View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

Yor release on delivery
Tuesday, March 11, 1969
Approximately 4 P.M. - E.S.T.


kemarks of


Board of Governors
of the
Federal Reserve System

At a Seminar
Sponsored by
The Department of Economics
of the
University of üichigan
Ann Arbor» '.debitan
»arch 11, 1969


I a : often appalled, amazed, amused, or horrified at what broad
conclusions can be constructed upon narrow foundations, when I read news
accounts interpreting the monetary statistics that the Federal Reserve
publishes each week,

llany stories present divinations that seem somewhat

akin to the oracular portents produced by prior civilizations after a
careful analysis of entrails, auspices, or other odd indices.

This is

not because the data are unimportant, but because figures are used to
carry a superstructure of analysis anc prediction that goes far beyond any
relationship to their worth or meaning.
Let me today speak to three misunderstandings which seem to
underlie numerous stories and which I believe lead to many errors, un­
warranted statements, and fault}7 analyses,
*** The first is a failure to recognize large random forces and
estimating errors present in rnost weekly adjusted data.
There are very few weeks— frequently even months— in which
much of the reported movement in monetary aggregates is not
primarily the result of statistical {noise9" by which I mean
something in the nature of static that I ’ll try to identify
more precisely later on.
*** The second, and more significant, are errors wnich attribute
the exact amount of the weeks1 or months' movements in the
monetary aggregates to a specific plan or action of the
Federal Reserve. Too many statements seem derived from an
incorrect interpretation of what the Federal Peserve does
based upon the highly oversimplified elementary textbook ex­
planations of the procedure by wnich banking systems create
money and credit. Too few statements recognize that in any
period the anount of money or bank credit created is the
joint result of a complex interaction among households, com­
mercial and industrial corporations, financial institutions,
the Treasury, and the Federal }eserve. When I point out that
in any one-month period, the Treasury, the ten largest banks,
and in a similar mAnner the ten largest corporations are likely
to have a greater impact on the narrowly defined money supply



than the Federal Reserve System, most people seem Indignant
that their carefully learned lesson from Economics I is not
a true statement of the operational facts.
*** Finally, the terms, ’
'money supply," or “money stock,1 are
used very loosely in much of both academic and popular dis­
cussions. There is often a failure to specify which monetary
aggregates are being discussed; which are considered signifi­
cant; or the advantages of one or another ’money supply" con­
cept as a basis of policy. This failure is important. The
differently constituted aggregates have significantly differ­
ent movements over short periods up to and including a year,
and the costs of influencing each may differ x^idely.
The problem of cost-benefit analysis is a major one.
To control a part of a whole may be expensive. Such costs
should be engendered only if the policy cannot be achieved
in more efficient ways.

"Koise, in the Tlonetarv Aggregates
Tables 1 and 2 show how widely reported changes in the money
supply can vary from the basic underlying trend of monetary policy.


one would expect they show that the longer the period under consideration
the smaller the impact of the non-policy-determined movements.


even over a quarter these other movements are large.
The non-policy-determined movements are actually of two very
different types.

The first, which I have labeled "noise," consists of:

operating misses; errors in estimating the actual data at the time
that operations for a period end; shifting seasonals, and irregular
movements which are temporary and the product of special factors.
The operating misses arise either because of errors in report­
ing, errors in sampling, or information not available when operations
must be ended.

For some time, the size of misses has been decreasing

The misses are small compared to the totals, but large com­

pared to weekly or monthly changes.

Table 1

(In billions of dollars; not seasonally adjusted)







Subject to

to 2.6

- .2
to .3

to 4.3

to 5.0

to 3.7

to 3.2

1 Week:
Average change
per period
Mean deviation
Standard "

to 5.2

to .7

4 Weeks:
Average change
per period
Mean deviation
Standard "

to 4.6

to .7

- .7
to .7

- .3
to .9

to 2.5

to 5.6

to 5.7

to 4.1

- .8
to 6.5

to .7

- .3
to .3

- .3
to 1.7

- .3
to 1.7

to 5.9

to 2.6

-1 .4
to 8.9


13 Weeks:
Average change
per period
Mean deviation
Standard "


* In order to make the orders of magnitude roughly comparable to demand deposits in terms of their reserve impacts
the reserve changes for operating misses, excess reserves, and total reserves have been multiplied by 6.6 while
time deposits have been multiplied by .28. Seasonal reserves are based on the credit proxy seasonal and auto­
matically carry the reserve ratio for the entire proxy.
** Seasonally adjusted.

