Full text of Economic Policy Review (Federal Reserve Bank of New York) : Summer 1986, Vol. 11, No. 2

View original document

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

The Quarterly Review is published by
the Research and Statistics Function
of the Federal Reserve Bank of New
\brk. Among the members of the staff
who contributed to this issue are
RICHARD G. DAVIS (on the recent
performance of the commercial banking
industry, page 1); LYNN PAQUETTE (on
estimating household debt service
payments, page 12); JOHN WENNINGER and
LAWRENCE J. RADECKI (on financial
transactions and the demand for M1,
page 24); and ALLEN J. PROCTOR (on
short-term borrowing by local school
districts, page 30).
Other staff members who contributed to
In Brief—Economic Capsules are ROBERT
N. McCAULEY (on large U.S. banks
moving international activity off their
balance sheets, page 42) and BRUCE
KASMAN (on prospects for the U.S.
international travel deficit, page 44).
A quarterly report on Treasury and
Federal Reserve foreign exchange
operations for the period February
through April 1986 starts on page 47.

The Recent Performance of the
Commercial Banking Industry

Banks are the primary institutions through which the
Federal Reserve conducts monetary policy. This is true
both in respect to the central bank’s efforts to influence
reserves, the money stock, and money market condi
tions, and in respect to its efforts to underpin the stable
functioning of the nation’s financial system. But banks
are businesses, and to perform their critical role in the
financial system and the policy process, banks must be
profitable. A number of developments in recent years—
some related to the overall economy and some specific
to the markets where banks operate— have had impor
tant effects on commercial bank profitability and the
performance of the industry generally. This article brings
together some of the main findings of a study under
taken at this Bank on the recent performance of the
commercial banking industry and the factors that have
influenced it. The aim of the study is to understand the
past rather than to attempt to foretell the future or to
prescribe for it. Nevertheless, an understanding of
commercial banking’s recent history is probably a nec
essary preliminary in any effort to assess its future
prospects or to evaluate policy options on matters that
affect it.

The recent slippage in bank profitability
Bank profitability overall has clearly declined in recent

This article is a slightly modified version of the concluding chapter
of Recent Trends in Commercial Bank Profitability—A Staff Study,
Federal Reserve Bank of New York (September 1986),
Documentation and a more detailed discussion of the points raised
in this article may be found in that volume.

years (chart). For all insured banks, return on assets
(ROA) declined in each year from 1980 to 1984 before
recovering modestly in 1985. Even with the 1985
improvement, ROA remained close to its lowest level of
the last 15 years. The recent performance of return on
equity (ROE) for all insured banks followed a similar
course, declining in each year from 1979 to 1984 wjth
a modest recovery in 1985 to a still-low level. By 1985,
ROE for all insured banks was about 11.4 percent,
substantially below the 13.9 percent 1979 peak and,
except for 1983 and 1984, the lowest since at least
1970.
Much of the study focused on a group of 17 multi
national bank holding companies, firms that hold some
37.9 percent of all bank assets and that are at the
center of some of the structural changes hitting the
banking industry. The profit performance of these 17
institutions has, of course, varied substantially from one
firm to another. But for this group of holding companies
as a whole, ROE slipped badly from its 14.4 percent
peak in 1979 and 1980 to a low of 5.4 percent in 1984,
recovering only to 11.2 percent in 1985, the lowest level
for any year since 1976. Even excluding the two holding
companies that lost money in both 1984 and 1985, ROE
in 1985 was still some two and one-half percentage
points below the 197 9-80 peak despite generally
improved profits for most of these holding companies
in 1985.
In contrast to the multinational holding companies and
insured banks as a group, the group of the 33 largest
regional holding companies examined in the study does

FRBNY Quarterly Review/Summer 1986

not show a recent downturn in profitability. For this
group, ROE held at a high plateau of about 14.6 percent
in each of the years 1979 through 1982. Profits did drop
somewhat relative to this level in 1983 and 1984, but
by 1985, ROE for the regionals as a group was essen
tially back to peak levels.
Overall, the 1980s have seen a slippage in profitability
for most groups of banks— although from relatively high

levels in terms of the postwar period as a whole.
Obviously there is a great deal of diversity in the profit
performance of individual banks and groups of banks.
And despite the overall decline, the profitability of
banking as a whole has shown greater stability than that
of other financial industries and of many nonfinancial
industries.
The market share performance of commercial banks

There has been some slippage in the overall role of
banks in the credit markets over the last ten years or
so. Moreover, there has been a marked decline in their
share of one particularly important credit market—the
wholesale loan market, as discussed below. But the
banks have been able to hold their own or even increase
their shares in most of the other markets they serve.
According to flow of funds data, the commercial
banks’ share of credit market debt claims against all
nonfinancial sectors of the economy peaked at just short
of 30 percent in 1974. Since then, this share has
declined virtually year-by-year to 25.2 percent in 1985.
However, the significance of this decline for the competi

FRBNY Quarterly Review/Summer 1986

tive position of the banks is hard to assess since the
overall decline represents a reduced share of holdings
of state and local and Federal debt. The banks’ share
of holdings of domestic nonfinancial private debt was
about unchanged on balance between 1974 and 1985.
Indeed, in most of the private credit markets where
banks compete, their share of the business has, as
suggested above, either held about steady or risen in
recent years. Thus the total share of home mortgages
(including allocated mortgage-backed securities) held by
commercial banks rose somewhat in the late 1970s and
has held fairly steady in recent years in the 25 to 27
percent range. The thrift institutions have been the
major losers in this market.
The banks’ share of commercial mortgages has also
risen over the last decade or so, from around 30 percent
to around 36 percent in 1984. Over the same period,
most of the other major participants in this market lost
share to the banks.
The banks’ share of consumer installment credit rose
over the 1970s but then suffered a significant decline
during the early 1980s due to a loss of share of auto
loans to captive finance companies. The banks’ share
of auto loans has been importantly weakened over the
last several years as the major domestic car manufac
turers have at times used reduced-rate financing as a
marketing tool. On the other hand, the banks’ share of
revolving consumer credit has risen almost continuously
since the mid-1970s as the use of bank credit cards has
spread at the expense of cards issued by retailers.
On the other side of the balance sheet, there are
various ways in which the market for deposits can be
sliced. One obvious way is with respect to the market
for “ transactions instruments.” The commercial banks’
share of deposits included in M1 (demand deposits,
negotiable order of withdrawal accounts (NOWs), and
Super NOWs) has of course declined somewhat in
recent years as the thrifts have entered this market, from
about 99 percent in 1978 to about 89 percent in 1985.
A second approach to defining the deposit market
might be in terms of the market for interest-bearing retail
deposit accounts. This market includes NOW accounts,
all savings and small time deposits, and money market
fund shares. The banks’ share of this total seems to
have suffered very little from the rise in the money funds
in the 1978-82 period. Instead, the thrifts were appar
ently the main losers in the rise of the money funds.
With the introduction of money market deposit accounts
(MMDAs) and Super NOW accounts around the end of
1982, the commercial banks’ share of this market has
subsequently risen a few percentage points, mainly at
the expense of the money funds. The share of the thrifts
in this market has continued to drift down, but more
gradually than in earlier years.

The banks’ share in the markets for various financial
services is somewhat harder to measure and to gen
eralize about. Owing primarily to the purchase of a
major discount broker by one of the multinational
holding companies, the banks as a group now appear
to have over half the revenue of discount brokerages
that are New York Stock Exchange members. On the
other hand, there has been a sharp fall in the banks’
share of pension fund assets under independent man
agement. In the underwriting field, banks are permitted
to underwrite municipal general obligation instruments,
municipal revenue bonds for housing and higher edu
cation, and private placements. These types of under
writing account for about 42 percent of all domestic
underwriting activity. Of these permitted activities, the
banks’ market share has fluctuated between roughly 12
and 20 percent in the 1979-84 period, with no discern
ible trend. Banks appear to have been advisors in only
about 2 percent of the 100 largest merger and acqui
sition deals of recent years. Data on smaller deals are
quite limited.

The decline of the wholesale loan market
O ne im p o rtan t m arket in which the b a n k s ’ role
undoubtedly has declined is the so-called “wholesale"
market. Precise docum entation of this decline is
impossible. For one thing, the concept of the wholesale
market is somewhat vague. As generally used, the term
means any large loan, whether to a national or multi
national corporation or to a foreign business or gov
ernment. However, “large” is not a precise concept. And
at times the wholesale market seems to refer primarily
to “good” large loans, i.e., to prime names. In any case,
there are no data designed specifically to measure this
market, so its dimensions have to be approached by
approximation.
The banks’ share of all nonfinancial business credit
has actually risen over the last decade or more, from
a little over 30 percent in the early 1970s to 35.2 per
cent in 1985. However, this rise obscures the banks’
weakness in the wholesale market, in part because it
reflects bank strength in commercial mortgage lending
and in part because of the steadily rising share of short
term business credit (in which the banks specialize)
relative to total business credit over the past ten years.
Since wholesale lending can be presumed to be con
centrated at the larger banks, one way to construct a
proxy for the decline in the banks’ share of this market
is to measure the sharp decline in the weekly reporting
member banks’ (WRMBs) share of all short-term non
financial business credit from over 43 percent in 1974
to about 27 percent last year. This is obviously a major
decline in the banks’ role. The slack seems to have
been taken up mainly by the commercial paper market

and by lending by branches and agencies of foreign
banks.
Because of statistical deficiencies, it is very difficult
to factor in precisely the role of the foreign banks in the
U.S. wholesale lending market, but it is obviously quite
large. We estimate that lending by all foreign offices of
foreign banks to U.S. commercial and industrial (C&l)
borrowers could represent as much as 39 percent of
such loans at WRMBs. Such limited data as we have,
however, do not really make clear whether there has
been any growth in the role of the foreign banks in our
wholesale loan market in the 1980s. (And, of course,
the role of our banks in foreign markets has also risen
over the years.)
The reasons offered by bankers and others for the
decline in wholesale lending seem to go to the heart of
the banks’ presumed comparative advantage as financial
intermediaries, at least with regard to this particular
market. Thus we are told that many prime corporate
borrowers have as good or better credit ratings than all
but a very few of the banks lending in this market.
Moreover, it is argued that for large borrowers, the
banks no longer have an informational advantage in
assessing the credit-worthiness of potential borrowers
in this market. Much of the relevant information is public
and readily available to any potential purchaser of the
borrower’s debt. Further, it is argued that the growing
importance of large pools of funds with sophisticated
management enables managers of these funds to
diversify credit risk as readily as can banks.
It is sometimes also argued that banks are at an
inherent disadvantage in acting as intermediaries
because, in addition to covering costs, they must also
price loans to realize an “adequate” return on capital.
This argument appears to be fallacious, however, since
the required return on bank equity can be thought of as
the needed reward for assuming credit and funding risk
over and above any purely “actuarial” component. The
“market” will also demand a reward for assuming such
risk. Whether the market will be willing to fund a given
credit at or below the level banks have to charge both
to cover costs and to achieve their “required” rate of
return on capital cannot be determined a priori but will
depend upon the circumstances. The point seems to be
that with respect to high-quality wholesale credits, at
least, the market has in fact increasingly been able to
outcompete the banks over the past decade or so.
The decline of the wholesale loan market has
obviously had profound implications for our largest
wholesale lenders, but a question arises about how
important it might be to the overall role of the banking
system in our financial markets.1 We can approximate

