount rate surcharge needs to be reviewed, since this
represented a rather significant effort to improve the flexibility of existing discount policy.
1980 Experience with a
Discount Rate Surcharge
A surcharge was added to the basic discount rate at
two points during 1980, first between March 17 and May 7, and
then again starting on November 17 and running into 1981.
In
each case the surcharge was applied to borrowings by banks with
deposits of $500 million or more, where the borrowings occurred
in two successive statement weeks or in more than four statement
weeks of a quarter
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 31 -
The spring surcharge of 3 percentage points was added
to the then prevailing basic discount rate of 13 percent.
fall surcharge
when initially imposed in mid-November
The
was 2
percentage points and was introduced in conJunction with a 1
percentage point increase in the basic discount rate, from 11
to 12 percent.
On December 5, after market interest rates had
experienced further steep advances, both the surcharge and the
basic discount rate were raised an additional percentage point
and were still in effect at the turn of the year
Table !.which compares the number of banks and the
amounts of borrowing that were subJected to the surcharge during
the spring and fall periods, shows that the incidence of the fall
surcharge was greater.
Much of this difference is attributable
to the fact that the fall surcharge, although introduced in
the middle of the quarter, looked back to the number of weeks
of borrowing in the quarter as a whole-in determining when
banks had exceeded the four weeks per quarter trigger that
subJected them to the surcharge.
The spring surcharge, on the
other hand, looked only forward.
Since it was introduced Just
two weeks before the end of the January-March quarter, there
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 32 Table 2
Discount Window Borrowings
SubJ ected to Surcharge }j
Statement
week endinq
Surcharge
(percent)
Amount subjected
to surcharqe
(millions of
dollars)
Number of
banks
Spring surcharge
period
03/26/80
3
77
7
04/02/80
3
68
3
04/09/80
3
55
2
04/16/80
3
101
4
04/23/80
3
110
5
04/30/80
3
135
5
05/07/80
3
11
1
11/19/80
2
124
5
11/26/80
2
463
24
12/03/80
2
669
30
12/10/80
3
182
14
12/17/80
3
395
18
Fall surcharqe
oeriod
1/ "Extended" borrowing by the First Pennsylvania Bank during the
-
period March 26-April 22, 1980, is excluded from these data.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 33 -
was no chance for the four weeks per quarter exemption to be
exceeded until five weeks into the April-June quarter (1 e
the statement week ending May 9),
Because this was also the
final week of the spring surcharge period, the banks that paid
the surcharge in the spring period did so almost entirely
because they borrowed in successive statement weeks.
Banks
subJected to the surcharge in the fall period, however, were
also substantially affected
because they had borrowed in more
than four statement weeks of the quarter
The two surcharge
periods were further differentiated by the application of direct
credit controls in the spring and not in the fall,
and by the
greater de~ands for bank credit that developed in t~e fall
b~cause the ecoaomy was substantially stronger than in the spring.
Rationale for surcharge strategy.
The rationale for
imposing a discount rate surcharge on top of the basic discount
rate has been to achieve a more selective restraJnt on use of
the discount window
and thus a somewhat smaller general response
in market interest rates than might have been expected from an
equivalent across-the-board increase in the basic discount rate
Under this selective approach, the greatest restraint is imposed
on large institutions that have been the largest and most active
users of the window, while institutions that have been infrequent
borrowers
and small institutions that have no effective access
to national money markets are still charged only the basic
discount rate.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 34 Banks whose past use of the discount window makes them
vulnerable to payment of the surcharge generally seek to avoid
it by increasing their reliance on the federal funds market.
N
As a result, the spread of the funds rate over the basic discount rate tends to widen relative to what it was before imposition of the surcharge
For the spring surcharge period this
process apparently led to a widening of the spread of the funds
rate over the discount rate that amounted to roughly two-thirds
of the surcharge itse~f
For the fall period, the
widening of the spread appears to have been about equal to the
surcharg9
Th8 fact that the spring discount rate surcharge was
announced Jointly with a direct credit control program, at a
time when there was growing evidence that the economy was
weakening, apparently triggered a broad market consensus that
a general interest rate downturn was at hand.
Over most of the
fall surcharge period, however, many market participants continued to anticipate significant further interest rate advances.
Impact of surcharge on reserve targeting.
To Judge
how imposition of the discount rate surcharge may have influenced
the effectiveness of the reserve-targeting strategy, two questions
need to be addressed
First, have deviations of borrowed reserves
from expected levels been any more of a problem in surcharge
periods than at other times?
Second, has the response of market
rates to changes in borrowed reserves provided any clearer
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 35 signal on the thrust of reserve-targeting policy during surcharge periods than at other times?
Actual borrowed reserves ran unusually high relative
to expected borrowing in the Mar~h 19, April 23, and April 30
statement weeks of the spring surcharge period.
But these
overshoots were essentially a reflection of other influences.
In the week ending March 19, borrowing rose in anticipation of
the widely presaged general credit control program, while at
the end of April overshoots of borrowing relative to the staff's
reduced estimates reflected the fact that the market had not
yet caught up with the Desk's shift to a strategy of more
liberal provision of nonborrowed reserves
In general, the existence of a surcharge does not seem
to have made the Desk's task of coping with misses in borrowed
reserves any more difficult than it was in weeks when there was
no surcharge.
Moreover, the 1980 experience suggests that a
discount rate surc~arge can be helpful in communicating the
current thrust of reserve-targeting policy to market participants
When a surcharge is first introduced, it has an announcement
effect similar to that of a change in the basic discount rate.
Consequently, if future use of the surcharge were to lead to
more frequent
changes in the structure of discount rates, this
could be expected to communicate the thrust of policy to market
participants more clearly than a strategy that relied largely
on shifts in the mix of borrowed versus nonborrowed reserves
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 36 as the catalyst for inducing desired adJustments in market
interest rates.
Use of a discount rate surcharge has sometimes been
advocated as a means of narrowing the effective spread between
the basic discount rate and market interest rates.
Under a
reserve-targe~ing procedure, however, this result could be
effectively achieved only if--after the surcharge
was imposed--the FOMC were willing to adJust its target for
nonborrowed reserves upward
That is, after the action, a
larger share of total reserves would need to be provided through
open market operations and a reduced share through bank reliance
on the discount window.
Such an approach might be rationalized
if it were thought that further general advances in market
interest rates ought to be constrained.
If after the imposi-
tion of a surcharge, the Desk did not change its target for
nonborrowed reserves (as was the case in 1980), banks seeking
to avoid payment of the surcharge would simply bid the federal
funds rate and other market rates higher, reestablishing spreads
over the surcharge that were not too different from the spreads
over the basic discount rate that had prevailed before.
Even though more frequent use of a discount rate surcharge could be expected to help reduce confusion in market
circles rega~ding the direction of interest rate responses to
reserve-targeting policy, there could still be considerable uncertainty as to the likely degree of interest rate response
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 37 The smaller reaction of market rates in 1980 to the spring surcharge than to the fall surcharge is a useful reminder that any
actual response of market rates can be significantly influenced
by the nature of other market forces prevailing at the
time.
The application of the 1980 surcharge only to large
banks permitted many users of the discount window--particularly
smaller banks--to continue obtaining Federal Reserve credit at
the basic rate
The basic rate sometimes ranged to as much as
7 percentage points below rates that depository institutions were
paying for funds in the market, or that thrift institutions were
paying to borrow from their special industry lenders.
To help
minimize such differences in the future, it may, therefore,
become necessary to apply any needed discount rate surcharge
to borrowings by institutions of all sizes--not Just to large
institutions
If this approach were followed, criteria
for applying the surcharge (on consecutive weeks of borrowing,
and number of weeks of borrowing within a quarter) would have
to be more liberal for smaller than for large institutions,
since the former have little access to money market sources of
funds.
In the past the substitution of a discount rate surcharge for a general change 1n the basic rate has always been
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 38 advocated as a means of limiting the risk of overkill in discount
rate actions
But if the surcharge were applied more generally
to institutions of all sizes, its potential advantage as a
selective instrument in discount rate administration would tend
to be reduced.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 39 Alternative Approaches to the
Management of Discount Policy
If looseness in the relationship between changes
in
borrowed reserves and money market interest rates should prove
to be greater than can be tolerated under existing procedures
for managing discount policy, are there alternative approaches
that might be more supportive of the reserve-targeting process?
As suggested earlier, the three discount policy approaches that
seem to capture the range of alternative possibilities most
effectively are: (1) maintaining a single discount rate consistently at a penalty above the federal funds rate, (2) tie1ng
a single discount rate at a fixed spread somewhat below the
federal funds rate, and (3) establishing a structure of discount
rates graduated according to the quantity of borrowing, with the
base rate in the structure set below and the peak rate at a
penalty above the federal funds rate.
Assessment of the pros
and cons of these various options against the background of
actual experience with existing procedures will help to place
the relative merits of both in better perspective
Maintaining the Discount Rate
Consistently at a Penalty!/
The rationale for shifting to a policy of maintaining
the discount rate always at a penalty above the federal funds
rate is to make the volume of borrowed reserves more predictable
by reducing it to minimal levels
Because of the penalty rate,
use of the discount window would be effectively constrained to
l/
This and the section on a graduated discount rate drew heavily
from a paper prepared by Perry Quick, "Discount Window Policies
without Administrative Pressure. 11
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 40 infrequent--but (on a day-to-day basis) relatively predictable-borrowing for emergency purposes
plus
a small amount of
frictional borrowing for adJustment purposes
largely
Since this
approach would make nonborrowed reserves virtually the same as
total reserves, the Desk's ability to hit its total re~erves
target should be less affected by deviations in borrowed reserves.
Adoption of a penalty rate arrangement would obviously
diminish the role of the discount window as a buffer for adjusting
the reserve positions of individual banks.
Banks seeking
to cover reserve deficiencies would rely largely on the federal
funds market as the less expensive source of credit and would turn to
the discount window only when their access to cheaper market sources
had been temporarily shut off, or when they had developed an
emergency need for funds
Similarly, banks with reserve sur-
pluses would rarely use them to repay debt at the discount
window since there would be few such loans outstanding.
With the discount window thus relegated primarily to a
lender of last resort role, the supply of total reserves would
be less adJustable than under the present system.
As a result,
any change in bank demands for reserves relative to the volume
of nonborrowed reserves being supplied through Desk open market
operations would exert a significantly sharper reaction on market
interest rates than is now the case.
Thus, the key question
at issue about this option is whether these sharper interest
rate responses would facilitate or inhibit the implementation
of monetary policy
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 41 The case for a penalty discount rate.
Those favoring
a penalty discount rate Jase their case on two maJor points.
First, they believe it would avoid certain key difficulties
that have emerged under the existing system.
Second, they
believe it would make monetary policy respond in a much more
timely fashion to the critical changes in money demands that
are linked to undesirable movements in economic activity.
Under existing arrangements banks tend to view their
borrowing at the discount window as Just another form of managed
liability.
The volume of Federal Reserve credit demanded,
therefore, tends to reflect relative spreads between tne discount
rate and other money market rates
However, borrowing by in-
dividual banks also reflects their varying individual assessments
of the implicit costs of running afoul of the administrative
guidelines discount officers use to control use of the window.
Since the Desk has had considerable difficulty forecasting how
the combined effects of these two types of influences are likely
to be reflected in the level of borrowed reserves for any given
statement week, a penalty rate system that minimizes their
importance should help to avoid inaccurate estimates of borrowed
reserves.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 42 -
Finally, under the penalty rate approach a shift in
banks' demands for reserves stemming from changes in the public's
demand for money would induce a much sharper movement in the
federal funds rate than before.
This would induce quicker,
offsetting reactions in bank lending strategies and lead to
more rapid general responses of deposit growth throughout the
banking system
The case against a penalty discount rate.
The con-
tinuous maintenance of a penalty discount rate would amount to
the pursuit of a total reserves or monetary base path
Adherence
to such a policy would, therefore, effectively eliminate the
important role of the discount window as a buffer for accommodating temporary needs for services that have no relation to
changes in underlying demands for money (such as changing demands
for excess reserves, or changes in the deposit mix affecting
required reserves)
Under such an arrangement, any discrepancy
between nonborrowed reserves provided by the Federal Reserve
and total reserves desired by banks would thus lead to very
sharp interest rate adJustments.
As a practical matter, a strict penalty system of
this type would not be possible under the existing system of
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 43 lagged reserve accounting, the only way the path for total
reserves could be maintained when Desk misestimates of
reserve factors failed to supply the nonborrowed reserves needed
to cover banks' required reserves would be for the federal funds
rate to rise temporarily to a level above the discount rate.
This would then induce the volume of borrowing needed to make
up the reserve deficiency.
Under present arrangements, the penalty rate approach
would, therefore, have important deficiencies.
stantially accentuate
It would sub-
the short-run volatility of interest
rates, leading at times to much sharper interest rate reactions
than were consistent with the FOMC's obJectives on money and
reserve growth.
Although it would effectively minimize the
volume of borrowed reserves, there would still be occasional
needs for adJustment borrowing because the Desk failed to provide
the appropriate volume of nonborrowed reserves
Consequently,
variability in the volume of borrowed reserves might remain unexpectedly sizable.
The increased volatility of money market rates likely
to develop under this approach could also be expected to make
staff estimates of excess reserves more difficult.
In weeks
when banks were uncertain whether money market rates might rise,
they would tend to carry a larger cushion of excess reserves
than at other times.