Table 2

(In per cent; not seasonally adjusted)

Deposi ts
Subject to












Operati ng









4 Weeks





13 Weeks






1 Week

Source : Table 1.
* Seasonally adjusted.



The seasonal factors are large.

In addition, they are domi­

nated by irregular factors, particularly over short periods.

In many

cases, it is hard to determine by analysis of historical data what is a
seasonal and what is an irregular factor.

The demand for money will

vary greatly depending on the day of the week in which a month, quarter,
or year ends.

The same is true of the day on which traditional dividend

and tax dates fall.

The change in tax collection dates and percentages

has been important in most recent years.

The day on which the Treasury

borrows and the form of its borrowings are critical.

Problems with

the debt ceiling and attempts to stay under it loom large.

While esti­

mates are made currently as to the impacts of these factors, they still
confuse the judgment of seasonal variation, particularly as observed
at the time operations take place.
The irregular elements include seemingly minor factors such
as the financing of a corporate take-over bid, a breakdown of a bank
computer, or a snow storm.

Each of these may cause even weekly average

changes to vary by over 100 per cent or more.

As an example of such

movements, examine pages 29 and 30 of the February 1969 Economic Indica­
tors and pages A17 and A18 of the January 1969 Federal Reserve Bulletin.
Each carries an estimate for the December 1963 change in the narrowly
defined money supply.

In one case the increase is reported as $1.2 bil­

lion which translates to an annual rate of growth of 7.5 per cent.


the second case the increase is estimated at $8.4 billion, or at an
annual rate of 53 per cent.

Neither figure is in error.

The first

weights the extremely unusual end-of-year changes in one way, the second


in a different way.

Neither gives a very good sense of the underlying

trend because of the dominant influence of very special factors that
were rapidly reversed.

These irregular forces were large enough, how­

ever, to bias strongly analysis of the two adjacent quarters in which
they occurred and for many purposes even the annual data for the two
Data calculated at the time operations end are the significant
data for operational purposes, but theoretically not for any policy im­

These estimates are subject to revisions as more information

becomes available, as full universe data replace samples, and as sea­
sonal forces are re-estimated.

These revisions between the money supply

as first reported and as currently reported are not shown on the table.
However, they averaged $152 million per week over the past three years.
They had a range of from $-1.4 billion to $1.0 billion.
ation was over $490 million.

Their mean devi­

Clearly, they make a significant amount of

noise which must be taken into consideration when one the re­
ported weekly changes.

In a somewhat similar manner,, we might note

that one part of the money supply, namely, non-member bank demand deposits,
is not subject to reserve requirements of the Federal Reserve nor is
information on these movements readily available.
rather great.

Their variance is

Their share of total demand deposits has been growing.

The weekly and monthly data for this component are estimates from other
types of data.

Specific information on how this component has changed

is available only semi-annually with a lag of four to eight months.


Tlie tables call attention also to the second and third types
of problems I noted in my introduction.

Many unsophisticated comments

and theories speak as if the Federal Reserve purchases a given quantity
of securities, thereby creating a fixed amount of reserves, which through
a multiplier determines a particular expansion in the money supply.
Ijuch of modern monetary literature is actually spent trying
to dispel this naive elementary textbook view which leads people to talk
as if (and perhaps to believe) that the central bank determines the
money supply exactly or even closely— in the short run— through its open
market operations or reserve ratio.
hard to dislodge.

This incorrect view, however, seems

Almost daily I read that last week or last month the

Fed increased the money supply by 5 per cent.
Such statements are simply inaccurate.

The growth of the money

supply in any period is the result of actions taken by the Federal Re­
serve, the Treasury, the commercial banks, and the public.

Over a long

period, the Fed may play a paramount role, but this is definitely not
the case in the short run.