’As an indication of the relative heft of the large banks most heavily
FRBNY Quarterly Review/Summer 1986

Profitability and Price/Earnings Ratios in
Financial Service Industries

Industry
Commercial b a n k in g ....................
17 multinational bank holding
c o m p a n ie s .............................
Finance com panie s......................
Mortgage companies ..................
Securities ......................................
Investment banks ....................
Other s e c u ritie s ........................
Life in s u ra n c e ...............................
Stockholder-owned ..................
M u tu a l........................................
Property and casualty insurance .
Stockholder-owned ..................
M u tu a l........................................
Insurance brokerage
Large firms ...............................
Small firm s .................................
Diversified financial firms ...........
Nonfinancial firms (S&P 400)

Average
after-tax return
on equity,
1980-84

Average price/
earnings ratio,
1977-84

12.2

6.3

12.9
12.6
13.1
18.7
26.0
15.8
13.4
15.2
10.5
7.4
7.7
7.4

6.1
6.2

18.3
9.2
13.1
13.7

*
7.9
*
*
*
6.4
*
*
7.1

12.8
*
8.3
9.6

For notes and sources, see Recent Trends in Bank Profitability— A
Staff Study, Federal Reserve Bank of New York (September 1986),
Chapter 14.
*Not applicable.

an answer by looking at the share of short-term nonfinancial business credit represented by C&l loans at
WRMBs. As of December 1985, actual C&l loans at
these banks were $255.2 billion or 27.3 percent of all
short-term business credit. If the WRMBs had main
tained their 1974 share of 43.5 percent, C&l loans at
these banks would have amounted to $406.6 billion in
1985. With such a higher level of C&l loans, and given
the actual 1985 total of outstanding debt of the nonfinancial sectors ($7,114.7 billion), such a figure would
mean that the banks had held 27.3 percent of this credit
total, or 2.1 percentage points higher than the share
they actually did hold in 1985.
Such a figure suggests what was implicit earlier in
looking at the overall share of banks in credit markets
generally: namely that the decline of the wholesale
lending market has by no means been a body-blow to
the banking system as a whole. Nevertheless, this
decline should not be underestimated. It has been very
important to the largest banks, banks that in some ways
occupy a pivotal position in the banking system as a
Footnote 1, con tin u e d
involved in this market, the 17 m ultinationals had 37.9 percent of all
bank assets as of last D ecem ber and 50.2 percent of all C&l loans.
The C&l loans at these banks con stituted 10.1 pe rcent of total
assets of all com m ercial banks.

FRBNY Quarterly Review/Summer 1986

whole. Moreover, the 2 percent figure could substantially
understate the qualitative importance of the decline in
wholesale lending for the role of the banking system as
the backstop to our credit markets.
For one thing, many bankers argue that the increasing
competitiveness of the wholesale loan market has more
or less forced the wholesale banks to reduce the
average quality of their C&l book. Some bankers were
able to offer internal data that support this argument.
In addition, examination indicates that the weighted
average “ betas” (a standard measure of stock price
volatility) of the C&l loan portfolio of a sample of 47 of
the largest banks are higher than those for the Standard
and Poor’s (S&P) 500 and indicates a deterioration in
loan quality since the mid-1970s. Deterioration in the
quality of C&l loans seems to have been a significant
drag on the stock market performance of the shares of
both multinational and regional holding companies. The
deteriorating status of less developed country (LDC)
loans also seems to have been an important factor,
especially for the multinationals.
Profitability of banks relative to other financial firms

As might be expected, there has been considerable
diversity in the profit performance in the various non
bank sectors of the financial industry that in at least
some respects compete with banks. These include
finance companies, mortgage companies, investment
banks, securities brokers, the various components of the
insurance industry, and large diversified financial firms.
In comparing the profit performance of these groups
with that of commercial banks, only a few generaliza
tions can be made, and even then, only with qualifi
cations. First, with the exception of the non-auto finance
companies, the profitability of all segments of the non
bank financial industry has tended to move through
w ider— often much w ider— ranges during the past
decade than has bank profitability. And the profit per
formance of many of these groups shows substantially
greater short-term variability than does that of the banks
as a group. Second, virtually all the nonbank compo
nents of the financial industry have shared the banks’
experience of deteriorating profit performance since
peaks occurring around 1979-80—or at least this was
true through 1984, the latest year for which complete
data were available. The major exceptions to this recent
downward trend have been the consumer finance and
mortgage companies.
The average profitability of the financial industry in the
1980-84 period has varied considerably from sector to
sector (table). Over this period, average after-tax ROE
was 12.2 percent for all insured banks and 12.8 percent
for the multinational bank holding companies. Some
other parts of the financial industry did considerably

better than that, including large investment banks, other
securities firms, and large insurance com panies.
Average ROE in the period for finance companies,
mortgage companies, and diversified financial firms was
about the same as for banks or only a little better. A
few sectors, notably property and casualty insurers, did
substantially worse than the banks. On average, the
reported after-tax ROE of nonfinancial firms in the S&P
400 was somewhat higher than that of the banks at 13.7
percent.

Stock market treatment of banks and other financial
firms
For more than a decade, the stock market has priced
the earnings of large bank holding companies at mul
tiples well below those for corporate earnings in general.
This is true both for the regional holding companies and
the multinationals, but especially for the latter. As of late
April 1986, the price/earnings (P/E) ratio of the S&P 500
was about 15.7, while that of a group of multinational
banks was 10.2 (excluding banks with current losses)
and 12.2 for a group of large regional bank holding
companies.
Bankers and financial analysts offer a wide range of
explanations for the market’s relatively adverse treat
ment of bank stocks over the past decade, and there
seems no reason to insist on a single explanation to
cover so long a stretch of time. One possible sequence
of factors, for example, could be a sense early in the
period that banks were simply “stodgy” investments with
poorer earnings growth prospects than industry in gen
eral. Subsequently, as noted below, there is reason to
believe that as inflation heated up in the later 1970s,
reported bank earnings became seriously overstated.
Thus P/E ratios for earnings properly computed to take
account of the overstatement of earnings due to inflation
may have been a lot higher than P/E ratios computed
simply from earnings as reported. For a time in the early
1980s, a difficult interest rate environment, as discussed
below, may well have hurt bank stock performance. Most
recently, concern about the quality of bank assets, and
the LDC loan problem in particular, has probably been
an important depressant. Some believe that the
shrinking wholesale lending market has also been a
pervasive factor in the stock market’s assessment of the
earnings prospects, at least for money center banks,
and therefore in the low P/E ratios it has accorded most
of these banks relative to other firms.
In assessing these possible explanations for the per
sistently poor stock market evaluation of bank earnings,
it may also be worth keeping in mind that with some
notable exceptions, the market has also valued the
earnings of most other segments of the financial
industry at lower multiples than have been accorded to

nonfinancial firms on average (table). Over the 1977-84
period, P/E ratios for nonfinancial firms, as represented
by the S&P 400, averaged 9.6 as compared with 6.3 for
a sample of 90 commercial banks. To be sure, the
banks’ average P/E ratio was near the low end of the
range even for financial firms. But except for insurance
brokers, no nonbank financial group had P/E ratios that
averaged as high as the average for the nonfinancial
firms.
We do not have P/E data for the large investment
banks over comparably long periods of time since many
of them have gone public only recently. Based on data
for April 1986, the stocks of publicly-owned large
investment banking firms show multiples well above
those for the multinational banks, though still not
especially impressive relative to stocks in general. Thus
as of April, the average P/E ratio for fiv£ publicly-held
large investment banks averaged 14.9, compared with
the 10.2 average for the multinational bank holding
companies cited earlier and 15.7 for the S&P 500. A
number of bankers contacted in the course of the study
were quick to point out the contrast between high mul
tiples for investment banks and the much lower ones
characteristic of major money center banks. However,
some of the very high investment bank multiples they
cited in 1985 (as high as 25) appear in retrospect to
have been temporary spikes. Indeed these multiples
have apparently declined further since the April data
cited above.
Overall, the low P/E ratios accorded most large banks
relative to nonfinancial firms and even to many financial
firms, of course, mean that equity capital is relatively
expensive for these banks. The practical implication is
that it is relatively difficult for banks to find new projects
(or expansions of old projects) with yields high enough
to justify the injection of new capital, and this is, of
course, a deterrent to the longer run expansion of the
industry.