On average, therefore, fluctuations in
the level of excess reserves could become significantly wider
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 44 and less predictable than they are now.
Consequently, there
would be a significant potential for the multiplier for total
reserves to become quite erratic under this scheme.
If this
contributed to more erratic short-term movements in money
market interest rates, it could distort and delay bank responses
to System policy initiatives.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 45 Adoption of a Tied
Discount Rate!/
As a second possible alternative to existing discount
policy, it has been suggested that the discount rate might be
tied 1n a constant spread relat1onsh1p slightly below the
federal funds rate, or 1n some fixed spread relationship to a
key alternative money market rate series.
Like the penalty
rate option Just discussed, the obJective of this approach
would be to insulate the volume of borrowed reserves against
changes in market interest rates
so that adJustments to per-
sisting deviations from targeted money growth rates would occur
more quickly.
The precise effects of any tied rate scheme on
the reserve-targeting process w1ll 1 of course, depend importantly
on the particular market rate selected as the tie, and the
mechanics of how the tie 1s applied.2/
Most recent proposals for a tied rate recommend that
the federal funds rate be used as the tie.
the fact that among banks that
This choice reflects
account for the lion's share
of borrowing at the discount window, purchases of federal funds
are the closest substitute for borrowed reserves.
Some older
tied-rate proposals that were designed primarily to minimize
1/
Much of the discussion 1n this section 1s drawn from a staff
paper prepared by Paul Boltz and James 0' Brien, "Tyinq the Discount
Rate to Market Rates of Interest" (Board of Governors of the Federal Reserve
2/
System} May 1980.
The option of tying the discount rate at a constant spread above the
federal funds rate is not considered because of the difficulties-described in the preceding section--that such an approach would have
in achievinq an eouilibriurn under lagqed-reserve accounting.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 46 -
the announcement effect of discount rate actions,. by making them
automatic, recommended
use of the 3-month Treasury bill rate as
the tie--or alternatively some broader index of 1- to 3-month
money market rates.
Rates on , .. to 3-month debt maturities would
also have certain advantages as a tie for reserve-targeting purposes, since they are less influenced than the rate on 1~day
federal funds by stochastic shifts in bank needs for reserves
that are unrelated to changing money demands,
If the federal funds rate were selected as the tie,
any attempt to link the discount rate to very recent levels of
the federal funds rate could produce large, possibly explosive,
movements in both the federal funds rate and other market rates.
For example, if today's discount rate were tied to yesterday's
federal funds rate, anything causing a change in yesterday's
funds rate would lead to a further change in today's funds rate
because of the tied increase in today's discount rate.
This
would induce still further changes in tomorrow's discount and
funds rate, and so on.
If the initial change was an increase in
the funds rate, rates would continue to move higher over the
statement week and into future statement weeks.
While the public
ultimately would begin to respond to large rate increases by
reducing its demand for money, this would probably not occur
very rapidly.
Any such rapid escalation of market rates
could pro-
bably be expected to force a change in either banking practices
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 47 or Desk procedures.
Banks might begin to hold larger amounts of
interest sensitive excess reserves, or the Desk might act to
offset the destablizing effects of the contemporaneously tied
discount rate by altering its provision of nonborrowed reserves.
If the Desk did act to halt the interest rate spiral, however,
this would reflect the imposition of a federal funds rate limit.
The technical problem of induced interest rate volatility could also be damped if the discount rate were tied to some
lagged value of the federal funds rate instead of a very recent
rate.
For example, if the discount rate were a monthly average
of the funds rate, demands for reserves would begin to be
adJusted before any induced sequence of interest rate movements
proceeded very far.
As a result, interest rate movements
would be better coordinated with counteracting changes in the
public's demand for money.!/
l/
Under a contemporaneous reserve accounting system, tyinq of
the discount rate to a current level of the federal funds
rate would create less of a problem, since the effects of a
change in the level of the discount rate would be partly
absorbed by some immediate change in the public's demand
for money and this would exert an immediate influence on
banks' demands for reserves. Nevertheless, the sequence of
interest rate adJustments could still be quite substantial
if the public's demand for money were highly interest inelastic.
If, for example, there were an initial increase
in the public's demand for money balances, this would raise
the federal funds rate and, with it, the discount rate.
Rat~s would continue to rise until they became high enough
to bring the level of desired money balances and back reserve
needs back to their initial level.
In practice, the increase
in market rates required to force the public's demands for
'
money back to this initial level right away could prove to
be very large indeed.
If this were generally so, the situation would be essentially the same as for lagged reserve
accounting.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 48 The case for a tied rate.
Those favoring a discount
rate tied automatically to a market rate typically recommend
either a link to a backward looking weekly or monthly average
of the federal funds rate or a link to a more current weekly
rate average for some market security, such as the three-month
Treasury bill.
The rationale for a backward looking federal funds
rate tie is essentially twofold.
First, it will allow for
some variation of the spread of the current federal funds rate
over the discount rate and thus, by ~olerating some increase
in the volume of borrowed reserves, limit the risk of an interaction with the discount rate that ratchets the federal funds rate
upward.
This will help to minimize the possible pitfall of
linking the discount rate too tightly to a current rate series
that is heavily influenced by strictly temporary shifts in
demands for reserves and not reflective of a basic trend
demand for money.
in the
At the same time, however, a lagged tie of
this type wilJ keep spreads of market rates over the discount
rate from reaching the unacceptably large proportions that have
developed at critical points under the existing system of establishing the discount rate on a discretionary basis.
By reduc1Dg
the average size of the spread of the funds rate over the discount rate, the risk that the administrative guidelines will
cease to be effective in controlling use of the disco1rnt window
will be minimized
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 49 -
The case against a tied rate
Those opposed to an
automatic tie of the discount rate to the federal funds rate
believe that this approach would inevitably encourage too close
a linkage to relatively current adJustments in money market
conditions.
They fear that such a tie would accentuate interest
rate volatility and lead to excessive, counterproductive swings
in growth of the monetary aggregates around their targeted
longer-run paths.
Moreover, they note that although a tied
rate approach would make the average volume of borrowed reserves
more stable than it is now the short-run variance of borrowing
could be exp~cted to remain sizable
Proponents of a tied rate approach argue that an automatic trigger is the only means of assuring that timely actions
will be taken to keep the discount rate in reasonable alignment
with market rates.
They believe the existing tendency for the
Federal Reserve to delay Judgemental discount rate actions is
essentially a reflection of the heavy publicity that surrounds
them and that the only way this publicity can be defused so
timely action can be assured is to make the discount rate move
automatically in tandem with market rates.
In practice, however, decisions on discount rate actions
involve ~uch more fundamental policy considerations than
concern about publicity.
The adoption of a tied rate approach
would make automaticity override these other considerations.
Where other central banks have introduced tied rate procedures
for setting their discount rates, they have had to wrestle with
this issue of whether the obJective of tying
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
should take
- 50 -
precedence over other policy considerations.
Generally, to accom-
modate other overriding policy needs, the rules for those other
ties have had to be breached frequently.
After a period of mixed
results, the experiments have usually been abandoned.
To assure that the automaticity feature of a tied
discount rate did not produce excessive volatility in market
interest rates and money growth, it would be necessary to
I
introduce a significant backward-looking focus in any particular federal funds rate tie that was selected.
Under such an
arrangement the discount rate would still lag appreciably behind
rapid changes in the federal funds rate.
To control use of the
window the administrative gu1delines would, therefore, still need
to be retained.
For this reason, there would still be room for
much of the same type of uncertainty regarding the relationship
between expected and actual levels of borrowed reserves that
there is under the present system.
More active use of the discount rate surcharge--possibly
applied to a broader size spectrum of borrowers--could achieve
some of the obJectives sought by a discount rate tied to the
I'
federal funds rate with a lag.
The risks of excessive rate
ratcheting would be less under this Judgemental approach, since
detisions on establishing the surcharge would not have to sacrifice the key element
of discretion that 1s so important when weighing the range of
policy considerations involved 1n any discount rate action.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 51 -
Use of a security rate as the tie, in lieu of the federal
funds rate, to try to
play
down
the importance of very temporary
changes in reserve demands would introduce additional technical
complexities that might prove troublesome.
shows that in periods as short as
For example, experience
an intermeeting period, most
market rate series will at times show temporary supply-demand
distortions relative to the structure of other similar rates.
Thus, any series used as an automatic tie would have to be
reviewed regularly to determine whether any temporary market
factors were creating distortions that indicated a need to set
the automaticity aside.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 52 Establishment of a Graduated
Structure of Discount Rates
The earlier discussion of a penalty discount rate
suggests that its relative inflexibility would amount to a policy
of virtually "no accommodation" to unexpected reserve shocks,
whether the shocks came from only temporary distortions in the
supply of reserves or from more fundamental increases in underlying money demands.
At the opposite extreme the former strategy
of targeting open market operations on the federal funds rate
amounted to a policy of "full accommodation" of any demands for
reserves that tended to bid up the funds rate.
With many analysts
faulting both of these systems--the penalty rate approach for
exacerbating interest rate volatility, and "fed funds" targeting
for frequently funding excessively rapid money_supply growth-logic suggests that there may be a middle, more nearly optimum,
ground that seeks to split the difference between these other
approaches.
I
It would reJect both no accommodation and a full
accommodation of reserve shocks, in favor of a partial acrommodatio11-0f course, the administrative guidelines now being
applied by discount officers are already designed to seek such
a middle ground, when taken in combination with the reserve targeting
system.
process»which forces more or less borrowing on the banking
As banks borrow, discount officers exert pressure to
repay on the particular banks that borrow too much or too frequently in the expectation that this will encourage them to constrain their loan and investment activity.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Even banks that seek
- 53 -
to avoid the imposition of these discount window constraints may
act to constrain loan and investment activity in order to limit
their use of the window.
In effect, under this existing approach
the implicit costs of additional borrowing at the discount window
rise as the borrower approaches the limits of either his own or
the Reserve Bank's constraints on his access to borrowed reserves.
Unfortunately, the earlier analysis of actual reserve
targeting experience suggests that these existing procedures for
administering the discount window do leave something to be desired,
because week-to-week levels of borrowing activity tend to be so
loosely linked to the spread of the federal funds rate over the
discount rate.
To try to improve on this sometimes erratic per-
formance, the Federal Reserve might introduce an alternative
middle-ground approach.
This alternative would establish an
explicit schedule of discount rates that rose in steps as the
quantity of an institutions's borrowing rose.
The graduated rate
structure would be uniformly applied at all Federal Reserve
Banks and would apply to an insitution's borrowing in each
statement week
Specifics of a graduated discount rate plan.
A set of
discount rates graduated to increase as the quantity of an
institution's borrowing expanded would have several dimensions
that could be adJusted in various ways.
Specific decisions
would need to be made on (1) the level of the base or lowest
rate in the structure, (2) the number of rates in the structure,
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 54 (3) the size (or sizes) of the spreads between rate steps, (4)
the volume of borrowing to be allowed at each rate step, (5)
whether these volumes should be defined in absolute dollar terms
or as a ratio to some measure of deposits or assets, and (6)
whether the rate steps should be changed Judgementally, or tied
automatically in a fixed relationship to the federal funds rate.
While these questions are complicated, experience with
the existing system of administrative controls would provide
some guidance on what ought to be introduced as the initial
settings in the graduated rate structure]
The settings could
then be adJusted as experience developedland to the extent
specific policy obJectives were modified
I
1
For example, the
I
slope of the graduated scale, as determi~ed by the size of
borrowing tranches and the increment in interest rates between
tranches, could be changed over time to reflect more or less
accommodation of borrowing.
To avoid the forecasting uncer-
tainties that arise under the present system of Judgementally
administered guidelines, the proposed system would have no
guidelines on the appropriateness and frequency of borrowing.
Under a system of this type
institutions would pre-
sumably increase their discount window borrowing until its
marginal cost equalled the prevailing federal funds rate--i.e
the cost of alternative funds
Thus, the spread of the funds
rate over the minimum borrowing rate would be determined by the
aggregate shortfall of nonborrowed reserves from total reserve
demand.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 55 For example, suppose the amount of borrowing allowed
at each step in the graduated rate structure was $25,000 with
increments of 25 basis points between each step, and that
initially nonborrowed reserves were provided in an amount such
that the shortfall from total reserve demand, given a funds rate
of rf' was $1.1 billion.
Suppose further that the minimum
borrowing rate (MBR) was set 75 basis points below rf
If there
were no administrative guidelines on frequency and duration of
~borrowing,and,institutions could borrow even when they were -net
sellers of federal funds, then all institutions would find it
attractive to borrow each week up to the point where the marginal
discount rate was equal to the funds rate
institution would borrow roughly $75,000,!/
In this case, each
For every 25 basis
point change in the spread between the funds rate and the MBR,
borrowing would change in the same direction by about $375
million
Of course, there still would be some uncertainty about
the intraweekly behavior of the funds rate.
In Judging how much
to borrow at the discount window early in the statement week,
banks would have to have some motion of the level of the funds
rate later in the week.
Errors in such expectations would still
lead to some variance in the relation between money market rates
and borrowing levels.
!/
If this agreement had been applied to the 15,000 member
commercial banks that were eligible to borrow at the Federal
Reserve in 1979, total borrowings would have amounted to
$1 1 billion
Regression equations of actual experience
suggest, however, that this number probably ought to be
raised by $500 million to allow for frictional borrowing
that is not interest sensitive
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 56 A system that sets the tranches in absolute dollar
amounts, as in the example above, relies on the interbank federal
funds market to distribute the borrowed funds to those institutions with actual reserve needs.