To the best of my knowledge, the Fed has not

and probably would have great difficulties controlling within rather
wide limits the growth of the narrowly defined money supply in any week
or month.

I think control over a period such as a quarter might be

possible, but I know no estimates of what the cost of obtaining a tight
control would be in terms of destruction of our financial structure,
output, or employment.



Tables 1 and 2 show movements in reserves absorbing or realizing

The first column shows changes in the narrowly defined money
The seventh column shows changes in the total reserves furnished

by the Fed.

The lines between indicate some of the factors which can

absorb or inject reserves to cause variations from the simplified text­
book case.
To place the relationship between these two columns in a better
perspective, we can note that over the past 10 years the rate of growth
of the money supply has averaged about 80 per cent of the rate of growth
in total reserves.

On the other hand, the coefficient of determination

(r^) for quarterly changes in this period is only .27; or, on the average,
nearly three-fourths of the quarter-to-quarter movements in the two
totals are not statistically related.

For year-to-year changes the r^

is .73.
The tables serve only as general indicators of rough orders of

They give no indication of causal relationships.

As dis­

cussed later, required reserves of member banks are based on the deposits
created two weeks earlier.

Increased reserves equal to the requirements

added by the deposit creation are, with the exception of important
marginal differences, normally made available by the Federal Reserve
through purchases or sales of securities.

In addition, securities are

also purchased or sold to offset seasonal or other changes in the socalled technical factors which add or subtract from member bank reserves.
These are mainly float, gold, currency, and Treasury balances, and other
accounts at the Federal Reserve.

In effect, the Federal Reserve creates or depletes reserves
to offset technical and seasonal changes in supplies or requirements.
However, as indicated by the second column of the table, in many weeks
shifts in reserves do take place simply as a result of misses in esti­
mating what technical as well as seasonal movements are occurring.


column is an indication of the difference between the final reported
net movements in reserves and those expected when open market opera­
tions ended for the week.

The figure for all three periods is the same

on the assumption that only the increases in the final week of a period
will influence that period’s results.
Host of the changes in the technical factors and for the
Treasury’s balances at the Federal Reserve can be considered as exoge­
nous, i.e., determined outside the system.

VJhile some movements might

be controlled, it is normally considered more efficient to offset them.
As an example, and as noted, the size of Treasury balances at member
banks varies widely from week to week.

While efforts are made to mini­

mize certain types of reserve impacts of these r;:ovemeiits, the totals
still remain large.

They are a major cause of week-to-week variations

in the money supply as funds move back and forth from private deposits
to the Treasury.
The utilization of the remaining reserves to meet the require­
ments for newly expanded time deposits, against currency, against demand
deposits, or held as excesses depends primarily upon portfolio balancing
decisions of the banks and the public.

The equilibrium from these

decisions is clearly influenced by the Federal Reserve.



however, may never be exactly reached and it may take a considerable
time before changes in demand schedules work themselves out.
On any given day, banks stay almost completely loaned up.
The individual bank, particularly the large money market banks, can
vary their loans and investments rapidly, with small costs and within
wide limits.

Thus, day-to-day changes in bank reserves tend to show up

almost immediately as changes in demand deposits and the narrowly de­
fined money supply.
We note from Table 1 that on the average in this three-year
period, the money supply increased by $150 million a week.
of weekly changes was, however, nearly $10 billion.

The range

If we look at the

mean or average deviations over the period, we note that the movement
in any given week was likely to vary from the average by $1.6 billion,
or by more than one thousand per cent.

As we move to the 4-week (month)

and 13-week (quarter) figures, we see that the amount of deviations
around the average change do not alter greatly.

On the other hand, the

longer the period, the larger the growth in the money supply.

As a re­

sult, the relative deviations decrease steadily.
When we look across the table, we see that in a single week
it xrould be difficult to predict what factors would accompany a change
in the money supply.

While they do not vary quite as much as the money

supply, the weekly movements of all of these factors except for time
deposits tend to be several hundred per cent as large as the average
change in the money supply
fall randomly around zero.

In i.'any weeks, of course, these other forces
Thus, their impact on changes in the money

supply in some weeks will net out.