Macro versus structural causes of the deterioration
in bank profits
In examining the various possible causes of the recent
deterioration in bank profitability, it is useful to make a
distinction between causes related to the general eco
nomic environment (“macro” causes) and causes spe
cific to changing competitive conditions facing the
commercial banking industry (“structural” causes). Such
a distinction might at least provide a starting point for
trying to determine whether pressures on bank profit
ability are largely temporary or long-run in character.
Thus macro developments, such as movements in the
general level of interest rates, the rate of inflation, and
real economic growth, tend to be associated with the
business cycle and, therefore, tend to reverse them

FRBNY Quarterly Review/Summer 1986

selves in tim e. Structural developm ents such as
changes in an industry’s technology, relative costs,
demand, the extent of competition, and the nature of
regulation, clearly tend to be of longer-than-cyclical
duration and may be fairly long-lasting. To be sure, the
correspondence between macro and temporary on the
one hand, and structural and long-lasting on the other,
is far from perfect. Macro developments obviously can
exhibit trends as well as cycles and can thus be longlasting. Similarly, structural changes are not necessarily
irreversible.
In any case, it seemed desirable to attempt a formal
statistical analysis of the business cycle and trend
components of movements in bank profitability. Not
unexpectedly, regression equations using business cycle
variables (such as actual GNP relative to trend) do in
fact "explain” a statistically significant fraction of the
variance of bank profits. Nevertheless, there has been
a clear tendency for actual profits to fail short of “pre
dicted" profits in recent years. Taken as a group,
equations including both trend and cycle variables offer
some limited evidence that there has been a downtrend
in bank profits in recent years after allowing for the
influence of cyclical variables.
Of course, statistical evidence of a recent downtrend
in profits over and above cyclical influences does not
tell us which bank product lines may be responsible for
depressing profits and, more generally, does not dis
tinguish between a possible longer-than-cyclical per
sistence of macro and structural problems. Perhaps
most important, this evidence does not predict how long
any downtrend in profitability might persist.

Some identifiable macro influences on recent bank
profits performance
Whatever the relative roles of macro and structural
impacts on the recent decline in bank profits, some
effects of each kind can be clearly identified. Chrono
logically, the first macro problem to hit the banks over
the past decade was the sharply higher inflation rates
of the late 1970s peaking in 1980-81. By reducing the
real value of net worth, inflation means that book
earnings overstate earnings for financial institutions,
because, unlike industrial firms, they are large net
holders of fixed-dollar-value assets. In theory, inflation
need not also hurt inflation-adjusted bank earnings if
nominal interest rates correctly anticipate inflation so
that real after tax interest rates are unaffected by
inflation. But it seems reasonably clear that the markets
did not fully anticipate the extent of the acceleration in
inflation that occurred. Thus there is a strong probability
that the inflation of the late 1970s not only resulted in
profits being overstated but also that the inflation
actually reduced true profits. We made no attempt to

FRBNY Quarterly Review/Summer 1986

measure the extent to which bank profits were actually
hurt by inflation, but we did estimate the extent to which
they were overstated as a result of inflation. As Henry
Wallich found in an earlier study of this problem,2 the
overstatement appears to be sizable, and it alters the
appearance of the statistical record in some important
ways.
First, inflation substantially overstated the average
level of bank profits for most of the high inflation years
of the 1970s and early 1980s. Adjusted ROE averaged
only about 7 percent rather than the roughly 13 percent
average shown by the raw data.
Second, ROE at all insured banks appears to have
been near its lowest level of the decade in 1979 and
1980 on an adjusted basis, not at its highest, as the
reported data suggest. Adjusted ROE rose fairly sharply
in 1981 and 1982. ROE in 1984 (the last year for which
we have data on an adjusted basis) was close to the
highest levels of the last decade, not at the lowest as
the reported data suggest.
Third, adjustment of profits for the effects of inflation
substantially alters the appearance of bank profitability
relative to that of nonfinancial corporations. On a
reported basis, bank ROEs were roughly equal to
average ROEs for nonfinancials until the 1980-84 period
when reported ROEs of nonfinancials dropped very
sharply. Adjusting ROE measures of both groups for
inflation, however, puts banks’ ROEs below those of the
nonfinancials (and often far below) in all but one of the
last 12 years.
Finally, adjusting earnings for the effects of inflation
for both multinational banks and the S&P 500, P/E ratios
for the banks were actually above the average for the
500 in the 1977-79 period, were close to the 500 in
1981, and were once again above the 500 in 1984.
It seems entirely reasonable that a “rational,” “effi
cient” stock market should “look through” the veil of
inflation to price earnings on an inflation-adjusted basis.
The numbers are at least consistent with the view that
this is what happened. But it is somewhat curious that
none of the bankers or bank stock analysts contacted
in the course of the study included the distorting effects
of inflation among the possible causes of consistently
low reported bank P/E ratios. In any event, inflation
should now be a much less important influence on bank
P/E ratios than it may have been a few years ago,
perhaps replaced, as suggested earlier, by the issue of
credit quality.
A second macro factor adversely affecting bank profits
in the late 1970s through about late 1981 was the
interest rate environment. The general uptrend in rates
over this period had, at least in the short-term, an

2Wallich, Henry C., "Inflation is Destroying Bank Earnings and Capital
Adequacy," Bankers Magazine (Autumn 1977), pages 12-16.

adverse effect on profits because of a general tendency,
that apparently increased over the period, for banks to
have more short-term liabilities than short-term assets
(defining “short-term” in a repricing sense). The exis
tence of this short-term “repricing gap” made the flat
tening and frequent inversion of the yield curve in this
period a further drag on profits. Finally, the market cost
of bank funds relative to market lending rates (as
measured by the spread between certificates of deposit
(CDs) and commercial paper rates) also became sharply
more adverse over this period.
It should be noted that the repricing effect of a onceand-for-all increase (for example) in interest rates on
interest earnings can be measured precisely only with
a complete knowledge of the maturity distribution of
assets and liabilities. It is necessary to make do with
a crude index based on the excess of liabilities repriced
within a year or less over similarly short-term assets.
This appears to capture the direction of effects but can
provide only an index, not actual magnitudes. As a
conceptual matter, it should also be noted that the
earnings impact of a once-and-for-all rise in rates varies
with the time horizon. It seems to be clearly adverse in
the short run, but at some point it must turn positive.
In the long run, the effect must be positive given the
existence of some fixed-rate liabilities and equity in the
banks’ balance sheets. The effect on the present dis
counted value of the bank of a change in rates thus
depends both on the time profile of earnings effects and
on the discounting factor. In practice, the stock market
seems to mark bank stock prices up or down inversely
with interest rate movements.
While the various aspects of the interest rate envi
ronment were, as noted, adverse to profits through
about late 1981, they have subsequently reversed and
seem to have been positive on balance since then. The
impact of the interest rate environment seems to be
clearly a cyclical affair. There is no obvious reason to
believe that there has been a permanently adverse
change in this environment for the banks.
One influence on profitability that could be related in
part to macro and in part to structural developments is
the increase in provisions for loan loss reserves. In a
purely arithmetic sense, the increase in such provisions
relative to assets has more than accounted for the
deterioration of ROA at all banks and at all major com
ponent banking groups. But the interpretation of this
statistical fact presents some problems.
In an ideal world where banks were able correctly to
predict the overall rate of loan losses—though not which
individual loans would go sour— loan loss provisions
would be set to maintain loan loss reserves at a level
equal to expected future losses. If a deliberate decision
were made to increase the average riskiness of the loan

portfolio, loan loss provisions would be raised accord
ingly. Since the pricing of loans should, if market con
ditions permit, include an allowance to cover “normal”
loss experience, a rise in loan loss provisions relative
to assets need not by itself indicate any decline in
profitability. If the asset portfolio were to be shifted
toward loans with a higher expected loss rate, the
earnings figures should show higher net interest earn
ings after the portfolio shift, partially or fully offset by
a corresponding rise in the loan loss provision. ROA
(and ROE) need not be significantly changed on
balance.
But in practice, actual default experience need not
correspond at all closely to expectations. Many kinds of
lending do not readily accommodate themselves to an
“actuarial” analysis of risk. Moreover, average past
experience will fail as a guide to future default rates in
the face of adverse developments in the general eco
nomic situation facing borrowers— just as mortality
tables will understate mortality rates during an epi
demic. The data indicate that fluctuations in loan loss
provisions are in fact highly correlated with current
charge offs. This suggests that these fluctuations are
probably at least as much influenced by the failure of
actual default experience to conform to expectations
as they are by changes in the expected default rate
resulting from deliberate changes in portfolio
composition.
On balance, it seems reasonable to conclude that
much of the recent rise in loss provisions represents a
deterioration in the quality of credits that was not
anticipated in setting the levels of loan loss provisions
in earlier years. To this extent, current rising loan loss
provisions reflect unexpectedly adverse loan loss
experience rather than a deliberate change in portfolio
strategy toward loans both involving higher expected
losses and offsetting higher interest earnings. Since it
seems reasonable to expect that cyclical downturns
would be a major cause of unpleasant surprises in loan
loss experience, the overall cyclical influence on bank
profits, identifiable in the regression results cited above,
would manifest itself in the form of a corresponding
cyclical influence on charge offs and thus on loan loss
provisions. Such a pattern is in fact supported by the
evidence.
At the same time, there are other ways, not neatly
tied to cyclical movements in GNP, that general eco
nomic conditions could produce unexpectedly adverse
loan loss experience with correspondingly adverse
effects on profitability. Even though economic expansion
has now been underway in the United States and the
rest of the industrial world for about three and one-half
years, substantial economic slack has rem ained.
Moreover, there has been a persistent shift away from

FRBNY Quarterly Review/Summer 1986

the inflation in commodity and asset prices of the late
1970s and early 1980s to an atmosphere of stable or
declining prices. Further, over much of the period, both
nominal and real interest rates have been at high levels
relative to historical norms, and high levels of the dollar
have been a problem for some domestic borrowers.
Thus despite the resumption of overall economic
expansion, the shift in the economic climate after the
earlier inflation has left many groups of borrowers under
continuing pressure. These, of course, include the
LDCs, agriculture, natural resources (notably energy),
and some sectors of heavy industry. So while loan
losses in these areas may have persisted in the face
of an improving overall economy, they would neverthe
less appear to reflect macro-economic conditions, at
least in substantial part.
There is, however, a possible structural side to the
story of rising loan loss provisions and their relation to
profitability. The decline in the demand for bank credit
by highly rated borrowers may well, as already sug
gested, have left many banks, especially the largest,
with a deterioration in the quality of their loan book
beginning as long as a decade ago. If so, this essen
tially structural development would show up as rising
loan loss rates and as correspondingly higher loan loss
provisions. So the rising level of provisions could reflect
a structural development, at least in part.
As noted above, if higher expected risks were fully
priced in setting rates for these more risky customers,
rising loan loss provisions would not necessarily reflect
a corresponding drain on profitability. However, such a
drain would occur if, in a highly competitive market,
banks failed to charge adequately for the increased level
of prospective defaults.
Unfortunately, efforts to distinguish between macro
and structural causes of rising loan loss provisions and
their effects on profits are complicated by an inherent
“collinearity” between the two. The reason is that higher
loss rates arising from an increasingly risky portfolio are
likely to be exacerbated by deteriorating overall macroeconomic conditions. Ideally, one would like to be able
to determine whether actual loss experience has been
worse than might be expected given the general macroeconomic climate and, if so, why. However, there is no
neat way to do this. The evidence cited above does
indicate that charge offs and loan loss provisions have
been greater than expected, based solely on cyclical
factors. However, the fact that this is true for all size
classes of banks, and not just the wholesale banks,
leaves open the question of why it has occurred.