The borrowing pattern that
results--with every institution borrowing a relatively small
amount from the window--is quite different from the historical
pattern for member banks, where borrowings for individual institutions range to millions of dollars
Clearly, if a large
institution suffered an unexpected deposit outflow of·several
million dollars near the end
of the day and was forced to the
window because there were few funds available in the funds
market, its marginal cost of borrowing would be quite high
especially if this occurred toward the end of the statement
week
This type of situation might also induce sizable intra-
day fluctuations in the federal funds rate, as institutions with
reserve deficits would be willing to pay a larger premium for
funds borrowed in the market to avoid the high marginal discount
rate.
An alternative approach,
needs vary with size of institution,
which recognizes that reserve
would be to make each bor-
rowing tranche a percentage of some size characteristic of the
borrower--for example, required reserves, deposits, or capital
The tranches could be set to yield roughly the same amount of
borrowing as in the illustrative example above, but a much
larger proportion of the borrowing would be done by the larger
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 57 institutions.
Given the assumption that all institutions would
borrow until their marginal discount rate equalled the funds
rate, the aggregate borrowing schedule would still be more predictable since at each funds rate expected total borrowing could
be derived by summing over the amounts each institution would
be likely to borrow at that funds rate.
If the amount of borrowing at a given spread between
the funds rate and the MBR were, say, higher than desired, then
1
borrowing could be reduced in three ways without resorting to
administrative pressure
reduce the spread between the funds
rate and the MBR, reduce the size of the borrowing tranches,
or increase the increment in the borrowing rate between each
tranche.
Existing Reserve Bank practices for administering
adJustment credit generally lead to an increase in administrative
pressure on individual borrowers, when, among other things, its
borrowing becomes more frequent.
The alternative discount window
procedure proposed here could, of course, be designed to continue
some sort of frequency guidelines, along with the graduated rate
schedule based on amount.
Such a feature introduces "progressive
pre3sure'' into the cost of discount borrowing
sporadic borrowing
due to temporary fluctuations in the supply and demand for
reserves is accommodated, with no unnecessary movements in market
I
rates, but borrowing of longer duration, which reflects more
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 58 -
permanent shifts in the need for
reserves, will bear a progres-
sively higher cost to borrowers over time and thus will be
transmitted to increases in market interest rates.
Although progressive pressure would permit the discount
window to act as an offsetting factor to underlying shifts in
reserve demand while inducing only a minimum of unwanted shortterm variability in market rates, it would also tend to weaken
the ~esk's ability to predict borrowings.
Any administrative
rule that makes the effective cost of borrowing depend on the
number of times banks borrow over a given interval creates uncertainty about the relationship between total borrowing and
money market conditions because the potential costs for each
borrower depend upon his past borrowing record and his expectations about future spreads between the discount rate and market
rates
The case for a graduated discount rate.
The princi-
pal advantage of a graduated rate approach is that it allows
the Federal Reserve to spell out more precisely how the discount
window will respond to demands for borrowed reserves instead of
trying to rely on 12 individual Reserve Bank discount officers
to provide uniform interpretations of a rather general set of
administrative guidelines
With a graduated rate arrangement,
the System can be more confident that any basic shift in demands
for reserves will produce a timely response in market interest
rates, which then feeds back appropriately on money demand.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 59 This closer relationship between borrowings and money market
conditions should make it easier for the Desk to Judge the
ultimate impact of its actions on the money stock.
While a structure of graduated discount rates like the
example described above may seem somewhat cumbersome on its
face, it could actually be expressed and explained as a relatively
simple arithmetic calculation
Moreover, because application of
the rate graduation would start afresh with each new statement
week and all administrative constraints on appropriateness and
frequency of borrowing would be eliminated, the role of the
discount window in borrowers' fund management strategies would
be simplified.
There would be no incentive to hold off borrowing
this week in order "to keep one's powder dry" for possibly
greater needs in later weeks.
A graduated discount rate procedure would also be
consistent with other possible innovative approaches to discount
policy
For example, the rate structure could be adJusted using
either discretionary actions or by tying
the discount rate
automatically to the federal funds rate.
If the structure of
discount rates were changed by discretion, the System could continue to use such changes to signal adJustments in general
monetary policy
In fact, such an approach would add new
dimensions for signaling purposes, since the Federal Reserve could
chanQe the minimum rate in the structure, the slope of the schedule, or
both.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 60 The case against a graduated discount rate.
The initial
problem with any plan that progresses the discount rate upward
with additional amounts of borrowing is that the Federal Reserve's
explicit authority to use this approach was withdrawn the Congress
in 1923.
The enabling provisions of the original Federal Reserve
Act included language stating that the "discount rate may be
graduated, ... or progressed, on the basis of the amount extended."
Immediately after the
First World War,
when heavy member bank
borrowing was thought to be exacerba~ing inflation, several
Federal Reserve Banks did elect to scale their rates upward on
the basis of amounts borrowed
In some cases the resulting
marginal charges reached levels of nearly 90 percent
Because these steep rates created a substantial backlash of complaint among bankers,
the Congress (in 1923) deleted the
wording of the act that had authorized graduation of rates by
quantity borrowed
Board lawyers are presently examining the
legislative history of subsequent discount window changes to
determine whether the authority to implement progressive discount rates might have been implic1 tly restored by the Congress
Absent such authorization, there is serious question whether the
Federal• Reserve possesses the authority to graduate discom1t, -, -,
rates by quantity without some new indication from the Congress
that it believes this approach is acceptable.
Even without this statutory impediment, other aspects
of a graduated rate proposal could create significant difficulties.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 61 For example, since the lowest discount rates in any graduated
structure would be below the federal funds rate, any institution
eligible to borrow at the Federal Reserve would be attracted by
this below-market opportunity.
Because there would be no guide-
lines requiring potential borrowers either to turn first to
alternative sources of funds or to limit their borrowing to
"appropriate" purposes, eligible institutions would be encouraged
to draw all the Federal Reserve credit they were entitled to, up
to the point where the marginal cost of such credit was roughly
equal to the cost of funds from the private market.
Information now being reported to Federal Reserve
Banks shows that roughly 20,000 depository institutions are
eligible to borrow at Federal Reserve Banks under provisions of
the Monetary Control Act.
Since it would be advantageous for
all of these to use the discount window under a graduated rate
scheme, the number of regular borrowers at the window would
mushroom.
In 1979, less than 2,000 member commercial banks
borrowed at the window during the year, and on a week-to-week
basis the number borrowing was seldom over 400, even during
periods of peak borrowing.
Since early September when the window
was opened to nonmember institutions on a regular basis, the
maximum number of such borrowers for any statement week has
not exceeded 50.
While borrowers under a graduated rate plan would no
longer have to be monitored closely to assure that their use of
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 62 -
Federal Reserve credit was appropriate, they would still need to
be monitored for creditworthiness.
Because of the number and
diversity of institutions involved, this would represent a maJor
undertaking.
Moreover, if most of the 20,000 institutions with
access to the window took full advantage of their eligibility
to borrow, the daily task of processing loans and loan collateral
would require substantially larger staffs at the Reserve Banks.
In addition, a graduated rate approach might be more
complicated than presumed.
In the initial stages of transition,
it would take considerable trial and error to determine appropriate sizes for the rate steps and borrowing tranches.
Consequently,
in this possibly protracted period of experimentation, forecasts
of expected borrowing at the window would probably be much more
erratic than they are now.
The presumption that past borrowing
patterns of member banks could serve as a useful model for predicting expected future results would probably prove to be too
sanguine.
Not only would the abandonment of all administrative
controls be likely to change the borrowing behavior of member
banks substantially but it would al so be difficult to predict how
soon and how completely other types of institutions would adapt
to their new borrowing rights
Finally, the different optional approaches that might
be followed in implementing any actual plan raise several additional questions
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
One suggested option is to tie the structure
- 63 -
of graduated rates in a fixed spread relationship,to ~he federal
funds rate, presumably so the base rate (or rates) will fluctuate
at a constant spread(s) below the funds rate.
If this approach
were followed, borrowers would be assured of continuous tranches
of credit at favorable rates.
As they perceived this to be
the case, they could be expected to begin allocating these
assured tranches of borrowing to somewhat longer-term forms of
lending.
In effect, they would be treating these tranches as
a basic borrowing privilege, which would operate in a limited
way as a partial substitute for capital.
To the extent this
proved to be the case, the approach might prove quite vulnerable
to attack as an unwarranted subsidy.
Borrowing from the unlimited tranche provided at a
penalty rate above the f/t'deral funds rate would have essentially the
same purposes as those served by a single penalty discount rate
system.
But the provision of credit in tranches Just below the
funds rate would tend to limit pressure on the penalty tranche
and thus produce less general volatility in market interest rates
than the single penalty rate system.
The alternative approach of establishing a structure
of graduated rates on a discretionary basis, without any fixed
relationship to the federal funds rate, would provide more leeway to adJust the number of rate steps, spreads within the rate
steps, and the size of the borrowing tranche at each step.
This
would add substantial flexibility to the graduated rate system
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- 64 -
as a policy instrument with subtle announcement effects.
But
it would add to its complexity and in the process probably make
forecasts of expected borrowing less predictable, at least
te~porarily.
I
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Appendix A
1../
Aggregate Borrowing and Money
Market Interest Rates
To test the linkage between borrowed reserves and the
spread of the federal funds rate over the discount rate, a
variety of econometric relationships were estimated relating
borrowing to this spread and to an assortment of other variables
that
measure various aspects of conditions in the money market.
Some of the more successful relationships are presented in
Table 1.2/
All of the results were obtained from weekly data
running from the beginning of 1970 through September 1979 ~/
As the table indicates, the spread variable alone
accounts for most of the explanatory power of the various
equations
Using a constant, the spread, and the lagged error,
somewhat more than 80 percent of the variance of member bank
borrowing can be explained.
The standard error of the regres-
sion is about $250 million or about one-third the average level
of borrowing over this period.
A similar equation to model 1
was tested in which the discount rate and the federal funds rate
were entered separately rather than subtracted one from the other.
When this was done, it was found that the coefficients on the
two rates were of opposite sign and of essentially the same
magnitude.
This result strongly suggests that it is the spread
This appendix was excerpted from Johnson and Spitzer, 11 Recent Behavior
of ~ember Bank Borrowing. 11
t-statistics for each coefficient appear in parenthesis.
The emergency borrowing by Franklin National Bank in 1974
was excluded from the data.'
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A - 2
between the two interest rates that influences borrowing and that
the absolute level of the rates has little if any detectable
influence.
Specifically, model 1 suggests that an increase in
the spread of 10 basis points will be associated on average with
an increase in member bank borrowing of $40 million.
A clear
implication of this result is that member banks historically
have responded in a systematic way to the rate advantage of
borrowing from the discount window.
Model 2 in T~ble 1 is the same as the first model
except that the spread variable is also used in the regression
in a squared form
l/
This specification improved the fit of the
model marginally and implies that successive increases in the
spread of equal amounts result in progressively smaller increases
in borrowing 2/
Some improvement in the fit also can be obtained by
adding the level of required reserves
Model 3 shows the effect
of required reserves without the spread squared variable and
model 4 includes both of these additional variables
The
estimated coefficients indicate that for every $1 billion increase in required reserves, member bank borrowing increases on
average by about $35 million
In effect, this variable scales
For values of spread less than zero, the spread squared is
'
multiplied by -1.
A somewhat implausible implication of this specification is
that increases in the spread beyond about 5 percentage points
are associated with progressively smaller levels of borrowing.
During the period over which the equation was estimated,
there was one brief period in 1974 in which the spread was
above 5 percentage points on a weekly average basis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A - 3
Table 1
REGRESSION RESULTS: BORROWING AS DEPENDENT VARIABLE 1/
(Weekly data prior to October 6)
Independent variables
and summary statistics
Model
1
2
3
4
5
Independent variables
Constant
Spread
493
(12. 3)
482
(14.0)
401
(15. 6)
(12.6)
577
-641
(-2.3)
(-LS)
392
(16.1)
567
(13.0)
570
(13.2)
-57
(-4.5)
-57
(-4.6)
.033
(4.1)
(3. 2)
-57
(-4.4)
Spread squared
. 037
(4.0)
Required reserves
-4 36
-761
(-2.4)
Change in required
reserves
.027
.034
(2.3)
Summary statistics
Standard error of
regression
254
250
251
247
246
-2
R
.836
.841
.841
.846
.848
Durbin-Watson
2. 28
2.29
2.25
2.25
2.24
.69
.65
.68
.68
,n2
Rho
1. The spread is measured in percentage points, and the reserve aggregates
are measured in millions of dollars.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A - 4
the regression and indicates that there is a weak tendency for
member bank borrowing to increase along with the level of reserves
The last version of the model shown in Table 1 includes
the change in required reserves in addition to the level
The
estimated coefficients imply that borrowing is greater in the
initial week in which there is an increase in required reserves
than in subsequent weeks
The desire to compare the borrowing relationship after
October 6 to earlier experience raises an econometric issue.