Uhen we look at the data for a month, we' see the effect of the
fact that insofar as these movements are random the average of their
deviations does not increase.

Some of the factors are not random

An obvious example is excess reserves.

These may be relatively

large in a one-week period, but banks are able to bring them closer to
zero as they have time to average several weeks.

On the other hand,

seasonal forces and government deposits at member banks seem to vary as
much or more by months as by weeks.
Finally, over a quarter most of the noise has worked itself
out of the system.

The average change in the money supply, since it has

a positive weekly growth, is comparatively larger.

In this longer period,

the change in government deposits has, with the exception of demand
deposits, the largest relative fluctuations.

Time deposits have a some­

what greater absolute movement, but related to their size they move far

Their total reserve impact is considerably smaller.
Finally, let us look at the last two columns.

The first shows

movements in the amount of reserves furnished to take care of seasonal
demands for credit.

These seasonal movements exist strongly in private

They are even greater in the government deposits.

There is

also a minor seasonal in time deposits.
The final column shows the changes in the seasonally adjusted
bank credit proxy.


While the growth in deposits subject to reserves

1/ The bank credit proxy is the sum of the deposit liabilities of
member banks subject to reserve requirements. Changes in it for short
periods are highly correlated with changes in bank assets. Its values
and those for changes in required reserves are developed simultaneously.
It is the most easily estimated of the monetary aggregates and has far
less 3noise” than most of the others.


10 -

is more than twice as large as is that for the money supply, it has a
good deal more stability.

For example, the standard deviations for

the weekly changes in the credit proxy (not seasonally adjusted) is
less than two-thirds that for the money supply.

Its relative average

weekly deviation is only about one-third as great.
For the seasonally adjusted series shown in the table its
deviations are still less.

Of course, somewhat similar relative improve­

ments would also be shown if the seasonally adjusted money supply were
included in the table.

We see in Table 2 that its absolute deviations

are only about three-quarters of that for the money supply.

In terms of

its own relative size, the deviations are only about one-quarter as great.
It is probable that some trade-offs would be possible which
would reduce the present fluctuations in the narrowly defined money sup­
ply at the expense of the more broadly defined money supply— which the
bank credit proxy is frequently considered to be.

It would, however, re­

quire a new type of operations, and new markets and institutions.
or not such a trade-off would be worthwhile is not at all obvious.


the opposite side, the fact that the seasonal and irregular movements
in the money supply are so great does not, of course, mean that it is
either necessary or worthwhile for the Federal Reserve to offset them.
As pointed out initiallys both of these are problems for cost-benefit
The data do, however, indicate that in any attempt to control
the money supply closely and directly through altering the rate of
reserve creation, a great deal of the effort would be spent in trying



to offset noise or irregular movements.

At the same time, if reserves

were allowed to change only in accordance with the small changes re­
quired for additions to the money supply, the financial markets would
have to absorb all of the other forces removing or releasing reserves.
This would cause interest rates to fluctuate widely.

I know of no

studies which attempt to measure the gains or losses from such a policy
probably because most monetary theorists agree it would not be a logical
or sound endeavor.

The Federal Reserve Money -larket Strategy
It is also clear that as a matter of fact the Federal Reserve
does not attempt to increase the money supply by a given amount in any
period through furnishing a fixed amount of reserves on the assumption
that they would be multiplied to result in a given increase in money.
(The multiplier it is recognized would not be a constant but would vary
from period to period depending on relative interest rates and the
actions of groups other than the monetary authorities.


advocates of a policy based on highly controlled reserve generation
recognize that monetary action must also be taken either to anticipate
changes in the multiplier or to determine it.)
Instead, the Federal Reserve follows what has been termed a
money market strategy:— ^

2/ For those interested in more detailed statements of seme of the
concepts and problems, cf., J. I I . Guttentag, "The Strategy of Open
Karket Operations," Quarterly Journal of Economics, Vol. LXXX, No. 1
(February 1966), pp. 1-38; and P. H. iiendershott, The Neutralized Honey
Stock (Homewood, Illinois:Richard D. Irwin, Inc., 1968), 159 pp.
(Footnote 2 continued on next page.)