Structural factors
Perhaps the most direct evidence of a structural change
in the competitive conditions affecting bank profitability

FRBNY Quarterly Review/Summer 1986

would come from indications that the profitability of the
traditional deposit-taking and loan-funding role of banks
had declined at the expense of newer, essentially
fe e -b a s e d a c tiv itie s . To docum ent such a shift,
however, would require revenue and cost data by
type of activity. For individual banks, determining the
profitability of individual activities at a high level of
detail is a difficult problem, despite access to internally
generated data. At the level of publicly available data,
there exists no direct information on the profitability of
different banking activities, even over very broad classes
of activities. As a result, it has been necessary to make
estim ates which to some extent rest on arbitrary
assumptions.
The approach taken was to distinguish bank activities
that involve deposit-taking and the funding of interestearning assets— i.e., “intermediary” activities narrowly
defined— from all other, largely fee-based activities.
Given this distinction between the two classes of activ
ities, the problem is somehow to allocate revenues and
costs between them. On the revenue side, net interest
income can be ascribed to the intermediary category,
but non-interest income arises from both intermediary
and other activities in ways that have changed over time
as bank pricing practices have changed. Consequently,
it was necessary to make some alternative assumptions
about the allocation of non-interest revenues between
intermediary and other activities that seem to encom
pass the range of possibilities. On the expense side, the
only information available by activity comes from the
Federal Reserve’s Functional Cost Analysis data, and
for various reasons, these, too, present some problems
in allocating costs between intermediary and other
activities.
While the analysis necessarily leaves the answers to
many questions uncertain, some facts do stand out.
First, under any reasonable set of assumptions, reve
nues from intermediary activities at large multinational
banks have been rising much more slowly than reve
nues from other sources. Thus while revenues from
intermediary activities appear to have risen on the order
of 17 to 24 percent per dollar of total assets between
1980 and 1985, revenue from other sources per dollar
of assets more than doubled over the same period.
While revenues from these other activities constituted
only about 30 percent or less of intermediary revenues
at these banks in 1980, this fraction had risen to roughly
50 percent in 1985.
Not surprisingly, the non-interest expenses (salaries,
furniture, equipment, and occupancy and other operating
expenses) of these "other” activities have also risen
substantially more rapidly than intermediary-related, non
interest expenses. Growth of non-interest expenses was
about 35 percent for intermediary activities and around

75 to 95 percent for other activities over the 1980 to
1985 period.
Since revenue and cost allocations have to be com
bined to allocate profits between intermediary and other
activities, the uncertainties in revenue and cost allo
cations are compounded in making profitability esti
mates. Hence the range of uncertainty is correspond
ingly enlarged. Under a combined set of assumptions
most favorable to the estimated profitability of inter
mediary activities, profitability per dollar of assets
showed no trend between 1980 and 1983 and then
dropped by more than half in 1984 and 1985. Under this
same set of revenue and cost allocation assumptions,
the profitability of other activities (again scaled by total
assets) rose irregularly between 1980 and 1983 and then
rose sharply further in 1984 and 1985 to a level that was
larger than the profitability of intermediary activities. Using
a set of revenue and cost allocation assumptions unfavor
able to intermediation, intermediary activities actually lost
money in 1984 and 1985 while the profitability of “other”
activities rose steadily between 1980 and 1985.
Thus the results do depend significantly on the rev
enue and cost allocation assumptions. And given the
unusually sharp deterioration in intermediation profits
and the sharp rise in other profits in the final two years
of the period (1984 and 1985), it is difficult to draw hard
conclusions about trends in the relative profitability of
these two broad types of activities at the multinational
banks over the longer run.
Both the relative level and the trend of profits in the
intermediary activities were importantly influenced in the
1980-85 period by the sharp rise in loan loss provisions,
an item quite properly treated as a cost of intermediary
activities. Excluding the loan loss provisions, there is no
clear sign of a downtrend in the profitability of inter
mediary activity over this period. In fact, favorable
assumptions suggest an erratic but discernible upward
trend while unfavorable assumptions suggest that
intermediary profits before loan loss provisions have
been generally unchanged. So for the profitability of
intermediation, as for the profitability of banking overall,
much depends on the extent to which recent increases
in loan loss provisions prove to reflect a permanent rise
in the level of such provisions beyond levels priced into
loan spreads and the extent to which they prove to
reflect a merely temporary effect of unexpected adverse
economic conditions.
One point should be made in the face of the agnos
ticism forced on us by a strict adherence to what can
be demonstrated from available data. The intermediary
activities covered by the estimates involve the full range
of deposit-taking and funding operations undertaken by
these banks, not just the wholesale lending operations
where profitability is widely believed to have declined.

So the data in no sense conflict with this widely held
perception about the wholesale market.
Indeed the picture painted by the revenue data, at
least, is entirely compatible with the generally received
view about the wholesale lending market. But there is
no evidence to suggest that the profitability of other
kinds of bank lending activities (funding consumer credit,
home and business mortgages, the middle and small
business loan market) are declining and, indeed, none
of the industry experts contacted in the course of the
study suggested that they are. What may well be true,
however, is that the natural market for these kinds of
lending products (absent geographic expansion) falls
greatly short of the deposit-gathering capabilities of
large money center institutions that formerly used such
capabilities to fund wholesale lending.
One important structural development examined in the
study was the effect of deposit interest rate deregulation
on bank net interest margins— motivated in part by the
superficially surprising fact that such margins have
generally tended to rise over the years in which deposit
rate deregulation has taken place. In looking at this
problem, it is necessary to disentangle the adverse
effect on interest margins of deregulation’s impact on
relative rates, given the general level of interest rates,
from the favorable effects on net interest margin of a
rise in the general level of rates. As was suggested
earlier, the long-run favorable impact of a rise in rates
stems from the fact that as long as banks have some
fixed rate liabilities (e.g., demand deposits) and equity,
the long-run repricing effects of interest rate rises must
be to increase net interest earnings. For example, shifts
out of rate-regulated instruments such as demand
deposits into market rate instruments such as super
NOWs hurt interest rate margins. If at the same time,
however, interest rates generally are rising, the
remaining zero-rate demand deposits become more
valuable.
What the computations reveal is that rate deregu
lation by itself hurt net interest margins significantly
between 1977 and 1984, by about 0.37 percent of
assets at multinationals, by about 0.56 percent at
regionals, and by about 0.18 percent at other insured
banks. A look at the year-by-year impact suggests that
most of it was completed by the end of 1982 and that
there has been little if any net deterioration in margins
as a result of deregulation since then— at least for the
45 large bank holding companies we examined in
detail. Apparently the adverse potential effects of
MMDAs and Super NOWs was largely offset by an
associated reduction of reliance on large CDs at these
banks.
According to our estimates, the effect of the overall
rise in the level of interest rates from 1977 to 1984 was

FRBNY Quarterly Review/Summer 1986

to raise net interest margins at 42 large holding com
panies by 0.87 percent of assets and by 0.97 percent
of assets at other insured banks. In other words, the
rise-in-rates effect much more than offset the deregu
lation effect between 1977 and 1984, accounting for
much, but not all of the overall improvement in bank net
interest margins over this period. By implication, a return
of general rate levels to the 1977 levels would reveal
the unfavorable effects of deregulation otherwise hidden
by the general rise in rates through 1984.
If one concentrates solely on the effects of deregu
lation on net interest margin and ignores the non
interest income and expense implications of deregula
tion, the computed effects on profits seem to be large.
Thus, for example, before-tax ROE of the multinationals
averaged 15.8 percent in 1984. According to our com
putations, net interest margin was reduced by deregu
lation by 0.37 basis points, as noted above. Had it not
been so reduced, these before-tax profits would have
been a much larger 22.8 percent of equity in 1984. The
problem with trying to translate these large effects of
deregulation on net interest margin into profit terms,
however, is that such computations ignore offsetting
concomitant changes in income and expenses brought
about by the onset of interest rate deregulation and rate
competition. Thus explicit interest rate competition has
meant reduced nonrate competition (and thus reduced
non-interest expenses for branches, human tellers, etc.)
and increased non-interest revenues in the form of
explicit service charges. Hence the net effect on
profitability of rate deregulation must be materially less
than its effects on net interest margins alone would
suggest.

Implications of the study
The most obvious question raised by a study of bank
profitability is whether the longer-term position of the
industry is deteriorating. The question is sometimes put
somewhat differently: Is the profitability of traditional
banking drying up so that the role and function of banks
as we know them must undergo substantial change?
Answers to such questions can be based on an
assessment of what has happened or projections of
what will happen. The present study obviously bears
mainly on what has happened. The answers it gives
are not unambiguous. On one side, there appears to
be evidence that major upheavals in the macroeconomic climate have had, and are continuing to
have, a significant adverse effect on bank profita
bility. T h e re can be little doubt that a period of
stable, low -inflation economic expansion, during
which the credit problems generated by past eco
nomic upheavals can be worked through, would do
much to strengthen bank profitability.