The regressions reported above were obtained from models that
implicitly assumed that the spread between money market rates
was determined by the Desk's actions and that the level of
borrowing was endogenously determined
In other words, the
System's actions fixed the relative price of borrowed and nonborrowed reserves, and banks were free to choose, within the
limits imposed by administrative practices, how much they would
borrow at the discount window
been reversed
Since October 6, the situation has
The System continues to fix the price of reserves
borrowed at the discount window but not the federal funds rate,
the price of nonborrowed reserves.
Instead the aggregate level
of borrowing is determined (ignoring the weekly variation in
excess reserves) by the System's adherence to nonborrowed
reserve targets
Under lagged reserve accounting, required
reserves less nonborrowed reserves necessarily equals borrowing
at the discount window (again ignoring excess reserves)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
With
A - 5
aggregate borrowing predetermined, the funds rate must adJust to
clear the market for nonborrowed reserves.
Thus, it is appro-
priate subsequent to October 6 to treat borrowing as exogenously
determined and the spread as endogenous.
The description above oversimplifies the differences
in behavior before and after October 6.
The System still main-
tians some limits on the range over which the funds rate is
allowed to fluctuate.
These limits sometimes have caused the
Desk to alter its nonborrowed reserve targets
With the level
of borrowing sometimes affected by the behavior of the funds
rate, the direction of causality is not entirely one way under
current operating procedures
Still, as was discussed earlier,
the actual level of borrowing in most weeks has been fairly
close to the level originally envisioned when setting the weekly
targets,
thus it seems that treating borrowing as exogenous
is generally appropriate.
It should be kept in mind that the same general demand
curve for borrowing is being estimated using data before and
after October.
1 above
Equation 1 below corresponds to model 3 in Table
One can transform this model by solving for borrowing
and obtain equation 2.
This is the sort of model that should
be estimated using data subsequent to October 6
Moreover, the
coefficients obtained can be related to the results obtained
from the era when the funds rate was being pegged.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A - 6
(1)
Borrowing= a+ b.
(Spread)+ c.
(Required reserves)
+ u
(2)
Sprea d
1
b
=
a+ b1
-b
Borrovnng - bC
(Required reserves)
u
Table 2 below reports results for a few models of borrowing behavior estimated with data from the first 33 weeks of
experJ.ence with reserves targeting.
The special borrowing of
First Pennsylvania Bank was excluded from total borrowing,
because such borrowing does not reflect funds rate pressures and
was treated by the Desk as beLng in the nature of nonborrowed
reserves
In addition to the same type of variables used in the
earlier models, a dummy variable set equal to one in the weeks
that the 3 percent surcharge was in effect also was included.
The regression results presented in Table 2 for the
first 33 weeks of reserve targeting indicate that borrooing does
help explain the spread but that required reserves do not.
It
should not be particularly surprising that required reserves
were not significant, since they were included as a scale variable
and there was little trend evident in this variable during the
relatively short interval over which this equation was estimated
The surcharge dummy also was significant and indicates that the
spread was about 2 percentage points higher in the weeks in
which it was in effect.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A - 7
Table 2
REGRESSION RESULTS:
SPREAD AS DEPENDENT VARIABLE 1/
(Weekly data subsequent to October 6)
Independent variables
and summary statistics
-
Model
1
2
3
4
,-0. 94
(-0.4)
-9.21
(-0.8)
-1.19
(-0.6)
-7.25
(-0.7)
.00084
(2.8)
. 00080
(2.6)
.00087
(3. 3)
.00084
(3.1)
Independent variables
Constant
Borrowing
Required reserves
.00013
(0.6)
.00019
(0.3)
Surcharge dummy
1.95
(3.1)
1. 91
(3.0)
Summary statistics
Standard error of
regression
. 984
. 991
.870
.880
-2
R
.795
. 792
.840
.836
Durbin-Watson
1.20
1.20
1.52
1.49
.93
.94
• 92
.93
Rho
1. The spread 1s measured 1n percentage points, and the reserve aggregates
are measured 1n millions of dollars.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A - 8
If the structure of this model had not changed subsequent to October 6, then the reciprocal of the coeff1c1ent on
the spread 1n model 3 {Table l) should be roughly
coeff 1c1ent on borrowing 1n model 4 {Table 2).
does not appear to hold, however.
equal to the
This rel a t1onsh1p
At-test 1nd1cates that the
coeff1c1ent on.borrowing 1s s1gn1f1cantly smaller than the
reciprocal of the estimated coeff1c1ent on the spread.
Spec1-
f1cally, model 4 of Table 2 1mpl1es that an increase 1n the
spread of 10 basis points was assoc1zted on average with an
increase of borrowing of $110 million as compared to the $40
million increase implied by the equations fitted with data before
October 6.
It thus would appear that for a given spread, banks
on average were more willing to borrow from the discount window
subsequent to October 6.
While 1t can be concluded with a high
level of confidence that banks have been more willing to borrow
recently, it should be noted that the point estimate of the
coefficient on borrowing obtained from recent data has a substantial variance and could be quite different from the "true"
value of this parameter.
It also appears that the borrowing relationship has
become more variable since the advent of reserve targeting
It
can be shown that the reciprocal of the coefficient on borrowing
1n model 4 {Table 2) multiplied by the estimated standard error
of that model should be roughly comparable to the standard
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A- 9
error of model 3 (Table 1)
assuming that the underlying vari-
ability of the relationship had not changed.
appear to be the case.
This does not
An F-test indicates that the standard
error of model 4 (Table 2) is significantly greater than the
estimate of the standard error obtained from data prior to
October 6.
Specifically, the point estimates indicate that the
standard error of the borrowing relationship was at least 35
percent greater after October 6 than before.
In sum, it appears
that the relationship between borrowing and money market rates
spreads has changed and has become more variable suhsequent to
October 6.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
IMPLEMENTING THE NEW OPERATING PROCEDURES
THE VIEW FROM THE TRADING DESK
January
1981
Paper Written for a Federal Reserve
Staff Review of Monetary Control
Procedures
by
Fred J. Levin and Paul Meek
January 21, 1981
IMPLEMENTING THE NEW OPERATING PROCEDURES:
THE VIEW FROM THE TRADING DESK
CONTENTS
Introduction and Summary
I.
II.
III.
Open Market Operations Under
the New Reserve Approach
A. Day-to-Dav Operations
B. Frequency and Timing of Desk
Operations
Experience
in Hitting Reserve
c.
Objectives
The Market's Perception of
Monetary Policy
A. Operational or Institutional Changes
IV.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Appendix: Technical Adjustments to
the Reserve Paths
PAGE
1
7
10
17,
19
26
34
A-1
Implementing the New Operating Procedures:
The View from the Trading Desk*
Introduction and Summary
Open market operations have functioned reasonably
well since October 6, 1979, under the new supply-oriented
reserve strategy.
The strategy itself entails a procedure
for translating the FOMC's desired growth rates for the monetary aggregates into intermeeting Desk obJectives for nonborrowed reserves (NBR).
The procedures allow for weekly adjust-
ment for changes in the relation between required reserves and
the appropriate money supply measures, and weekly decisions on
whether to adjust the NBR path to speed up the correction of
undesired money growth.
Over all. the Desk views the approach
as a significant improvement in generating the kinds of portfolio decisions and interest rate changes needed to resist
sustained and potentially destabilizing movements in the
monetary aggregates.
The present memorandum analyzes how the Desk has
implemented the new procedures since their inception, highlighting the operational changes that have resulted from the
modification in strategy.
It also examines how market parti-
cipants have tracked monetary policy under the new procedures
and whether operational or institutional changes could reduce
the variability of interest rate response without compromising
*Fred J. Levin and Paul Meek were primarily
responsible for the preparation of this study. Robert Van Wicklen
provided able research support.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
2
the new strategy.
(An appendix discusses some technical
issues involving intermeeting adjustments to the reserve
paths.)
(1)
A summary of the principal findings follows:
The new strategy shifted the focus of weekly Desk
operations from the federal funds rate to nonborrowed
reserves.
The Desk depended more heavily on staff NBR
estimates in determining the amount and timing of operations.
Daily market operations were concentrated
principally b~tween
11:30 a.m. and 12:15 p.m. in order
I
I
to emphasize that they were addressed to managing
I
reserves and were largely independent of the existing
I
I
I
level of the 1ieder,al funds rate.
(2)
• Reserve forecast errors continued to be quite
large--with the miss in the weekly average of market
factors equal to about $675 million on the first day
of the statement week and $145 million on the last day
of the statement week.
The variability of Federal Re-
serve float continued the chief source of forecast
error.
(3)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
As a supplement to the reserve projections, the
Desk found the federal funds rate a somewhat useful,
although not consistently reliable, indicator _of- - , "
reserve availability.
It did point in the right direc-
tion about two-thirds of the time when sugqestinq a miss in
the reserve proJections; for errors of $500 million
or more it was accurate on
seven of nine occasions.
3
(4)
The number of Desk operations to supply or absorb
reserves in the market through repurchase agreements and
matched sale-purchase transactions were about one-third
less than in the previous year.
Such temporary actions
were notably less frequent before the weekend than had
been true earlier.
(5)
The absolute deviation in nonborrowed reserves from
path averaged $184 million over fifteen separate reserve
periods.
There was a tendency for nonborrowed reserves
to come in below path in contrast to total reserve~ which
ran above path.
The average shortfall in nonborrowed
reserves was $130 million.
(6)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
To a considerable extent, deviations from the nonborrowed reserve path were tolerated, after consultation
with the Board staff and Chairman, whenever it was potentially misleading to market participants to seek
sharp changes in member bank borrowing in the final
week of a reserve period from what had recently prevailed or was likely to result from new FOMC instructions.
Deviations also arose when borrowing ran sub-
stantially above or below expectations built into the
reserve paths during the early part of the final week.
Confronted with the alternative. of sharp declines or
increases in borrowing and interest rates after the
week.end, the Desk sometimes accepted a miss in the nonborrowed reserve path.
In the broader context, this
4
smoothing process has not interferred with the shifts in
I ,
interest rate levels resulting from the changing impact
of the demand for reserves relative to the nonborrowed
reserves path.
(7)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
The perceptions and actions of market participants
have been significantly affected by the variability of
interest rates associated with the supply-oriented
strategy.
Market participants continue to use the
fecPral funds rate as t~e pri~~ry indicator of policy's
current thrust, despite the Desk's focus on reserve
considerations.
An analysis for the past 14 months of
the relation between borrowing and the spread between
the fed~ral funds rate and the discount rate suoaests
that the relationship is considerably more variable than
in 1972-74.
In addition, changes in borrowing appear to
have less impact on the rate spread in the recent period
than earlier.
The supply-oriented strategy produced a
rapid increase in borrowing from a low level between midAugust and mid-September 1980, but the impact on the
federal funds rate was
MO(h.5t.
In such a situation.
when banks have had little recourse to the window for
some months, the System could perhaps be somewhat more
aggressive in lowering the NBR path to speed up achievement of the FOMC's monetary obJectives.
However, in 1980,
subsequent increases in the discount rate and restoration
of a rate surcharge constituted effective actions in the
same direction.
5
(8)
The variability of the relationship between borrowing and the rate spread adds to the market's difficulty
in tracking the Federal Reserve's current policy posture.
This variability contributes to the short-run volatility
of interest rates, which,in turn, may well add to underwriting costs in the money and capital markets.
However,
the market's overreactions tend most of the time to reinforce the System's efforts to attain its objectives.
(9)
If
ool icymakers
consider it desirable to reduce the
noise content of the market's perceptions of System policy,
one might reduce the variability of the rate response to
borrowing changes by modifications of (a) the discount
window or (b) the routine conduct of open market operations.
Whether one considers this desirable depends upon
where one strikes the balance between reducing interest
rate volatility and insuring that short-term interest
rates are free to respond to changes in both reserve
pressures and market expectations of further change.
(10)
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Two issues involving procedures for adJusting the
reserve paths between Committee meetings to reflect
changes in the money-reserve relationship deserve further
attention.
First, there is the question of whether the
current practice of typically making only partial adjustments to the paths each week for technical factors should
be cont~nued or whether all potential adJustments should
be incorporated into the paths (even though this would
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
6
mean revising the adjustments weekly on the basis of
new information).
Second, there is the question of
how much policy weight should be given between meetings
to the broad versus the narrow money measures.
Currently,
because of the structure of legal reserve requirements,
the primary emphasis is on the narrow aggregates.
Moreover, their relative weight will increase further
as the provisions of the Monetary Control Act are
phased in.
The Committee may want to consider whether
path procedures should be modified so as to increase the
weight given to M-2 in operations between meetings.
7
Open Market Operations Under the New Reserve Approach
Since October 6, 1979,the Desk has worked between
Committee meetings with paths for total and nonborrowed reserves
designed to be consistent with the Committee's short-run objectives
for growth of the monetary aggregates.
Operationally, the major
focus has been on hitting the nonborrowed reserves path, the
reserve measure which is reasonably subject to short-run Desk
control.
The Desk's primary concern has been to achieve average
path levels for nonborrowed reserves over blocks of weeks
encompassing either the full intermeeting period or two separate
subperiods when the meetings are relatively far apart.
The Desk begins each intermeeting period with a path
for nonborrowed reserves (the total reserve path estimated by
the Board staff less the Committee's initial assumption for
borrowing at the discount window).
Each week, as new information
becomes available, senior Board staff and the Account Management
review, und revise,if appropriate, the reserve paths to maintain
their consistency with the Committee's aggregate objectives.