12 -


The operational directives of the Open i'arket Committee
specify alterations in the margin between required re­
serves and the amount of reserves furnished by the
System. These margins are considered significant in
their direct impact on bank operations, but probably more
important, they influence the interest rates on money
market instruments.


The amount of marginal reserves to be furnished and
the money market rates sought are picked so as to
influence the direction and rate of change of a more
remote intermediate monetary variable.


The desired rate of change in the intermediate mone­
tary variable is that judged to be the most effective
in aiding the economy to move toward its ultimate
optimum goals.

A possible side advantage of this strategy is that it can be
followed even though it might be impossible to get agreement among the
members of the FOFC either as to ultimate goals, or to the form or level
of an intermediate monetary variable, or as to how to define what strategy
is being followed.
Each decision-maker may believe one or the other of the follow­
ing types of variables is most significant at a given time:
Intermediate Monetary Variables
(1) Ilonetary or credit aggregates such as: the money
supply narrowly or broadly defined; deposits of
financial institutions; member bank liabilities or
credit; broader concepts of credit flows, liquid
assets, wealth, and lending.

2/ (Continued) The present discussion is my personal construct. As
indicated in the text, many and even most members of the FOKC might dis­
agree with my construct. They xrould build an entirely different one of
their own to express their view of what are obviously identical opera­


(2) Relative and absolute real or nominal interest
(3) The general atmosphere of the credit markets and
banking as reflected in expectations; demand for
credit; the amount of credit being supplied;
rates of change.
Because significant relationships exist among all these vari­
ables, influencing one will move others in the same direction although
not necessarily to the same degree.

As a result, if there is an agree­

ment as to the operational variables the manager is directed to follow,
there need be no meeting of minds with respect to which intermediate
monetary variables should be controlled or as to the proper degree of
The movements of these intermediate variables can be influ­
enced by a change in the level of any of the policy instrument variables
within the power of the Fed.

These are primarily:

Policy Instrument Variables
(1) The purchase or sale of open market securities.
(2) Repurchase agreements on securities.
(3) The discount rate.
(4) Regulation Q ceilings.
(5) Required reserve ratios.
A change in an instrument variable reacts with other forces in
the credit markets and the economy to shift the demand and supply for

At each Open liarket meeting, estimates are made as to the effect

changes in particular instrument variables will have on those money mar­
ket variables which respond most clearly to Federal Reserve policy, namely:


Honey larket Variables
(1) Borrowings of member banks from the Federal
(2) Net free reserves.
(3) The Federal funds rate.
(4) Call money rates to government bond dealers.
(5) The three-month bill rate.
The expectfed movements in the money market variables are
accompanied by estimates of growth in the intermediate monetary variables.
Each possible setting of the money market variables, given the projected
state of the economy, the banking system, Treasury operations, etc., is
expected to lead to a unique growth rate for an intermediate monetary
Debates may occur with respect to desired goals ; desired move­
ments of the intermediate financial variables.: the importance of specific
instrument variables:, or as to the correctness or errors in the judgment
models— which are used to estimate changes in the economy, as well as
the changes in the intermediate variables, and the money market results
of shifting the instrument variables.
All these considerations are summed up when the manager of
the Open liarket Account is instructed to buy or sell securities in order
to achieve specific (within a range) values for the money market vari­

The manager of the Account operates in the securities markets


At times, because of outside influences, the specified

relationships for all variables cannot be achieved simultaneously.


this occurs, the manager uses his discretion in an attempt to achieve


those settings which he believes are most consistent with the goals of
the Cominittee.
This intent to control intermediate monetary variables through
the money market variables is shown by the inclusion in most directives
of a proviso clause.

The manager is provided the growth rate for the

bank credit proxy (within a range) expected to result from the directed
settings of the money market variables.

If the proxy moves outside the

projected limits, he is instructed to operate in the open market so as
to alter the money market variables in order to influence the credit
proxy toward its projected path.