FRBNY Quarterly Review/Summer 1986

On the other hand, some of the pressures on the
banks clearly would not go away, even under the most
favorable of macro-economic scenarios. Probably the
most critical issues raised by the structural develop
ments we have examined center on the decline of the
wholesale loan market. The figures indicate that the
decline in wholesale lending has already led to a
modest but significant slippage in the banking system
as a whole in the national credit markets. And the
development has raised acute strategic issues for many
of our largest banking institutions that have been most
heavily involved in this market. Their success in forging
successful new strategies in the face of the decline in
this traditional market has so far been mixed.
But perhaps more important than its impact to date,
the most interesting question raised by the decline in
wholesale lending is whether it may prove to be a par
adigm for the future transformation of other traditional
banking markets. The development that needs closest
attention in this connection is the spread, or potential
spread, of the process known as “securitization.”
Securitization, mainly in the form of a huge expansion
of the commercial paper market, made possible the
decline of the wholesale bank loan market. Securiti
zation has become a major factor in the mortgage
market (though the banks themselves are important
holders of mortgage-backed instruments), and securi
tization of numerous other types of loans, especially
consumer loans, appears to be in an active, if early
stage of development.
The question raised by the securitization process is
whether a range of developments— including a widening
ability of investors to assess credits and diversify risks
and innovations in technology that facilitate the pack
aging of securitized loans— are about to produce a
broad-based erosion in the profitability of intermediating
credit through the banks. Some analysts have already
come to the conclusion that such an erosion is indeed
in prospect. Some go further to argue that, as a con
sequence, the banking system will eventually evolve to
produce “banks” that are more like money funds,
offering transactions instruments on one side of the
balance sheet and holding essentially riskless money
market instruments on the asset side, with traditional
banking credits securitized out to the market or held by
nonbank institutions.
But the fact that an outcome is conceptually possible
does not mean it will materialize. Banks retain tremen
dous advantages as specialists in assessing and
diversifying credit risks and in funding them through an
array of highly attractive deposit instruments. The
examination of past developments we have conducted
does not tell us to what extent these advantages may
erode in the future. It does suggest, however, that out

side the wholesale lending market, there has been no
substantial slippage to date.
What is clear is that the profitability of the banking
system, and hence its continued ability to play its
present role in the credit markets and in monetary and

financial policy, cannot be taken for granted. The prof
itability of the banking system is likely to be a continuing
factor in the consideration of a wide range of policy
issues relating to banks and to financial institutions and
markets more generally.

Richard G. Davis

This article is excerpted from a 15-chapter book entitled Recent Trends in Commercial Bank Profitability—A Staff Study recently
completed by the Federal Reserve Bank of New York. For details, see the inside back cover.
— _— .— „--------------------- ;...— ~
— — —— — .......... ..................— ..-

FRBNY Quarterly Review/Summer 1986

Estimating Household Debt
Service Payments
Household debt rose very rapidly during the last several
years, with the ratio of debt to disposable personal
income reaching an all-time high of 0.74 in 1985-IV and
1986-1 (Table 1).1 This high level of household debt may
cause concern for at least two reasons. First, to the
extent that individuals may become subject to liquidity
constraints, a high level of debt may reduce future
consumer expenditures, aggregate demand, and real
economic activity. Second, a high level of debt may
increase consumers’ default rates and adversely affect
the soundness of the financial system. However, liquidity
constraints and default rates depend not only on the
level of debt outstanding but also on the level of debt
service payments.2 Because data on aggregate debt
service payments are not collected, this article estimates
a debt service payment series from 1975-1 through
1986-1.3
Aggregate debt service payments behaved quite dif
ferently from debt outstanding in recent years. Although

The author thanks James August, Paul Bennett, Christine Cumming,
A. Steven Englander, Howard Esaki, Ellen Evans, Edward Frydl,
Andrew Silver, and Charles Steindel for valuable comments and
suggestions, and Michael Weitz for excellent research assistance.
’Federal Reserve Board, Flow of Funds.
2Of course, neither the debt outstanding measure nor the debt
service payments measure takes into account such factors as
demographics, wealth holdings, or the distribution of debt among
. income groups, all of which would also affect the assessment of the
consumer debt burden.
•3See Goldman Sachs, Pocket Chartroom (November 1985), for
alternative estimates of debt service payments on consumer
installment debt.
12

FRBNY Quarterly Review/Summer 1986

the estimated ratio of home mortgage plus consumer
installment debt service payments to income was higher
in 1986-1 than in any of the previous ten years, it did
not rise as rapidly over the past decade as the debtto-income ratio. While home mortgage debt service
payments increased faster over the last decade, on
average, than home mortgage debt, consumer install
ment debt service payments increased much less than
consumer installment debt.
This article also analyzes why the debt and debt
service payments ratios grew at different rates by
examining how changes in loan extensions, maturities,
inflation, and interest rates affect each measure. Cycles
in the level of loan extensions generated the cyclical
pattern of both debt and debt service payments. How
ever, while the debt-to-income ratio surpassed previous
levels in 1985 and 1986-1, extensions of most types of
household loans were not higher, relative to income,
during the current expansion than during the 1975-79
expansion. Rather, longer maturities on consumer loans
and lower inflation rates contributed substantially to the
recent increase in the debt ratio.
Changes in maturities, inflation, and interest rates had
different effects on debt service payments. Longer
maturities on consumer loans decreased rather than
increased the debt service payments ratio. In general,
lower levels of inflation, if accompanied by lower nom
inal interest rates, affect the debt service payments ratio
much less than the debt ratio. However, interest rates
on consumer loans remained high in recent years, rel
ative to inflation, contributing to the growth of debt
service payments.

Estimating debt service payments

Data on the aggregate debt service payments due on
home mortgage and consumer debt outstanding are not
collected. Therefore, debt service payments of the
household sector from 1975-1 through 1986-1 were
estimated for home mortgages, which consist of mort
gages on one to four family homes, and the components
of consumer installment debt—automobile, mobile home,
revolving, and “ other consumer installment” credit. Debt
service payments were not estimated for consumer
noninstallment debt.4
The estimates of debt service payments are based on
past levels of extensions, debt, and average maturities
and interest rates at which loans were issued. In gen
eral, for each type of debt, the stream of debt service
payments due on loans issued during a given period is
calculated based on estimates of the amount of loans
extended during that period, and of the average interest
rates and maturities at which the loans were issued.
Prepayment rates are set so that quarterly changes in
debt are equal to extensions of new loans minus esti
mated repayments of principal. (For a detailed expla
nation of the methodology used, see the box.) In addi
tion, estimates of consumer installment debt outstanding
are adjusted to account for any precomputed finance
charges (that is, the interest component of the debt
service payments due on debt outstanding) that finance
companies may include in their reported levels of debt.
This adjustment has very little effect on the growth of
debt outstanding over the past decade, and reduces
consumer installment debt by 6 percent in 1986-1.

Table 1

Ratios of Home Mortgage and Consumer Debt
Outstanding to Disposable Personal Income,
1975-86*

Date
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986-1

.
.
.
.
.
.
.
.
.
.
.
.

Home
mortage
ratio

Consumer
installment
ratio

Other
consumer
credit ratio

Total

0.40
0.41
0.43
0.45
0.47
0.47
0.47
0.46
0.46
0.48
0.51
0.50

0.15
0.15
0.16
0.17
0.17
0.15
0.15
0.14
0.15
0.17
0.19
0.19

0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04
0.04

0.58
0.60
0.63
0.66
0.69
0.66
0.65
0.64
0.65
0.69
0.74
0.74

.
.
.
.
.
.
.
.
.
.
.
.

*Debt is as of the end of the year. Disposable personal income is for
the fourth quarter, at a seasonally adjusted annual rate. Data for
1986 are as of the first quarter, seasonally adjusted.
Source: Federal Reserve Board, Flows of Funds.

Chart 1

E s tim a te d Ratios of H om e M o rtg a g e
and C o n s u m e r In s ta llm e n t D e b t and
D e b t S e rv ic e P a y m e n ts to D is p o s a b le
P ersonal In c o m e

Results

Chart 1 depicts the estimated ratio of required debt
service payments on home mortgage and consumer
installment debt to disposable personal income from
1975-1 through 1986-1, along with the estimated ratio of
home mortgage and consumer installm ent debt to
income.5 In general, the estimated ratio of debt service
payments to income did not increase as much during
this period as the debt-to-income measure. While the
debt ratio was 22 percent higher in 1986-1 than in 1975-1,
the debt service payments ratio was only 8 percent
higher.
Charts 2 and 3 depict the two estimated ratios by type
of household debt (home mortgage versus consumer
installment). Chart 2 shows that debt service payments
in s ta llm e n t c re d it is c re d it sche duled to be repaid (or with the option
of repaym ent) in two or more installm ents. N oninstallm ent cre dit is
cre d it sche duled to be repaid in a lum p sum rather than through
p e riodic de bt service paym ents. Home m ortgage and consum er
installm ent de bt acco unt for 94 percent of the home m ortgage and
consum er de bt owed by households at the end of 1985.
E s tim a te d de bt service paym ents are also presented in Table B-2 of
the box.

0.150
0.145

i i i l m l i i i l m lil o.i40

0.52
1975 76

85 86

D ebt s e rv ic e paym ents and d is p o s a b le pe rso n a l
income are at annual rates. R atios include home
m ortgages ow ed by h o useho lds, personal trusts,
and n o n p ro fit o rg a n iza tio n s.
S ources: Federal R eserve Board, Flow of Funds
and Federal R eserve Bank of New Y ork s ta ff
e stim ates.