Then the Desk must translate the reserve paths into weekly
operating obJectives for nonborrowed reserves.
in the following way:
This is done
First, the staff proJects the demand
for total reserves--that is, required reserves based on actual
or estimated deposits plus excess reserves.
Second, the
average projected demand for total reserves over the period
is compared to the average nonborrowed reserve path over
the period.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
This, given actual levels of borrowing in earlier
8
weeks, provides an estimate of average borrowing over the
remaining weeks if the average nQnborrowed reserve ~ath is
to be achieved.
Finally, this steady level of borrowing is
subtracted from the projected demand for total reserves in
each of the remaining weeks to give a series of weekly nonborrowed reserve objectives.
Using this averaging approach, the procedure tends
to smooth weekly changes in money market conditions but not at
the expense of delaying the interest rate response to a monetary
overshoot or a shortfall.
Indeed, the approach tends to speed
up the response by about one week from what it might otherwise
be under lagged reserve accounting.
For example, if deposit
growth exceeds path in the first week of a reserve period,the
Desk's procedure calls for an estimated higher level of borrowing
in the second week to be sustained over the remainder of the
period.
Otherwise, given lagged reserve accounting, the rise
in borrowing would not be expected to occur until week three,
when the demand for required reserves was boosted
and the
resulting increase in borrowing would be sharper.
As the Desk works to keep the supply of nonborrowed
reserves in line with its average path level, money market rates
tend to adjust automatically ,.whenever.growth of the monetary
...
- " .,.~
~
aggregates deviates from the Committee's objectives.
For example,
in the spring of 1980, when monetary growth fell below these
objectives, the demand for total reserves (required reserves
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
9
plus banks' normal desired holdings of excess reserves) exceeded
the nonborrowed reserve path by less than the initial borrowing
assumption.
With the Desk supplying nonborrowed reserves in
line with its path, money market rates eased dramatically as
banks found reserves plentiful and thus were encouraged to cut
back on their borrowing from the discount window.
More recently,
in fall 1980, as monetary growth ran above the Committee's
objectives and, hence, the demand for total reserves exceeded
the nonborrowed reserve path by more than the initial borrowing
assumption, money market rates rose sharply as banks were forced
to step up their borrowing from the discount window in order
to meet their reserve requirements.
In each case, the resulting
changes in money market rates tenoed to encourage banks and
the public to alter their portfolio behavior in a way that
worked in time to bring monetary growth back in line with the
Committee's objectives.
At times, as seemed appropriate, the senior Board
staff and Account Manager, in consultation with the Chairman,
accelerated the adJustrnent process by changing the nonborrowed
reserve path relative to the total reserve path.
The nonborrowed
reserve path was lowered when the demand for total reserves was
running significantly above path and raised when the demand
for total reserves was running significantly below path.
This
tended to encourage even sharper changes in borrowing and money
market rates, and thus tended to promote a speedier return to
desired growth rates for the monetary aggregates.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Increases
10
in the discount rate also worked to hasten the adjustment
process.
Normallyv the System does not respond to a discount
rate change by altering the nonborrowed reserve path.
result, banks have bid up the federal
As a
funds rate to maintain
roughly the same spread previously prevailing between the two
rates.
Day-to-Day Operations
In facing its operational task each day, the Desk has
before it an objective for average nonborrowed reserves to be
achieved in that week and in subsequent weeks.
It also has
projections, prepared independently by the Board and New York
staffs, of the supply of nonborrowed reserves for several weeks
ahead, assuming no further actions are undertaken by the Desk.
These projections reflect estimates of technical factors such
as Federal Reserve float, currency in circulation, and Treasury
balances held at the Reserve Banks.
(The proJections normally
assume that the pool of foreign account short-term investment
funds will be arranged internally with the System, although
the Desk can choose to execute all or a portion of these orders
in the market as one way of supplying reserves.)
A comparison of the nonborrowed reserves objective
with the projected supply of nonborrowed reserves leads to an
indication of the amount of reserves that needs to be added or
withdrawn to meet the objective for the week.
These comparisons
are the primary determinant of open market operations under the
current approach.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
However, a number of factors make the Desk's
11
t1sk considerably more complicated than simply p~oviding or
absorbing reserves according to the difference between the
projected reserve supply and the reserve objective.
For one thing, the level of borrowing within a
statement week can, on occasion, deviate significantly from
what was expected in the construction of the nonborrowed reserve
path.
For example, if borrowing runs far enough above expecta-
tions during the early days of a week, mathematically there
might be no way of attaining the average borrowing level implied
by the nonborrowed reserve objective for the week as a whole.
The Desk is then faced with the alternative of adhering
strictly to its nonborrowed reserve objective and, hence,
allowing huge excess reserves and an overshoot in total reserves
or else absorbing some of the excess reserves and coming out
below its objective for nonborrowed reserves.
On the opposite
side, if borrowing runs well below expectations during the
early part of a week, the Desk is faced wi~h the alternative
of forcing huge borrowing on the final day or else accommodating
some of the shortfall and coming out above its obJective for
nonborrowed reserves.
The problem with forcing a big bulge
in borrowing, or encouraging overly abundant reserves-, on the
final day of a week is the risk that banks may overreact in
managing their reserve positions in the following statement
week.
An unduly tight Wednesday can carry over to sharply
higher money market rates and greater-than-desired borrowing
levels in the next week, while an unduly easy Wednesday
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
12
conversely may tend to result in sharply lower money market
rates and under borrowing in the next week.
The Desk sought to avoid these difficulties by
directing the intraweekly pattern of open market operations
in a way that would encourage the amount of borrowing desired
over the week.
For example, if borrowing were starting out
the week on the high side of expectations, the Desk might add
reserves before the weekend in order to ease market pressures
and reduce borrowing~
even though there was no projected
reserve need for the week as a whole.
Then,having provided
reserves early in the week, it would absorb the resulting
excess later on. Despite the Desk's efforts, however, there
.
were many weeks when borrowing deviated significantly from
expectations, and the Desk chose to accept a miss on its
weekly nonborrowed reserve objective rather than provide
overabundant total reserves or force a huge end-of-week rise
in borrowing,
(For a further discussion, see below.)
Another operational problem is that the reserve
projections themselves are subJect to a wide margin of error.
For example, on the first day of the statement week the average
absolute projection error of weekly average nonborrowed reserves
(apart from System actions to affect reserves) over the October
1979-80 period amounted to about $675 million, or about l 1/2
percent of nonborrowed reserves (see Table 1).
The accuracy
of the projections naturally improves over the week as data
become available each day on the actual supply of reserves on
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Table 1
FRBNY Resei:ve Projection Errors
by Major Corrponent
for Selected Years*
{Weekly average, millions of dollars)
Nonborrowed
One week ahead
(Thursday) forecasts
reserves
~rrket fac!ors)
8
A
Addendum:
CUrrency
Treasury
Float
deposits
I'll I
in
circulation
lei
Other
factors
,~r
!Al
628
31
597
IA! l~I - IAI
iel
i~I
lel
873
409 1,681
152
453
158
required
reserves
1977-78
711
1,867
592
1978-79
887
1,435
861 1,189
334
692
140
519
221
776
65
707
1979-80
673
1,057
619
350
535
170
551
176
646
155
724
918
I-'
N
Pl
One day ahead
(Wednes~~}- forecasts
41 1,681
52
453
55
628
17
597
182 1,189
49
692
54
519
73
776
35
707
140
37
535
30
551
73
646
48
724
1977-78
118'
1,867
110
1978-79
176
1,435
1979-80
144
1,057
873
918
* From third staterrent week in October to second statenent week in October of following year.
jel = mean
absolute forecast error.
l~l = nean absolute
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
change.
13
the previous day.
However, even on the last day of the week,
the average absolute projection error of weekly average nonborrowed reserves came to about $145 million over the period
and last day errors of $500 million or more for the weekly
average were not unknown.
The major factor accounting for reserve projection
errors continues to be misestimates of float.
Indeed, the
forecast errors of float amount to almost as much as the
errors of all other technical factors combined.
A number of
studies have been undertaken over the years to improve the
accuracy of the float projections.
The most promising approach,
however, would seem to be through institutional changes in the
check collection system and Federal Reserve procedures that
would reduce its variability.
When large projection errors occur,they can hamper
the Desk's efforts to achieve its nonborrowed reserve objective
and mislead market participants on the System's policy stance.
For exampl~ an overestimate of reserve availability, which
leads the Desk to supply less than the reserves needed to
meet the nonborrowed reserve objective, can leave banks scurrying
to the discount window to meet their reserve requirements,
accompanied by a sharp rise in the federal
funds rate at the
end of the week as banks seek out other sources of reserves
as well.
The market could interpret this as a tightening of
System policy, and other short-terro rates would then adjust
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
l
upward accordingly.
14
The result might be that banks would
continue to borrow heavily in the following week, and it
might take a week (or several weeks if there is a string
of projection errors in the same direction) until borrowing
could be brought back in line with the level assumed in
constructing the nonborrowed reserve path.
Because the projections are subject to a high degree
of error, the Desk also looks at other indicators of reserve
availability, such as conditions in the federal
funds market
and the volume of dealer offerings at various rates when it
solicits propositions for repurchase and matched sale-pur9hase
transactdons.
Given the level of the discount rate and a rough
notion of the demand
function for borrowing, the Desk has some
idea of the range within which the funds rate could be expected
to trade consistent with the mixture of nonborrowed and borrowed
reserves that it is seeking.
If actual trading levels for funds
deviate significantly from expectations,this can be an indication
of a miss in the reserve projections.
The behavior of the funds market, however, is also
not a wholly reliable indicator of reserve availability.
buying or selling federal
In
funds, banks themselves are working
with projections of their own positions that
uncertain and changeable throughout the week.
are highly
The funds rate,
therefore, can move contrary to actual reserve supplies in
the banking system if only a few large banks misjudge their
positions.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
Moreover, the money center banks,in particular,
15
try to anticipate what actions other participants might take
that would influence the funds rate, including actions by
the Desk.
They are also conscious that their own actions,
or inactions, may affect the funds market significantly.
Thus,
banks tend to hold back in bidding for funds if they expect
that the Desk will soon be entering the market to add reserves,
or to sell funds if they believe the Desk is likely to be in
the market to drain reserves.
The observation that the Desk
has a normal time period in the day in which most of its
transactions are initiated leads banks to be especially likely
to defer action during that time.
Ironically, the more the
Desk confines action to a specified time of day to divorce
operations from the federal
funds rate, the less reliable the
rate becomes as a reserve indicator because of the tendency
for banks to await Desk action.
For these re,.asons, the Desk
is cautious in using the-funds rate as a reserve indicator.
The data in tables 2 and 3 attempt to measure how
useful the funds rate has been as a check on reserve projections
since the inception of the October 6 program.
They show the
number of days in which the Desk either took or deferred
actions that would have been called for by the New York reserve
projections because federal
funds were trading at rates out
of line with expectations,-~hus suggesting a miss in the
'
reserve
projections.
For each of these days, they measure
the error in reserve projections as reflected by the difference
between projected average free reserves for the week and actual
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
15a
Table 2
Federal Funds Rate as Indicator
of Reserve Supplies,
Surmiary
October 8, 19791 throuQ!i December 17, 1980
Days when f edera 1 funds trading
influenced open market operations
by suggesting error in FRBNY
reserve projections
Total
period
Nuni:ler of days
31
Federal funds rate:
Useful indicator
Fa 1se indicator
Subperiod when projection
errors of weekly average
:reserves exceeded $500 million
21
10
9
FedP"cll funds rate:
Useful indicator
False indicator
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
7
2
15b
Table 3
Federal Funds Rate as Indicator
of Reserve Supplies
October 8, 1979 through Decerrber 17, 1980
Dates when f edera 1 funds
trading influenced open
narket operations by
suggesting error in FRBNY
reserve projections
Funds rate suggested that
reserve projections:
overstate actual supplies+
llllderstate actual supplies -
Reserve projection error:
Projected less actual free
reserves, excluding open
market operations
(weekly average, millions of dollars)
1979
Wednesday, October 17
Ivbnday, October 29
Tuesday, October 30
Wednesday, October 31
Friday, November 23
Monday, November 26
Tuesday, November 27
Wednesday, November 28
Tuesday, December 4
Tuesday, December 11
+
+
+
+
+
-94
+1,173
+444
+381
-1,993
-745
-637
-311
-616
+279
1980
Wednesday, January 9
Monday, March 17
M:Jnday, April 7
Tuesday, April 8
Friday, April 11
Wednesday, May 14
Wednesday, June 18
M:mday, July 7
Thursday, July 31
Friday, August 1
Monday, August 4
Tuesday, August 26
Wednesday, August 27
Thursday, September 11
Friday, Septeni:)er 19
Friday, October 3
Thursday, November 13
Wednesday, November 19
Thursday, November 20
TUesday, November 25
Monday, December 1
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
+
+
+
+
+
+
+
+
+
+38
+38
-+46
-81
+810
+56
+86
-126
·+10
, -239
-209
-363
+
+
+
+
-138
+429
-35
+693
+1,781
+319
+2,505
+260
+43
16
free reserves (excluding any subsequent open market operations) . 11
Of the 31 days in which Desk action was affected by the funds
rate relative to expectations, there were21 days in which it
provided useful information on reserve supplies--that is, when
the funds rate was relatively high, projections were in fact
overstating actual reserves, and when the funds rate was
relatively low, projections were in fact understating actual
reserves.