The proviso clause is an attempt to

correct for errors which may arise if the relationships among the money
market variables and the intermediate monetary variables have not been
projected correctly, or if errors were made in projecting the other
financial and economic variables which also influence the proxy's growth.
This picture of operations can be expressed symbolically:







Intermediate monetary variable
Borrowed reserves
Free reserves
Q ceiling
Treasury bill rate
Federal funds rate
Call money rate to dealers
Economic activity
Liquidity preference of corporations»
banks, financial institutions, etc.
Treasury cash management
Discount rate
Required reserves
Open Market operations

rb ; rf ; rc

K (Rb , Rp, Q, rb , rf , rc , GNP, L, T)

r (rd , Rb , Rp , GNP, L, T)



The change in the intermediate monetary variable, however de­
fined, is determined by the interaction of the Federal Reserve controlled
variables, certain money market rates strongly influenced by the Federal
Reserve, changes in output and pricesr movements in the financial sector
and liquidity functions; and the Treasury as in (1.0).
The Federal Reserve action may influence directly the IIIV•


also will influence money market rates as in (2.0).


Rb, Rf



- I1SV


R O' RR, S)


The change in the intermediate monetary variable approximately
determines the change in required reserves two weeks later (3.0).


the change in required reserves, the manager of the Open liarket Account
can (within the limits of his operating misses) determine exactly the
level of net free reserves (4.0).

The banking system, given a level of

net free reserves, determines its own level of borrowings and excess re­
serves simultaneously.
Uhen the manager is directed to influence the money market
variables and through them intermediate monetary variables, he cannot at
the same time control the changes in total reserves.

Most reserves

additions will follow directly from the previous changes in the IMV
(credit proxy).

The manager will operate so as to furnish slightly

more or less than the change in required reserves (4.0) so as to interact
with the market (2.0) to obtain the settings he is attempting to achieve.
This means in most cases, he will furnish most (say, SO per cent or more)
of the changes in required reserves which have been previously deter­
mined by the various market interactions.


An Elastic Currency
It is now possible to restate one logical reason for follow­
ing the money market strategy.

We saw liow great are the misses, the

random movements, and the influence of other forces on reserves when
compared to the changes required for growth in the narrouly defined money

If one attempted to increase reserves according to an exact

schedule, the market would have to shift rapidly in order to accommodate
seasonal forces, errors in operation, Treasury cash operations, and the
type of irregular movements which the Federal Reserve now accommodates.
An attempt to control growth in the money supply directly
through controlling the amount of reserves created runs into the diffi­
culty that in any quantity-price relationship if one controls the quan­
tity tightly the price must be allowed to move freely and through an
extremely wide range.

In addition to many other considerations, the

problem would have to be faced of uhat costs and what structural changes
the economy would experience if interest rates fluctuated widely as the
result of an attempt to control directly a single use of monetary re­
Our financial structure and capital markets are extremely well
developed and efficient.

The amount of funds bought and sold in our

money markets averages well over 10 to 12 billion dollars per day.


amount of money raised on a gross basis by the economy totals over $600
billion for maturities of under one year and over $220 billion with
maturities of over a year each year.

In such a system, major advantages

result if the monetary aggregates react flexibly to absorb the daily,
weekly, and monthly seasonals, shock, and other irregular forces.


13 *

This need for flexible reactifttts in the monetary aggregates
was a major factor in the formation of the Federal Reserve.
always been a central interest in its operations.
flexibility may be greater today than in the past.

It has

The need for such
Our capital markets

operate with an extremely low ratio of equity capital.

Ue have developed

highly specialized financing institutions and techniques.

The under­

writing of our public debt is done at extremely low margins.

These are

possible because the market does not have to shoulder the risks of
widely fluctuating interest rates from irregular short-term movements.
The additional reserves created to satisfy the purely seasonal or
irregular demands for short-term funds disappear quite rapidly.


influence total demand or the supply and demand equilibrium for financial
funds only slightly.

It is not evident why one should want rates in

the money markets to fluctuate in response to their movements.
i!ost decision models and loss functions would, I believe,
show that beyond certain limits it is highly advantageous for the Govern­
ment to assume the risks from irregular movements.

The position of

these limits will depend at any time on the ability of the private
sector to assume such risks, on the shape of loss functions, on the
variance of movements and similar matters.