FRBNY Quarterly Review/Summer 1986

Methodology for Calculating the Debt Service Payments Ratio
To estimate debt service payments for each type of
consumer installment credit (excluding revolving credit),
loans issued in each quarter are assumed to have an
interest rate equal to the estimated average contract rate
on new loans issued in that quarter, and a maturity equal
to the estimated average maturity of new loans issued
in that quarter. Under these assumptions, the required
monthly debt service payment for loans issued in each
quarter is calculated. For each type of debt, prepayment
rates are set so that, at the end of each quarter begin
ning 1974-1V, the amount of debt outstanding implied by
the debt service payments calculations equals the
amount of debt outstanding as estimated by the Federal
Reserve Board in Statistical Release G.19.
For a given type of loan, let
E, = extensions during month i,
Di = debt outstanding at the end of month i,
rt = the interest rate on loans issued in month i,
m, = the maturity of loans issued in month i,
M, = the maximum (original) maturity on any loan
outstanding in month i,
Xl(t= debt service payment due in month t on loans
issued in month i (assuming no prepayments),
= E, * r, / 11 - (1 + r,) _m,]f
for i < t « i + mi and
= 0 otherwise.
Rtt= principal portion of debt service payment due
in month t on loans issued in month i
(assuming no prepayments),
= EiV < 1 + ri) " ‘-7 t(1 + rl)m> -1 ],
for i < t ss i + m( and
= 0 otherwise.
p, = prepayment rate on debt in month t,
Pw= percent of loans issued in month i that were
prepaid prior to month t,
t-1
= 1 - n
(1 — p,) for t > i + 1 and
j —i +1
= 0 otherwise.
Then aggregate debt service payments due in month t
equal
t- 1
2
X,,t * (1 - PM),
i = t-M ,
while aggregate required principal payments due in
month t equal
t- 1
T, = X
Riit* (1 - P,t).
i = t-M ,

FRBNY Quarterly Review/Summer 1986

Prepayment rates (p,) are calculated as follows. Let diit
equal the estimated amount of principal remaining at the
end of month t on loans issued in month i,
t
= E, - X
R i/O -P y)
j =i+ 1
t
- 2
P, * idy., - R ,/(1 -P M)1,
j = r+ 1
for t > i;
= E, for t = i.
That is, the amount of principal remaining at the end of
month t on loans issued in month i equals the original
amount issued minus the sum of required principal
payments made through month t minus the sum of pre
payments made through month t. Then, the amount of
total debt outstanding as of month t equals
t
or =
I
dM.
i = t - M,
Note that D? is a function of the prepayment rates p,,
j =s t. Prior to 1975, all values of p, are required to be
equal (i.e., p = p, for all j < 1975). Then p is set so
that, for t = December 1974, D, (actual debt) equals D^.
At the end of each quarter beginning 1975-1, the monthly
prepayment rate during that quarter is set so that
t
Dt - Dt_3 =
2
[Ej - T, - (D V, - Tj) *p,]f
i= t-2
where T( and D?. , are functions of pt, for i = t-2 ,t. That
is, the prepayment rate during each quarter is set so that
the actual change in debt equals the estimated value of
extensions during that quarter minus required principal
payments minus prepayments.
Estimating debt service payments using the method
described above requires data on extensions (Et), debt
(Dt), average (original) maturities (mt), and interest rates
(rt) for each type of loan. The principal data sources for
these series are listed in Table B-1. Unfortunately, data
are not available for each series for all time periods. In
particular, for mobile home loans, a maturity series is
available only for credit on new homes at finance com
panies, and only through 1982-IV. Maturities on all mobile
home loans are assumed to equal maturities on loans
made by finance companies, and after 1982, are
assumed to increase at the same rate as the maturities
on new car loans. (Between 1975 and 1982, average
maturities on new car loans rose 21 percent, compared
with 29 percent for new mobile home loans.) Also, the
average maturity on all mobile home loans is assumed
to equal 90 percent of the average maturity on new
mobile home loans, since data reported in the American

Methodology for Calculating the Debt Service Payments Ratio (continued)
Bankers Association’s Retail Bank Credit Reports indi
cates that maturities on used mobile home loans are
lower than on new home loans.
Data on average maturities on “ other consumer
installment” loans are very limited. Through 1982-IV,
finance companies reported the average maturity on
loans for “ other consumer goods” (excluding autos,
mobile homes, and recreational vehicles). This average
maturity increased 44 percent between 1975 and 1982.
While it corresponds to only a part of the “ other con
sumer installment” category, this maturity series does
suggest that average maturities on other consumer
loans, like maturities on auto and mobile home loans,
may have risen during the 1970s and early 1980s. On
the basis of these data, maturities on other consumer
installment loans are assumed to have increased at the
same rate as used car loans (roughly 31 percent
between 1975 and 1982).
Data on extensions of consumer installment credit after
1982 were collected only from finance companies.
Therefore, data on changes in outstanding debt are used
along with estimated repayments (including prepayments)
of principal, to estimate extensions of each type of debt

in each quarter after 1982. The prepayment rates after
1982 are assumed to equal their average values during
1982. That is,
Er = Dt - D,_, + Tt + p82*(D,_1 - Tt),
where E? equals estimated extensions of debt in month
t and p82 equals the average value of p, during 1982.
While the estimated prepayment rates varied somewhat
from quarter to quarter, they did not exhibit a trend over
the 1970-82 period, and the averages during 1982 did
not differ much from the averages during the preceding
years. Predicting post-1982 prepayment rates using a
more sophisticated moving-average process would not
change the debt service estimates considerably.
According to the Federal Reserve Board, finance
companies generally include the interest component of
debt service payments owed in their reported holdings
and extensions of debt. Therefore, in calculating debt
service payments on loans made by finance companies,
all data on extensions and debt outstanding that finance
companies report to the Federal Reserve Board are
assumed to include not only the principal but also the
interest component of the debt service payments owed.
Actual extensions by finance companies (i.e., excluding

Table B-1

Principal Data Sources For Debt Service Payments Calculations
Debt ou tstanding
— Federal Reserve Board, Flow o f Funds (home m ortgage debt).
— Federal Reserve Board, S tatistical Release G.19 (consum er installm ent debt by holder and by type).
Extensions
— U.S. D epartm ent of H ousing and Urban Developm ent, “ Survey of M ortgage Lending A ctivity,” M onthly M ortgage Loan Transactions
and Com m itm ents, 1970-79 (home m ortgage originations).
— Federal Reserve Board, S tatistical Release G.19 (consum er installm ent extensions through 1982).
— Federal R eserve Board, S tatistical Release G.20 (consum er installm ent extensions at finance com panies).
M aturities
— Federal R eserve Board, S tatistical Release E.4 (average m aturities on new auto loans at com m ercial banks through 1982).
— Federal Reserve Board, Statistical Releases E.4 and G.19 (average maturities on new car and used car loans at finance companies).
— Federal Reserve Board, Statistical Release E.10 (average maturities on mobile home loans and other consumer goods loans at finance
com panies through 1982).
— A m erican Bankers A ssociation, Retail Bank C redit Reports (distribution of m aturities on new car loans, most com mon maximum
m aturities on other consum er loans at com m ercial banks).
— Federal Home Loan Bank Board, "C onventional Home M ortgage R ates" (average m aturities on conventional new home and existing
home m ortgages).
Interest rates
— Federal R eserve Board, S tatistical Release G.19 (rates at com m ercial banks on new car, personal, and m obile home loans, and on
cre d it card plans. Rates at finance com panies on new car and used car loans).
— Federal Reserve Board, Statistical Release E.10 (rates at finance companies on mobile home and other consumer goods loans, through
1982).
— Federal Home Loan Bank Board, "C onventional Home M ortgage Rates” (average con tract interest rates on fixed-rate loans, percent
of loans that have ad ju stable rates).
Prepaym ent rates (hom e m ortgages)
— Helen F Peters, Scott M. Pinkus, and David J. Askin, "Prepayment Patterns of Conventional Mortgages: Experience from the Freddie
Mac P ortfolio," S econdary M ortgage M arkets (February 1984).
— M ortgage Security P repaym ent Rate Profile, Salomon Brothers Inc., various issues.
— Thomas N. Herzog and Dominick C. Stasufli, "Survivorship and Decrement Tables for HUD/FHA Home Mortgage Insurance Programs
as of D ecem ber 31, 1983," U.S. Departm ent of Housing and Urban D evelopm ent (M arch 1984).

FRBNY Quarterly Review/Summer 1986

Methodology for Calculating the Debt Service Payments Ratio (continued)
these precomputed finance charges) are calculated
based on the estimated average maturity and interest
rate at which loans were issued. That is,
x,' = EJ/m,,
where Xf is the monthly debt service payment on loans
issued by finance companies in month i, and E,' equals
extensions as reported by finance companies in month
i. Then “ actual” extensions by finance companies are
calculated as
E l = X<*[ 1 -

(1 + r t) " mi] / r.

Debt held by finance companies is adjusted as follows.
Let Z, equal the “actual” amount of debt held by finance
companies, and let D,' equal the amount of debt as
reported by finance companies. Then Z, is estimated as
D,' * k„ where kt equals the estimated ratio of principal
payments remaining, in month t, on loans issued by
finance companies, to remaining debt service payments.
In general, these adjustments have a very small effect
on the estimates of debt and debt service payments.
Since not all finance companies include precomputed
finance charges in their reported debt holdings, while
some other holders of debt may, debt and debt service
payments may be slightly understated or overstated.
The method used to calculate debt service payments
on home mortgage debt is very similar to that used to
calculate payments on consumer installment debt.
However, for home mortgage debt, prepayment rates are
varied by year of origination and by age of the loan,
based on average prepayment rates for Federal Housing
Administration mortgages and for loans in various mort
gage security pools. That is, the prepayment rate in
month t [year y(t)] for a loan issued in year y(i) equals
sy<i).y(t) * Qt> where sy(i)y(t) is the prepayment rate in year
y(t) for mortgages issued in year y(i), and is estimated
based on aggregate mortgage prepayment rate data. The
adjustment factor q, varies by quarter and is set so that
the estimate of debt outstanding implied by the debt
service payment calculations equals the estimate of debt
outstanding as reported in the Federal Reserve Board,
Flow of Funds. That is, the variable q, in the home
mortgage calculations is determined in a manner anal
ogous to that of p, in the consumer installment calcu
lations. The contract interest rate on adjustable rate
mortgages was reset annually using the one-year
Treasury bill rate.
For revolving credit, required principal payments in a
given month are assumed to equal a specified fraction
of the amount of debt outstanding at the end of the
previous month. For revolving credit held by banks and
savings institutions, required principal payments are
assumed to equal 5 percent of outstanding debt. For

FRBNY Quarterly Review/Summer 1986

credit held by retailers, this fraction is assumed to equal
8 percent, while for credit held by gasoline companies,
this fraction is assumed to equal 20 percent. These
assumptions are based on the required minimum pay
ment schedules of various bank, retail, and gasoline
company credit cards.