The funds rate was a more accurate barometer on days
when projection errors turned out to be sizable.
For errors
of $500 millio~ or more the funds rate was accurate on seven
1
out of nine days.-
Aside from using the f~deral
funds rate as an indicator
of reserve supplies, the Desk also monitors the rate to insure
that it stays within the limits of the Committee's broad range.
1/ Free reserves was chosen as the reserve measure
in order to pick up misses in estimates of required reserves.
While these were generally small, as one would expect given
lagged reserve accounting, there were a few sizable errors
that reflected misestimates of reserve ratios that the Desk
would have accommodated in full if correct figures were known.
Using nonborrowed reserves rather than free reserves as the
reserve measure would not significantly alter the general
conclusions reached in the text.
2/ It is interesting to note that the funds rate
was a better indicator of reserve supplies in the early
months of the new reserve approach when the Desk was first
establishing the practice of largely confining Desk entries
to a regular time period during the day and the market was
unsure of the Desk's new strategy.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
17
The Desk looks at these limits more as defining acceptable
weekly average federal
funds rate trading rather than limits
for individual days or points of time within days.
Since
the ranges have been fairly wide, varying between 4 and 8 1/2
percentage points, there have,in fact,been only a few occasions
when the funds rate actually breached or threatened to breach
the limits.
On each occasion, tne effect on open market opera-
tions was only temporary as the rate moved back within the
range or the Committee adjusted the range.
Frequency and Timing of Desk Operations
With the Desk no longer aiming for a particular
federal
funds rate objective, the new reserve approach seems
to have had a significant impact on the frequency of System
operations in the market.
The number of Desk interventions
to arrange repurchase and matched sale-purchase transactions
over the October 1979-0ctober 1980 period, for example, was
down about one-third from the level of the previous year (table 4).
Declines were registered each quarter and were not concentrated
~
in any particular part of the year.
Judging from a four-month
sample ( table 5) the sharpest drops were in the number of
relatively small transactions (defined as less than $1 billion).
Under the federal
funds rate strategy such transactions were
often used to signal the System's funds rate intentions even
though projections showed no need for action.
Under current
procedures small transactions are usually arranged when
the reserve need itself is small.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
While other factors
17a
Table 4
NUmber of Desk InteJ:ventions in Market to Add or Drain
Reserves on a Tenporary Basis
1977-78
1978-79
1979-80
51
41
Jan.-March
39
60
60
38
April-Jtme
68
60
July-sept.J/
74
90
54
Total period
241
261
3.75
8
30
9
.Additional rounds
63
56
15
Preannounced
for Thursday
15
7
outright transactions
20,
19
18
Foreign RP.s in market.
32
47
54
Period
ocP.
-Dec.
42
Mano items:
Extemed rounds
Y.
Begins with thim statement week in OCtaber.
21 Ends with second statement week in October.
Notes
1. IncludEsonly RP and MSP transactions in market (both for foreign customers
and System Account). Outriqht transactions for System Account in market
shown as memo item.
2.
Exclude;pmannounced transactions far Thursday. These shown separately as
IlB10 item..
3.
COUn1Sextended murds as one narket entr:y. Number of extended rounds shown
as meno item.
4. COOn1s sinultaneous announ.cenent of transactiai with various maturities
as one market entry~
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
17b
Table 5
Number of Desk Interventions in Markett
to Add or Drain Resexves on Tenporary Basis:
by Day of Week fQlC' Varioos AlJIOtmts, of ReserVes Jj
______(JUn_e ~.E,@_§e~ 1979-SOL
$1 billion or less
Day
More than $1 billion
1979
1980
1979
1980
Thursday
4
1
13
6
Friday
7
2
16
11
MJnday
7
1
13
11
Tuesday
6
2
14
12
Wednesraly
4
3
14
11
29
9
72
51
TotD1
Table 6
Nmt)eJ: of Desk Interventiais in Market
to Add or Drain Resei:ves on ~ a z y Basis:
by Time of Day ]J
(Jlme 1::hrough S e ~ 19 7 9-0 0)
Time
1979
1980
Before 11,: 30
34
3
11:30 to 12:15
36
49
After 12:15
Total
31
lof
8
60
1/ Pericxls oover fran third statement week in October to second etate:ri.e.nt
week in October of following year.
See notes to table 3.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
18
can influence the frequency of Desk operations (for example,
the variability of reserve projections from day to day) it
seems likely, given the extent and persistence of the decline,
that the adoption of the new procedures is the dominant
explanation.
The new procedures also seem to have had a significant
impact on the timing of operations, both in terms of the day
of week and the time of the day.
Tbe Desk is now much less
likely to intervene in the market before the weekend, especially
on Thursdays when reserve projections for the week are the most
uncertain and the Desk has to rely on a tentative estimate of
the nonborrowed reserve objective made the previous week.
A
practice has also developed of initiating most temporary
transactions (repurchase agreements or matched sale-purchase
transactions) between 11:30 a.m. and 12:15 p.m., following
the regular morning conference call,which reviews reserve
projections and market developments.
The Desk has tended to
operate chiefly within that period to reduce the significance
that the market attaches to the federal
at the time of market entry.
funds rate prevailing
While some transactions still
occur outside of this period for various reasons, they are
relatively few ( table 6 ) . 3/ In contrast, under the federal
3/ An early entry to arrange repurchase agreements,
for example, may occur wnea the Desk anticipates a market shortage
of collateral, while a late entry (to arrange repurchase agreements
or matched sale-purchase transactions) may reflect a revision in
reserve projections. Early and late entries may also occur when
the federal funds rate is threatening to breech the limits of the
Committee's broad range.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
19
funds rate strategy the Desk was prepared to intervene in
the market well beforell:00 a.m.
or after 1:00 p.m. if funds
deviated by 1/8 to 1/4 percentage point or so in either
direction from its objective.
Experience in Hitting Reserve Objectives
Table 7 attempts to summarize the System's record in
achieving its path objectives for nonborrowed and total reserves.
Specifically, the table shows average deviations from paths
and the factors underlying these deviations, for the 15
reserve periods running from the four weeks ended October, 31,
1979, to the four weeks ended November 19, 1980.
The reserve
periods, which vary in length from three to five weeks,
generally correspond to intermeeting periods except, as noted
earlier, that intermeeting periods were split into two
reserve periods when meetings were relatively far apart.
,
Devi-
ations ,are measured from adjusted (rather than original) path
values since operationally these were the objectives that
the Desk sought to hit.
For nonborrowed reserves, deviations from average
path levels ranged from a low of $13 million to a high of $670
million over the fifteen separate reserve periods.
Ignor~ng
~
sign, they averaged $184 million per reserve period,
or about 0.4percent of the average level of nonborrowed reserves
( table 7).
As explained more fully below, nonborrowed
reserves tended to come in below path values in contraEt to the
experience for total reserves which ran above path values.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
On
19a
Table 7
Comparison of Actual Reserves to Path:
Four Weeks Ended October 31, 1979,to Four Weeks Ended November 19, 1980
Average
deviation per
reserve period
(million of dollars)
Average absolute
deviation per
reserve period
(million of dollars)
Average absolute
deviation as percent of reserve
measure
(percent)
Deviation from nonborrowed
reserve path1
Total
-129.7
Accepted or intentional·
Transition between
reserve periods
-39.3
Weekly deviation of borrowing -24.o
Special borrowing
-8.4
Monetary aggregates growth
-19.9
Federal funds rate
constraint
-1.2
Unintentional:
Dealer propositions
+9.5
Projection errors2
-46.4
57.0
50.7
.14
19.9
.12
.02
.05
1.2
.00
11.2
56.5
.03
8.4
.14
Deviation from total
reserve pathl
+81,3
325,0 3
Required
Excess
+122.2
-40.9
335.8
67.3
.78
Nonborrowed
Borrowed
-129. 7
183.9
+210,9
404.9
.43
.94
Total
.75
.16
1 computed from final data and adjusted paths.
2Calculated as residual.
3rndividual components do not sum to total because of interaction of components.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
3
20
average, actual nonborrowed reserves fell below path by $130
million per reserve period. 41
In accounting for deviations between actual and path
values for nonborrowed reserves, it is useful to distinguish
between accepted or "intentional" misses and unintentional misses.
Accepted or intentional misses, which accounted for over twothirds of the deviations, represented decisions to tolerate or even
aim for reserve supplies either above or below average path values.
They arose from a variety of considerations, but mainly reflected
deviations from expectations for borrowing in the final week of a
reserve period and a desire to maintain continuity in the degree of
adjustment pressure on the banks in the transition from one
control period to the next around the time of FOMC meetings.
Unintentional misses, which accounted for less than a third
of the deviations, resulted primarily from reserve projection
errors and, to a lesser extent, from the inability of the
Desk to arrange the volume of open market operations planned
because of insufficient dealer propositions. 5/
~/ Using preliminary estimates available just after
reserve periods ended shows nonborrowed reserves about $109
million, on average, below path. Final figures for reserves
tend to be lower than preliminary estimates largely because
of negative "as of adjustments."
5/ A number of the sources of reserve misses interact.
and thus a-fully satisfactory separation is not possible. The
various categories shown in the table and discussed below
are only meant to be suggestive.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
21
Of the intentional misses, the largest arose from
System efforts to deal with cumulative deviations in nonborrowed
reserves from desired levels as the FOMC meeting approached .
.
During several periods cumulative misses in weekly nonborrowed
reserve objectives would have required large changes in
borrowing in the final week had the Desk sought to achieve the
average nonborrowed reserve path for the period as a whole.
• For example, the implied borrowing level in the February 6, 1980,
week would have had to drop to zero, with excess reserves of
$900 million, to meet the average nonborrowed reserve path
for the three-week period.
In previous weeks borrowing had
run about $1 billion to $1.8 billion; then at the February meeting
the Committee chose a borrowing level of $1 1/4 billion for the
start of the next period.
and the federal
Producing a sharp drop in borrowing
funds rate, in this instance, would have
confused market participants as to the System's restrictive
posture.
It would also have made it less likely that banks
would have borrowed in accordance with the reserve objectives
for the next period.
Under such circumstance~ the Desk, after
consultation with the senior Board staff and the Chairman,
chose to aim for a weekly nonborrowed reserve objective that
tended to smooth the implied change in borrowing rather than
promote rapid changes in borrowing and rates that would have
to be reversed in the following week.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
22
The second largest source of intentional misses
in the nonborrowed reserve path arose when borrowing deviated
significantly from expectations within the final week of a
reserve period.
As noted above, if borrowing ran well above
expectations during the early days of a week, the Desk was then
faced with the alternative either of allowing huge excesses
and an overshoot in total reserves by strictly following the
path or else of absorbing some of the excess and coming out
below path.
If borrowing ran well below expectations during
the early part of a week, the Desk was faced with the alternative of forcing huge borrowing on the final day of the week
or else of coming out above its obJective for nonborrowed
reserves.
Because borrowing more often ran above expectations
than below, especially in the periods of rising interest '
rates, there was a tendency for the Desk to come out below
its weekly nonborrowed reserve objective.
If thi~ occurred·
in the final week of a reserve period, it meant that nonborrowed
reserves would also come out below the average path level for
the period as a whole.
Moreover, even if it occurred earlier
in the period, it meant that there might be a large gap to
fill in the final week, with the implication of a large drop
in borrowing in that week, even when the ~OMC's new instructions
might involve a rise in the subsequent week.
As noted
earlier, in dealing with the problem of transition between
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
23
reserve periods the Desk chose to avoid large swings in
borrowing around the final week, and so it would aim for
nonborrowed reserves to come out below the path value.
(As
defined here this latter source of miss would be captured
under "transition between reserve periods.")
There were several other less important sources of
intentional misses in the nonborrowed reserve path.
The Desk
treated special borrowing--for example, borrowing resulting
from a reserve need created by a computer failure--like a
provision of nonborrowed reserves and, hence, aimed to come out
below path by the amount of such borrowing.
(Borrowing by a
large bank facing a special extended need for accommodation
during May through August was also treated like nonborrowed
reserves, but because it was recurring and after a while fairly
predictable it was formally built into the paths.)
On two
occasions the Desk cited the projected strength in the aggregates
as a reason for accepting some shortfall from path.
the federal
Finally,
funds rate constraint accounted for a relatively
small deviation in only one of the reserve periods, the four
weeks ended May 21.
At the end of that period, the Desk drained
reserves, even though projections indicated no need for action,
when federal
funds traded at rates below the lower limit of the
Committee's funds rate range,
In other periods, while the
behavior of the funds rate on occasion had a very temporary
and modest influence on open market operations, it did not
constrain the Desk's ability to hit average path levels,
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
24
Among the unintentional misses, there were many
occasions when dealer propositions proved skimpy, and the
Desk was unable to arrange the volume of open market operations
plannedo
When this occurred on the final day of a reserve
period, which happened twice during the year, the Desk missed
its average nonborrowed reserve path.
A far more important
source of miss ~ere projection errors of reserve supplies.
According to the estimates shown in table 7, they
accounted for an average miss of about $57 million (ignoring
sign) per reserve period.
However, this may understate the
size of errors from this source because of the interaction
of projection errors with other factors.
For exaP1.ple, an
overestimate of reserve supplies early in a statement week
that
led the Desk to hold back its provision of reserves
might result in upward pressure on the federal
funds rate
and borrowing levels that exceeded expectations.