Table B-2

Estimated Home Mortgage and Consumer
Installment Debt Service Payments
In billions of dollars
Home
m ortgage

C onsum er
installm ent

Total

1 9 7 5 - 1 ....................... ........... 12.3
1975-11....................... ........... 12.6
1975-111 ............................... 12.9
1975-IV
........... 13.2
1976- 1
........... 13.6
1976-11....................... ........... 14.0
1976-111 ............................... 14.3
1976-IV
........... 14.8
1977- 1
........... 15.3
1977-11....................... ........... 15.8
1977-111 ............................... 16.4
........... 17.0
1977-IV
1978- 1
........... 17.5
1978-11....................... ........... 17.8
1978-111 ............................... 18.5
........... 19.4
1978-IV
1979- 1
........... 20.2
1979-11....................... ........... 21.1
1979-111 ............................... 22.0
1979-IV
...........23.0
1980- 1
........... 23.9
1980-11....................... ........... 24.8
1980-111 ............................... 25.6
1980-IV
........... 26.5
........... 27.4
1981- 1
1981-11....................... ........... 28.2
1981-111
........... 29.2
1981 -IV ............................... 30.0
...........30.8
1982- 1
1982-11....................... ........... 31.5
1982-111 ............................... 32.2
........... 32.6
1982-IV
1983- 1
........... 33.2
1983-11.................................. 33.8
1983-111 ............................... 34.7
1983-IV
...........35.9
1984- 1
........... 37.2
1984-11....................... ...........38.6
1984-111
........... 40.0
1984-IV
...........41.3
1985- 1
...........42.4
1985-11....................... ........... 43.3
1985-111 ...............................43.9
1985-IV
...........44.8
1986- 1
...........45.6

31.8
31.5
31.7
32.0
32.5
32.5
33.1
33.8
35.2
36.2
37.1
38.2
40.0
40.9
42.8
44.1
45.8
46.6
48.1
49.0
50.3
50.4
50.8
51.4
51.7
52.2
53.7
54.2
54.2
53.5
54.4
55.4
55.6
55.2
57.3
59.6
62.5
63.3
67.1
69.7
73.1
74.2
77.2
80.1
83 4

44.1
44.1
44.6
45.2
46.1
46.5
47.5
48.6
50.5
52.0
53.5
55.3
57.5
58.7
61.3
63.4
66.0
67.7
70.1
72.0
74.2
75.1
76.4
77.9
79.1
80.4
82.8
84.2
85.0
85.0
86.5
88.0
88.1
89.1
92.0
95.4
99.7
101.9
107.1
110.9
115.5
117.5
121.1
124.9
129.0

Quarter

on home mortgages increased steadily over the past
decade, and grew faster, on average, than home mort
gage debt outstanding, particularly during the 1980-82
period. Chart 3 illustrates that consumer installment debt
increased much more sharply, on average, than debt
service payments. While the consumer installment debt
ratio was 24 percent higher in 1986-1 than in 1975-1, the
debt service payments ratio was 4 percent lower. The
debt ratio was 11 percent higher in 1985 than in 1979
(its peak during the previous business cycle), while the
debt service payments ratio was 3 percent higher in
1986-1 than at its previous peak.
Factors that affect debt and debt service payments

Changes in a variety of factors— amounts of loans
issued, maturities of new loans, interest rates, and
inflation—caused the debt service payment and debt
ratios to grow at different rates over the past decade.
Understanding how these various factors affect debt and
debt service payments, both in the short and long run,
will aid in the assessment of consumers’ current debt
burden as well as in projecting how quickly debt and
debt service payments are likely to grow in the future.6
To illustrate the effects of these factors, examples are
presented based on the “ ty p ic a l” loan, where the
required monthly debt service payment is constant (in
nominal terms) over the loan maturity.7 That is, the
principal portion of the monthly debt service payment
increases over time as the interest portion decreases.
Effects of an increase in loan extensions
An increase in monthly consumer borrowing will lead,
in the long run, to proportionate increases in aggregate
debt service payments and debt outstanding, assuming
that interest rates and m aturities remain constant.
However, in the short run, debt will increase much more
quickly than debt service payments. For example,
assume that in each month prior to some month m0,
$100 of new loans is issued at a 36-month maturity and

C hart 2

Estim ated R a tio s of H o m e M o rtgag e
D e b t and D e bt S e rv ice P a y m e n ts to
D isposable P e rs o na l In c o m e
R atio
R atio
0 . 6 ------------------------------------------------------------------------------------- 0.08
Home m ortgage de bt
0 .5 ----------------------

■*------- Scale

------------------------- -

- o.07

0.4 l - V ------------------------------------------------------------- — 0.06

0 . 3 __________________Home m ortgage d e b t------------------- q .05
se rv ic e paym ents
S c a le ------- ►
0 . 2 -------------------------------------------------------------------------------------- 0.04

11 I I I I I I 1 1 1 . 1 1 L 1 I 1 1 1 I I I I 1 I I I I I I I I 1 I I I 1 I I I 1 I I I

1975 76

I I I p

0 3

85 86

D ebt se rvice paym ents and d is p o s a b le personal
incom e are at annual ra tes. Ratios includ e home
m o rtga ges ow ed by ho useho lds, personal tru sts,
and n o n p ro fit o rg a n iz a tio n s .
S ources: F e deral R eserve Board, Flow of Funds
and Federal R eserve Bank of New Y ork staff
estim ates.

C hart 3

E s tim a te d Ratios of C onsumer In s ta llm e n t
Debt and D e b t S e r v ic e Payments to
D is p o s a b le Personal Incom e
R atio
0.18

0.16
0.14

*The attributio n of changes in d e b t and de bt service paym ents to the
factors listed above does not take into account the effect that one
factor, such as interest rates, may have on another factor, such as
loan extensions. Nevertheless, this exercise is useful in explaining
why de bt and de bt service paym ents grow at different rates over
time. Explaining how inflation and interest rates affect borrow ing
p a ttern s is a sub je ct for future research.

0.12

0.10

0.08
7Most consum er loans are issued at a fixed interest rate for a given
m aturity and require a de bt service paym ent that is constant over
time. The main exce ptions are revolving cre dit, with required debt
service paym ents that may de clin e over time, and adjustable rate
loans, with de bt service paym ents that may vary over the life of the
loan in response to changes in interest rates. Most of the exam ples
presented below illustra ting the relation between de bt service
paym ents and debt ou tstand ing generalize to cases where the
interest rate varies over the life of the loan.

Debt service paym ents and d is p o s a b le personal incom e
are at annual rates.
Source: F ederal R eserve Bank o f New Y ork sta ff
estim ates.

FRBNY Quarterly Review/Summer 1986

an interest rate of 0.5 percent per month, with no pre
payments. Then prior to month mOJ the rate of amorti
zation implied by the loan terms will generate an
aggregate debt level of $1904, and aggregate debt
service payments of $109.50 per month. Now suppose
that beginning in month m0, the amount of new loans
issued doubles to $200 per month. In the long run {i.e.,
after 36 months have elapsed), both debt and debt
service payments will double as well. However, in the
short run, for example after only six months have
elapsed, debt service payments will have increased less
than 17 percent, while debt outstanding will have
increased over 33 percent (Table 2). That is, the ratio
of the remaining principal on loans issued in the past
six months to total debt is, in general, greater than the
ratio of debt service payments on loans issued in the
past six months to total debt service payments. Because
much of the original principal on the “ older” loans has
already been repaid, those loans account for a relatively
small portion of debt outstanding. The debt service
payment on the “ older” loans, however, remains con
stant over the maturity of the loan. Therefore, recently
issued loans account for a larger proportion of debt than
of debt service payments, causing changes in the rate
of borrowing to affect debt more quickly than debt
service payments.

Chart 4 presents the ratios of estimated extensions
of home mortgages and consumer installment loans to
income from 1975-1 to 1986-1.8 Extensions rose, relative
to income, during the 1975-79 expansion, declined
during the 1980-82 downturn, and then increased again.
The cycles in extensions account for the cycles in debt
and debt service payments. However, with the exception
of revolving credit, the estimated ratios of extensions to
income have not been higher, on average, in the current
expansion than in the 1975-79 expansion. While Chart
4 depicts gross rather than net extensions, it neverthe
less suggests that a trend in the rate of borrowing does
not, at least by itself, explain why the debt ratio is cur
rently at an all-time high.
Table 2 also demonstrates that if the amount individ
uals borrow each period increases, debt outstanding (as
well as debt service payments) will increase at a
decreasing rate over time. Thus, for example, even if
borrowing were to continue at the relatively high level
of the past year, the growth in the consumer installment
•Data on extensions of consum er installm ent loans were collecte d
after 1982 only from finance com panies. E xcept for revolving credit,
extensions of loans were estim ated after 1982 based on changes in
debt outstanding, estim ated required prin cip a l payments, and
estim ated prepaym ents (box). Extensions of revolving de bt were not
estim ated after 1982.

C h a rt 4

Table 2

E s tim a te d R a tio s of H o m e M o r t g a g e and
C o n s u m e r In s ta llm e n t Loan E x te n s io n s
to D is p o s a b le P e rs o n a l In c o m e

The Effects of Changes in Amount of New Loans
and Maturity on Aggregate Debt and Debt Service
Payments

R atio
0.14

In percent
C o n s u m e r in s ta llm e n t

Increase in loans Issued*

f \

r \ \\1 ji1

0.12
*
/
II
i

>* / \
/ ^
I

1 /A

¥\\
1/
i

0 .0 8

(e xclu d in g re v o lv in g )

H
I
i t
r\
V ,/W
\
\

0 .0 4

Home m o rtg a g e lA v
o ria in a tio n s
\

a * ' / \
Av
v' / V
» \* fl
/
'
,V */
\ / \
/

/
S

I 1 1 1 L. 1 L1 1 1_LL 11[„„l 1,1 1, 1 1.1 1,1 1 1 1 1 1 1 1 1 1 1 1J - l l .,1 i 11 1 1.1 i 1 1
1975 76 77 78
79
80
81 82 8 3
84 85 86

S o u rc e s : U.S. D epa rtm e nt of H ousing and U rban
D e ve lo p m e n t (hom e m ortgage o rig in a tio n s ); F e deral
R eserve B oard (c o n s u m e r in s ta llm e n t e xte nsions
th ro u g h 1982); and Fe deral R eserve Bank o f New Y o rk
s ta ff e s tim a te s (c o n s u m e r in s ta llm e n t e x te n s io n s
a fte r 1982).