If the
Desk at the end of the week then chose to deliberately miss
'
on its nonborrowed reserve objective because of the higherthan-expected borrowing, this would show up in the accounting
as ''weekly deviations of borrowing
.
rather than a proJection
•1'
6/
error. --
'
6/
On the other hand, because the estimates of
projection-errors were computed in the table as a residual-that is, after accounting for deviations from all other factors-there is the possibility that they could overstate their impact
to some extent. An alt':=rnative P1.ethod of co-npm:1.ng t'rie devia-t.io~1
arisi.:.1.g f:'.:'om res·3rV3 rr.:,jection rr1isses is to j_oo,: at one~day
forecasting errors of weeklv average reserve supplies on Wednesdays
divided by the average number of weeks in reserve periods. This
produces an estimate of about $37 million per reriodi which could
be thought of as a lower bound estimate of the deviation one
might expect from this source.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
25
For total reserves the deviations from path in
absolute value were much larger than those for nonborrowed
reserves.
Over the fifteen reserve periods, they ranged
from a low of $65 million to a high of $888 million and
averaged $325 million (ignoring sign) or 0.75 percent of
total reserves.
On average, total reserves exceeded path
by about $81 million per reserve period.
Deviations from
the total reserve path mainly reflected misses in required
reserves and, hence, largely mirrored deviations from the
Committee's monetary aggregate objectives.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
26
The Market's Perception of Monetary Policy
A key issue under the new procedures is how the
System's supply-oriented strat~gy affects the perceptions
and actions of bank and financial market participants.
How
efficiently do market participants translate the levels of
nonborrowed reserves (NBR) achieved by the Desk into the
portfolio decisions and interest rate movements that will'
promote attainment of aggregate objectives?
Can operational
or institutional changes improve the speed and appropriateness
of the financial system's response to NBR changes?
The design of System operations is clear enough.
By
supplying nonborrowed reserves in line with a desired path for
growth in reservableM-2 deposits, the System assures the public
that deviations in monetary growth will produce corresponding
increases or decreases in the amounts that will have to be
borrowed at the discount window to cover reserve requirements.
So long as the demand for total reserves exceeds the NBR path,
increases or decreases in such borrowing bring upward, or
downward, pressure on the federal
funds rate and encourage
portfolio adjustments that should work toward offsetting the
overshoots, or shortfalls, in money growth.
If the System
wants to speed up the adjustment process, it can lower or
raise the NBR path and/or change the discount rate.
Whenever
the demand for total reserves falls below the NBR path, then
the procedure calls for supplying reserves in excess of the
volume demanded, producing a drop in the federal
funds rate
toward whatever lower limit the FOMC sets as a constraint on
operations.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
27
While the broad outlines of this strategy are well
understood by financial market participants, they find it
harder to assess the current and prospective thrust of monetary policy under the present procedures than previously.
Before October 1979, the Desk's operations gave the market
an immediate reading on a change in posture with regard to
the
federal funds rate.
At present, market participants can
follow the actual behavior of borrowing"at.the window on-a
weekly basis--and assume that on average it reflects the
Desk's new operating procedures.
But the federal
funds rate
associated with each borrowing level can range fairly widely.
Confronted with a choice between borrowing or the federal
funds rate as an indicator of the Federal Reserve's operational stance, market p~rticipants have a natural preference
for the latter.
Not only is the rate observable continuously
but market participants also know the direction the rate has
to move to correct monetary growth when it is too rapid or
too slow.
,The rate also bears directly on the costs incurred
by banks in borrowing
short term and by investment houses in
financing their positions.
For market observers, the supply-
oriented strategy boils down to a procedure for generating
movements in the
federal funds rate.
The new strategy is thus marked by some ambiguity
in the signals
the market reads of policy's current thrust
from the funds rate.
The Desk can hit its NBR targets--and
the levels of borrowing that correspond--reasonably well over
an intermeeting interval.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
But the federal
funds rate, the
28
market focuses on as a policy indicator, can vary widely for
a given level of borrowing.
Changes in the federal
funds
rate appear to be strongly influenced not only by the borrowing level itself
but also by past borrowing experience
and by market expectations of future rate developments.
For an earlier period, July 1972-Decernber 1974, the
relationship between member bank borrowing and the spread
between the federal
funds rate and-the basic discount rate •
7/
is shown in chart 1. -
In this period,, the rate spread
exhibited considerable dispersion in relation to borrowing
when the latter moved beyond about $1.5 billion.
the
But since
federal funds rate was the control instrument, it was a
,
matter of indifference to both the Desk and market participants that borrowing was bouncing around.
Of greater inter-
est for current purposes is the fact that the relationship
shifted to the right in the October 1979-November 1980 peri-'
od.
jl/
As the chart shows, member bank borrowing after
October 6, 1979,involved a smaller average impact on the
federal
funds rate for a given change in bo~rowinq than
7/ The period is chosen because borrowing ranged
widely and-discount rate policy allowed a sizable positive gap
to develop between the federal funds rate and the discount rate.
In the 1977-79 experience, the discount rate was kept in closer
alignment with the federal funds rate, so that borrowing infrequently rose much above $1 billion.
8/ Discount window borrowing excludes emergency
borrowing In both periods. The period in which the federal
funds rate was below the discount rate was eliminated from
analysis of the recent period since the level of borrowing in
such a period is typically not sensitive to the size of the
negative spread prevailing.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
CHART 1. ADJUSTMENT BORROWING VS. THE SPREAD BETWEEN THE
FEDERAL PUNDS RATE AND THE PEDERAL RESERVE DISCOUNT RATE~
- - - - - - - - - - • 7172 TD lZ/7-t lfl.ftTIUNSHir
- - - - 10/79 10 11/&) RELA'rION9fIP
(C
-----------------------
la.]
f-t
a:
0::::
Ln
(>
f-t
z
::::)
<) <)
D
c.:,~
(/)
......,
□
(/)~
I\,)
z
......
Ill
::::)
CX)
L'.
u.J N
H
a:
n::
CJ)
□
z
::)
,....
La...
i<>
CJ
......__~◊--'----~--~---------,.,___ _ _ _ ___
0.0
500.0
1000.0
1500.0
2000.0
2500.0
ADJUSTMENT BORROWING
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
N..lJLl J97Z TD ~EMBiR 197i - 3 N£EK 11:JVING ft~Rf&f
3000.0
3500.0
29
the earlier period.
:11
The relationship between borrowing and
the rate spread also exhibited a much looser fit after October 6 than in 1972-74.
(An R2 of 0.48 compared with one of
6
the 0.83 for the earlier period for a linear relationship in
the logs.)
As noted below, one factor at work was the intro-
duction of a surcharge on the discount rate for frequent use
'
of the window by large banks.
On closer examination, the recent experience is not
all of one piece.
The da,ta suggest, instead, a basic rela-
tionship fitted to 23 of the 41 weeks of the period when the
yield spread was positive.
logs graphed in chart 2.)
(See linear relationship in the
Th~ basic relationship (R 2 equal
to 0.8~ suggests a greater shift in bank borrowing behavior
since 1972-74 than the regression line of chart 1,
all the weeks of positive spread.
which used
The basic relationship
suggests that member bank borrowing of about $2.3 billion was
needed to produce a 2 percent rate spread in the recent period,
whereas $1.5 billion of borrowing was associated with such a
spread in 1972-74.
The market's perceptions of rate vari-
ability relate much more, however, to three distinctive
episodes, in which the spread departed dramatically from the
basic relationship.
'}J An even stronger shift in borrowing behavior was found in the paper
by Peter Keir, written for this staff review. That paper suggests that a change of
$BO mill ion in borrowing is associated with a move of 10 basis points in the federa
fu~ds rate after.October 6, whereas a change of only $40 million was associated with
such a move earlier. See Peter Keir, "Impact of Discount Policy Procedures on the
Effectiveness of Reserve Targeting, 11 p. 27.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
CHART
I
~
2D
~
ADJUSTMENT BORROWING VS. THE SPRLAD BETWEEN THE
FEDERAL fUNDS RATE ANO THE fEDERAL RESERVE DISCOUNT RATE*
ONfflIN9 NfEK5 H l"UNJ, MTC Wft5 BEL<M "JHt: 0]5DOUNT
we
- - - - 10/18 ro 12/80 REDR:SSI0N RELR:rU)ISHlP E><CLLDJND OU'D..]ERS
0 INO.UCID [N ftf2ESSIDN
A EXO.WEO rRJH RE~ION
w------------------------
w
~
cc
et::: IJ1
~
t-t
I
z
:::)
I
!
CJ)
/
I
.........
D
I
j.
(f)~
jf
9/80 TO 12/80
:::)
z
1--f
/
~/
1/
/
E
/
WN
~
E--f
Ai -=p--~"jG,---0
0:::
6---0..
a:
e, er,
(f)
D
~
I
D~
(...J
z
::)
,
r-t
,~
11/79 TO 1/80
la..
□
0.0
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
500.0
1000.0
1500.0
2000.0
2500.0
ADJUSTMENT BORROWING
II
nl'!''TT'1AS:-I' , a?CI '1n nF"l"S:-MRF'IRP , aAn -
"ll: Nn.JTM Nnll TtJI! Rvr:'IPA::r'
3000.0
3500.0
30
In the first, borrowing declined from the $1.9 billion level of four weeks in November 1979 to the $1.3 billion
level of the five weeks in January 1980, but the rate spread
actually increased by about 1/2 percentage point.
A part of
the explanation doubtless lies in the fact that high OctoberNovember borrowing levels had forced many banks into the
window so that maJor banks were prepared to continue bidding
aggressively for
window.
federal
funds in order to stay out of the
Year-end expectations of economic weakness in 1980
may also have led the major banks to shorten up a bit on their
funding, exerting a bit more pressure on the money market.
Also at work may have been the banking system's response
under the new regime to financing seasonal loan demands,
which typically peak in December.
The rate spread did fall
back somewhat in mid-January, but it remained on the high
side for some weeks.
Perhaps one influence on the banks after
mid-January was the increase in loan takedowns as borrowers
became increasingly concerned that some form of credit restraint would be a part of the anti-inflationary package
then being put together by the administration.
In this atmos-
phere the increase in the discount rate from 12 to 13 percent
on February 15 led banks to bid up the
federal funds rate a
corresponding amount almost immediately so that the spread
relationship was not affected.
The second episode began with the announcement on
March 14 of the credit control program and a 3 percentage
point surcharge on the discount rate for large banks.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
~
31
Nearly the full amount of the surcharge seems to have been
translated into a higher federal
funds rate within three
weeks as evidenced by the sharp rise in the spread.
The
rapidity, with which this occurred, probably reflected the
restrictive thrust of the credit control program at a time
when bank loan demand was running very high, as well as the
sustained recourse of the banking system to discount window
borrowing.
The sharp rise in the federal
funds rate appar-
ently reflected the efforts of the larger banks to stay out
of the window.
Surcharge borrowing was, in fact, very low.
Once the money supply began falling far below path in April,
borrowing fell rapidly under the new procedures and the
federal funds rate fell below the discount rate by the first
week in May 1980.'
The third, and in some respects, most interesting
episode began in August 1980, when a surge in money supply
led to an'irnmediate rise in discount window borrowing as the
'l.0/
demand for total reserves exceeded the NBR path.-
However,
since member banks had been essentially out of the window for
some months, upward pressure on the federal
modest.
funds rate was
After having risen from 9 percent to the 10 percent dis-
count rate in late August, the federal
funds rate rose only
10/ Between early May and early August, the discount
rate surcnarge was eliminated and the basic rate was lowered
in thl:;-ee. ateps to 10 percent. Each of these changes reduced
the spread between the discount rate and the federal funds
rate without affecting the latter, whereas discount rate increases translate quickly under reserve targeting into a corresponding rise in the federal funds rate.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
32
about 1/2 percentage point further by mid-month.
Member bank
borrowing rose from the frictional levels prevailing in midAugust to around $1.2 bil~ion in mid-September, an increase
that was rapid by historical standards for a period of economic recovery.
the
federal
Market participants took the moderate rise in
funds rate as an inadequate response to the con-
tinued rapid expansion of the money supply after August's
19.3 percent annual rate of growth in M-lA.
A major revision of expectations regarding the
.
outlook 'for the economy and inflation had already begun by
mid-September.
While the feoeral
150 basis points since July,
funds rate had advanced
six-month and one-year Treasury
issues had,risen 300 basis points.
Long-term bonds had
risen over 1 percentage point in yield, as evidence accumulated of a resumption in economic growth and as buoyancy in
producer prices eroded earlier hopes of near-term progress
in reducing inflation.
federal
The market wa5 disappointed that the
funds rate did not rise more vigorously.
Talk that
the Federal Reserve was not following through on its moneiary
I
objectives probably contributed to the widespread resurge~ce
of inflationary expectations.
I
j
The rise in the Federal Reserve discount rate from
I
10 to 11 percent on September 26 seemed to dissipate suchl
concerns and contribute to expectations of still higher rates
ahead.
The
federal funds rate moved up by more than l pet-
I
centage point so that the spread between it and the discount
I
I
rate widened.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
In succeeding weeks the spread rose from
33
1 percentaqe ~oint to over 2 percentage ooints in late
October-early November, for similar levels of borrowings.
As borrowing increased further under the new procedures, the
spread continued to rise
l
percentage point or more above
the fitted rel~tionship of chart 2.
The rise in the dis-
count rate by 1 percentage point on November 17 and the
institution of a 2 percentage point surcharge were quickly
translated into further increases in both the federal
rate and the spread, much as in March.
funds
And a second 1 per-
centage point increase in basic rate and surcharge on December 5 had a similar effect.
, The extraordinary rate increases between midOctober and mid-December, with federal
funds rising from a
bit over 12 percent to 19 to 20 percent--appear to reflect
an interaction of bank and market expectations with Federal
Reserve actions.
The increase in the spread after the
September discount rate increase seems,related to both the
general shift in expectations concerning interest rates and
the rapid ,increase in business loans at banks since midsummer.
The fact that rapid money supply growth threatened
achievement of the FOMC's 1980 objectives fed expectations
that rates would move higher.
The markets quickly translated
these expectations into higher rates in a self-reinforcing
process.
Participants repeatedly talked up the likelihood
of discount rate increases as the federal
funds rate rose
further above the discount rate--apparently on the theory
that catch-up increases were needed under the flexibility
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
34
principle specified in the announcement of the new strategy
in October 1979.
This interpretation became a part of the
market's assessment of Federal Reserve dedication to monetary
restraint.
The rise in the spread was taken as indicating a
further need for discount rate change rather t~gn g meas~re
I
of the pressure of banks' efforts to avoid recourse to the
window.
Operational or Institutional Changes
An important question raised by the experiencelwith
I
I
1
the supply-oriented strategy is whether changes in the SI stem's
procedures or the operation of the discount window could increase the predictability and speed of market and bank r sponses to changes in the supply of nonborrowed reservesl
One part of the question relates to drawing the reserve paths,
I
I
the other to the predictability of the rate effects flowing
from the supply-oriented strategy.
The data embodied in the basic relationship shown
in cnart 2 raise doubts whether the System was vigorous enough
in August and in September in lowerinq the NBR path as a result of
overruns in the demand for total reserves in relation to
path.
The rise in borrowings from frictional levels to over
$1 billion between mid-August and mid-September may have been
rapid by past standards, but the resulting increase in the
federal funds rate of about 5/8 percentaoe ooint aopeared
small to the market in relation to the overshoots reported
weekly in the money supply.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
With the benefit of hindsight,
35
a quick rise in borrowing levels nearer to $2 billion would
have carried greater conviction, both to the markets and
discount window borrowers as well.
A related issue is the trade-off between increases
in borrowing levels and increases in the discount and/or surcharge rates.
Had the System lowered the NBR path more vigor-
ously in August and September, the federal funds rate would have
risen more rapidly.
Some bankers argue that attaining a
critical mass of borrowing--say $1.5 to $2.0 billion--exerts
availability effects on bank behavior that can substitute for
the restraint generated by the rise in the federal funds rate.
Whether high borrowing levels could have achieved the same
degree of restraint at somewhat lower levels of interest
rates is problematical.
The possibility deserves further
analysis.
Would improving the predictability of the interest
rate effects of a reserve strategy be desirable?
The answer
depends upon where one strikes the balance between reducing
interest rate volatility and insuring that short-term interest
rates are free to move.
The present strategy involved a clear
break with the 1978-79 strategy, in which stability considerations weighed rather heavily.
Policymakers in still
earlier periods--for example, 1973-75--were often more aggressive in moving the federa 1 funds rate than was the case in 1978-79.)
1978-79.}
The volatility of short-term interest rates from
day to day and week to week over the past year 1.s a natural
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
36
consequence of the change in operating strategy.
If market
participants base their interest rate forecasts more than
before on projections of economic activity in an inflationary environment, then short-term rates, including the
federal
funds rate, are likely to reflect shifts in expecta-
tions as new data on the economy come
in.
The difficulty
that market participants have had in reading Desk operations
adds to uncertainty and the scope for'rate movements.
One can argue that expectational shifts have
generally tended to work in the direction needed to effect
desired changes in the growth of the aggregates.
of sluggish growth, the federal
In periods
funds rate has tended, to fall
more rapidly because of expectations.
When aggregate growth
has been excessive for a sustained period, the funds rate has
tended to rise faster than one would expect on the basis of
the increases in adjustment borrowing that have occurred.
The variability in the relation between borrowings and the
federal
funds rate has seemed to operate in the direction
needed to bring monetary growth back to path more quickly.
On the other hand, the present procedures involve
a considerable amount of short-run rate volatility as the
markets seek to appraise the Federal Reserve's current
posture as well as the economic outlook.
This probably has
imposed considerable cost on the Treasury and other borrowers
in financial markets, who ultimately pay for the greater
underwriting risks encountered in rapidly changing markets.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
37
The question is whether changes in System procedures or
financial institutions could moderate this volatility without
impairing the advantages of the supply-oriented strategy.
An institutional way of imposing a more predictable
relatidnship between discount window borrowing and the f edera 1
funds rate would be to provide access to the window through ·a
tferetl structuring of credit lines, which would be available
to banks at escalating rates without administrative constraint.
The interaction of the demand for total reserves
with'the supply of nonborrowed reserves controlled by the Desk
would, as now, determine the aggregate volume of borrowing.
The structure of credit lines and related discount rates
would also essentially determine the federal funds rate
since banks would be able to arbitrage between the discount
window and the money market.
Under this system, changes in
borrowing levels because of either a rise in the demand for
total reserves or a lowering of the NBR path would lead to
a fairly predictable change in the federal funds rate and
reduce this source of uncertainty to the market.
A tiering
of-credit lines -and discount rates would, of course, pose'
'
serious equity questions by differentiating access to the
window among various classes of institutions.
Designing the
gradient of the rate schedule would be difficult, and adjustment of the schedule would be cumbersome if one wanted to
reduce, or increase, the rate impact of changes in borrowing levels.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
38
The same reduction of instability in the relation
between borrowing and the federal funds rate miqht be achieved
with less impact on institutional relationships by introqucing
1
a flexible
or 2 percentage point, band for the federal
funds rate into the Desk's conduct of open market operations.
The Desk could pursue its weekly nonborrowed targets mor,e, ,,
flexibly, being somewhat more aggressive in supplying reserves
toward the upper end of the band and in absorbing reserves ,at
the lower end of the band.
The NBR weekly target would •remain
the primary obJective, but the Desk would provide some resistance to over- or undershooting, especially during times of
rapid dynamic change in borrowing levels or the discount rate.
The purpose would be to smooth the transition in rates from
one level to another, in order to reduce short-term instability.
To be sure, such an approach could generate concern
among market and academic observers that the System was re-.
verting to a policy of restricting movements in the federal
funds rate.
However, System practice could soon allay such
concerns by continuing to allow rates to move when money
growth was excessive or deficient.
The shift to the supply-oriented'strategy reflected
a need to restore credibility to System policy after an extended period in which oolicymakers did not allow interest
rates to move sufficiently to cJntrol money growth.
The
System's subsequent commitroent to pursuing its monetary
objectives through this strategy, augmented by discount rate
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
39
policy, has been much in evidence.
The short-term volatility
of interest rates, as distinct from their cyclical variability,
has been one consequence.
It seems possible that this volatil-
ity could be reduced somewhat through the conduct of open
market operations without involving significant risk of reverting to the former federal funds rate strateqy.
The gain
would be a reduction of short-run uncertainty and the associated rise in underwriting costs by smoothing of the interest
rate changes produced by a supply-oriented strategy.
One
might, of course, diminish in the process the self-reinforcing
effects of expectations, which at times can speed up the
monetary control process.
Where one strikes the balance in
such matters depends on the relative weights one attaches to
these considerations.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
APPEN~IX
A-1
Technical Adjustments to the Reserve Paths
Regular procedures have been developed for adjusting
the reserve paths over the intermeeting period for technical
changes affecting the money-reserve relationship.
The original
paths, estimated shortly after the FOMC meeting by the Board
staff, incorporate explicit assumptions for excess reserves
and required reserves needed to support the growth of reservable
liabilities not contained in ~-2, such as large time deposits,
net interbank deposits, and Treasury deposits.
They also
incorporate assumptions regarding the mix between deposits and
cu~rency, the composition of deposits by type and maturity,
and their distribution among banks by size and membership status.
As new data become available each week, the paths are reviewed
(generally on Fridays) by the senior Board staff and the Account
Manager to see whether changes are needed to maintain their
consistency with the Committee's aggregate objectives.
For
example, if net interbank deposits are growing more rapidly
than originally projected, less reserves would be available
to support the de~osit components of M-2. Such a circumstance
would normally call for an upward adjustment to the paths.
Similarly, if currency growth is running above expectations,
the growth of other components of M-2 would need to be restrained
an~ hence, a downward adjustment to the reserve paths would
be called for.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
J
A-2
While the procedures for adjusting the paths for
technical considerations are for the most part straight~orward,
two issues raise difficult conceptual questions and deserve
further study.
On~ involves the question of how quickly and
by how much the paths should~be adjllsted as new information
becomes available.
The more typical practice has been to apply
only a portion of potential adJustments to the paths each week
since the technical factors are volatile from week to week
and, hence, full application may have to be reversed in succeeding
weeks.
There is also the practice of making adjustments only
when their size is deemed significant, and so the paths are
generally not adJusted if the potential amount is less than $50
million.
These practices, however, can be questioned on several
grounds.
In the first place, many of the adjustment items--for
example, reserves to support net interbank deposit and Treasury
deposits--are fully reflected in the projected demand for total
reserves.·
Since it is the difference between the projected
demand for total reserves and the nonborrowed reserve path
that largely determines the estimated borrowinQ level for the
week, the practice of not fully adjusting the path for these
items introduced unnecessary changes in borrowings, and hence
in money market rates, that are not related to growth of the
monetary aggregates relative to the Committee's objectives.
Secondly, even those adjustment items that do not directly
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A-3
affect the demand for reserves (for example, those involving
currency growth) still represent the staff's most up-to-date
judgment of the changes which are necessary to keep the path
in line with the Committee's monetary aggregate objectives.
Nd:to adjust the paths in full for these items, seems to imply
some special significance to the estimates made immediately'
after the Committee meeting even though they were based on
less information. Finally, ignoring even small potential
adjustments to the path can have significant implications for
money market conditions.
This is especially true when the
adjustments occur either in the latter weeks of a reserve
period, as they have a multiple effect on the borrowing estimate,
or when banks are already relying heavily on the discount window
and, therefore, are especially reluctant to increase their
borrowings further.
The other issue that deserves further study--and
which could also benefit from Committee consideration--involves
how much policy weight over the intermeeting period should be
given to the broad versus the narrow aggregates.
Currently,
the primary emphasis under the reserve approach is on the narrow
measures.
Deviations in their growth from the Committee's
shor~-run objectives affect
1
the demand for reserves relative
to the nonborrowed reserve path and thus influence
the borrowing
levels and money market rates that develop over the intermeeting
period.
Many of the components of M-2, however, are nonreser,vable
and so their growth has no policy significance between meetings.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
A-4
These include personal savings and small time deposits at
thrift institutions, money market mutual fund shares, overnight
repurchase agreements, and overnight Eurodollars (held at
Caribbean branches),which together constitute nearly one-half
of M-2. Moreover, the short-term policy weight attached to M-2
will diminish further under the provisions of the Monetary
Control Act as reserve requirements on personal savings and
small time deposits at commercial banks are phased out.
When
the Act is fully implement~d for member banks in 1984 and,
hence.transactions balances become virtually the only M-2
component subject to reserve requirements, open market operations
between meetings, in effect, will be focused almost entirely on
control over the narrow monetary aggregates--if current procedures
are maintained.
There are two ways the Committee can increase the
role of M-2if it so chooses.
One approach would be for the
Committee to give more emphasis than currently to M-2 at each
meeting when it selects its short-run aggregate growth
objectives and determines the initial borrowing level used in
constructing the nonborrowed reserve path.
The aggregate
objectives themselves could be specified in terms of growth
.
of the narrow measures. However, in choosing these objectives
and the initial borrowing level, the Committee could give
greater recognition to whereM-2 (and perhaps still broader
money and credit measures) stand in relation to their yearly
range~.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
For example, if M-2growth was running above its
A-5
yearly range
.
, this would be strong reason for considering
relatively low short-run growth objectives for the narrow
'
aggregates and/or a relatively high borrowing level--even if
the narrow aggregates were behaving satis~actorily.
The other alternative woul~ be for the Committee to
indicate to the staff some specific short-run weighting scheme
for its aggregate obJectives similar to what was done under
the federal funds rate procedure.
For
examole, the Committee
'
•
could direct the staff to construct and maintain the reserve
paths in a way that gave equal weight to achieving its short-run
obJectives for growth of, say, M-lB and M-2.
The staff's
procedures for drawing up the initial paths would remain the
same, but subsequent adjustments would insure that the Committee's
priorities are being reflected in the paths rather than a
weighting scheme based on the structure of legal reserve requirements.
This would involve making adjustments between meetings
to the reserve paths for deviations of growth from the Committee's
objectives in nonreservable
deposits similar to those that are
;
currently being made for currency.
Such an approach can have
significant effects on short-run policy.
For example, had the
above mentioned 50-50 weighting scheme been in effect in 1980,
it would have called for sizable downward adjustments' to the··'
nonborrowed reserve paths over the summer months.
This would
have given an earlier start in letting interest rates adjust
upward over the balanc,e of the year.
https://fraser.stlouisfed.org
Federal Reserve Bank of St. Louis
- ~ ... -
"II
.c
@FedFRASER