FRBNY Quarterly Review/Summer 1986

Between
month 0
and month:
S ix .............
Twelve . . .
Eighteen . .
Twenty-four.
Thirty . . .
Thirty-six
Forty-two .
Forty-eight .

Increase in m aturityt

Change in
Change in
Change in
Change in
debt debt service
debt debt service
outstanding
payments outstanding
payments
33.9
57.7

33.3

16.7

76.3
89.7

50.0
66.7

97.6
100.0
100.0

83.3
100.0
100.0

100.0

0.8
2.9
6.4
11.4
17.8
25.7
31.9
33.7

3.7

- 7.6
-1 1 .4
-1 5 .2
-1 9 .0
- 22.8
- 10.0
2.9

•Prior to month 0, $100 of new loans is issued in each month, at a
maturity of 36 months and an interest rate of 0.5 percent per month.
Beginning in month 0, $200 of new loans is issued per month. No
loans are prepaid.
fP rior to month 0. $100 of new loans is issued in each month, at a
maturity of 36 months and an interest rate of 0.5 percent per month.
Beginning in month 0, new loans are issued at a maturity of 48
months. No loans are prepaid.

debt ratio (and eventually in the home mortgage debt
ratio) would slow considerably, assuming maturities
remain constant.
Effects of changes in maturities
Between 1975 and 1985, the maturities on certain types
of consumer loans increased substantially. For example,
the average maturity on new car loans issued by finance
companies increased from 38 months in 1975 to 52
months in 1985 (Table 3). This lengthening of maturities
may account for much of the difference between the
growth patterns of consumer installment debt and debt
service payments, since in the long run a change in
maturity has a much larger effect on debt than on debt
service payments. An increase in maturity at first
decreases aggregate scheduled repayments of principal,
thereby increasing aggregate debt outstanding. While
aggregate debt service payments decrease in the short
run, they may be slightly higher in the long run.
To illustrate this point, consider a situation where
issuances of new debt are constant over time, the
interest rate is also constant at 0.5 percent per month,
no prepayments occur, and the maturity at which debt
is issued equals 36 months prior to month mD and 48
months in month mQ and thereafter (Table 2). Aggregate
debt service payments at first decrease, because the
debt service payment on each loan issued at a maturity
of 48 months is less than the payment on each loan
issued at a maturity of 36 months. However, debt issued
at the longer maturity remains outstanding for a longer
period of time. Thirty-six months after the increase in
maturity, aggregate debt service payments begin to
increase, since debt issued 36 months ago is not yet
paid off. Forty-eight months after the increase in matu
rity, aggregate debt service payments have reached
their new long-run level, and are slightly higher than
aggregate debt service payments in month m0, reflecting
the fact that interest is owed on a larger amount of debt.
In the more general case where prepayments occur, a
change in maturity may have a different effect on debt
and debt service payments than in the “ no prepay
m ents” example, depending on how prepayments
change. However, as in the example presented above,
the effect on debt will be very different from the effect
on debt service payments.
To determine how much of the divergence between
the consumer debt and debt service payment series has
been due to increases in the maturities of consumer
loans, the debt and debt service payments ratios were
reestimated, assuming that maturities did not change
after 1974.9 That is, household debt and debt service
*Data on m aturities of m obile home and "o th e r consum er installm ent”
loans are lim ited, and a num ber of assum ptions were made in
con structin g the average m aturity series (box).

Table 3

Average Maturities on Auto Loans Issued by
Finance Companies
In months
Year

New car loans

Used car loans

..................... 38
1975
1976
..................... 39
1977
......................41
1978
..................... 43
1979
......................44
1980
......................45
1981
......................45
1982
......................46
1983
......................46
1984
..................... 48
1985
......................52
1 9 8 6 * .............................................. ......................51

29
30
31
33
34
35
36
37
38
40
41
43

’ January through May.
Source: Federal Reserve Board, Statistical Release G.19.

payments were estimated under the assumptions that
the amounts borrowed, and interest and prepayment
rates, were equal to their actual levels, but that after
1974, maturities were equal to their 1975-1 levels. If
maturities had not increased, the consumer installment
debt ratio would have risen only 8 percent between
1975-1 and 1986-1 rather than 24 percent (Chart 5).
Furthermore, the debt service payments ratio would
have increased 3 percent, instead of falling 4 percent.10
Thus, increases in m aturities on consum er loans
account for much of the difference between changes in
the consumer installment debt ratio and changes in the
debt service payments ratio over the past decade.
The growth in maturities also explains, to some
extent, why the consumer installment debt ratio was
higher in 1985 and 1986-1 than during the 1975-79
expansion, while extensions of most types of consumer
loans relative to income are estimated to be lower. If
maturities had not increased after 1974, the consumer
installment debt ratio would have increased only 2 per
cent between 1979 and 1985 rather than 11 percent,
while the ratio excluding revolving credit would have
decreased 7 percent rather than increased 5 percent.
Changes in maturity account for virtually none of the
growth in home mortgage debt or debt service payments
between 1975-1 and 1986-1, partly because maturities on
home mortgages did not exhibit a pronounced trend
10By assuming that prepaym ent rates remain the same under the
shorter maturities, the amounts prepaid decrease, since less debt is
outstanding at a given time. As a result, aggregate debt service
payments are slightly higher (rather than slightly lower, as in the
example given above) under the shorter maturity.

FRBNY Quarterly Review/Summer 1986

C h a rt 5

E s tim ated Ratios of C o n s u m e r In stallm ent
D e b t and D e b t S e rv ic e P a y m en ts to
D is p os ab le P ersonal In c o m e
Ratio
0.18

0.16

0.14

0.12

0.10

0 .0 8
1975

8 5 86

Debt se rv ic e paym ents and d is p o s a b le p e rsona l incom e
are at annual ra tes.
S o u rce : F e deral R e s e rv e Bank of New Y ork s ta ff
e stim a te s .

during this period. Furthermore, since home mortgages
are issued at long maturities, any change in the maturity
on new loans will affect debt outstanding very slowly.
Table 2 illustrates that the response of debt and debt
service payments to an increase in maturity is gradual.
As a result, recent increases in maturities will continue
to affect the growth of consumer installment debt and
debt service payments over the next few years. For
example, if borrowing relative to income continues at the
level of the past year through 1987-IV, and maturities
remain the same, the consumer installment debt ratio
will increase 4 percent between 1986-1 and 1987-IV.
However, if maturities had not increased after 1974, the
debt ratio would increase less than 1 percent.11
Effects of changes in inflation and nominal interest rates
Large fluctuations in inflation and nominal interest rates
over the past decade also affected debt and debt
service payments differently (Table 4). An increase in
inflation— holding real interest rates, maturities, and
am ounts borrowed (in real term s) constant, and
11For this exam ple, interest rates on consum er loans were assum ed to
fall roughly 100 basis points betw een 1986-1 and 1987-IV.
P repaym ent rates were assum ed to remain at their average levels
over the last fou r quarters, w hile incom e was assumed to grow
roughly 6 pe rcent per year.

FRBNV Quarterly Review/Summer 1986

assuming no prepayments—decreases the real values
of both debt and debt service payments in the long run.
However, if nominal interest rates rise with inflation, real
debt service payments decline much less than real
debt.12
To illustrate this point, suppose that prior to month mD
the inflation rate is zero, and beginning in month mD it
increases to 0.5 percent per month (Table 5). The real
interest rate is set at 0.5 percent per month and the
maturity at 20 years. The amount borrowed is constant
in real terms and no prepayments occur. In the long run,
debt will decrease 25 percent, in real terms, while the
real value of aggregate debt service payments will
decrease only 10 percent. If the interest rate is adjust
able rather than fixed, real debt service payments will
rise considerably in the short run, before declining. In
either case, the change in the rate of inflation affects
real debt and debt service payments quite differently.
To demonstrate how the level of inflation has affected
the debt and debt service payments ratios over the past
decade, the ratios were reestimated assuming that in
every year after 1974, the inflation rate was 2 percent
lower than its actual value. Prepayment rates, and the
real values of interest rates, the amounts borrowed, and
income were left unchanged.13 Under this “ lower infla
tion” scenario, the growth of the home mortgage debt
ratio between 1975-1 and 1986-1 increases from 21 to
32 percent, but the growth of the debt service payments
ratio decreases from 36 to 32 percent.14 Thus, the high
level of inflation experienced during the past decade
slowed the growth of the debt ratio, while raising the
debt service payments ratio.
A change in the long-term inflation rate affects the
consumer installment ratios much less than the home
mortgage ratios, because consumer loans have shorter
maturities. If inflation and nominal interest rates had
been 2 percent lower after 1974, the consumer install
ment debt ratio (excluding revolving credit) would have
12H olding the real interest rate constant, an increase in inflation will
reduce the long-run level of real debt service paym ents by
increasing the rate at w hich real paym ents on the de bt are m ade.
That is, with no inflation, the required debt service paym ent on a
given loan is constant in real term s over the life of the loan. With
inflation, the real value of an in divid ual's m onthly de bt service
payment decreases over time, since the nominal value is constant.
Thus, the real value of the loan is paid off more qu ickly in the latter
case, and therefore, the long-run level of ag gregate de bt service
paym ents is lower in real term s.
13Nominal interest rates on loans issued after 1974 were reduced by 2
percent. Ex p o st real interest rates on loans issued before 1975 are
increased by this simulation.
14By assum ing that prepaym ent rates remain the same under the
"low e r inflatio n" scenario, the am ounts prepaid (in real term s)
increase, since more debt is outstanding in real term s. Therefore,
the decrease in inflation lowers the debt service paym ents ratio
slightly, rather than increasing it slightly as in the exam ple presented
above.

Table 4

Interest Rates on Home Mortgages and Consumer Loans,
and the Rate of Growth of the Consumer Price Index (CPI), 1975-86
In percent
Interest rates*
Home
m ortgagesf

Year

Auto loans at
commercial banks}:

Personal loans at
commercial banks§

Bank
credit cards

Rate of growth
of the CPI//

1975 ....................... ....................

11.4

13.1

1976 ....................... ....................

8.8

11.1

13.0

17.2
17.1

4.9
6.8

1977 .......................

7.0

....................

8.8

10.9

13.0

16.9

1978 ....................... ....................

9.3

11.0

13.2

17.0

9.0

1979 ....................... ....................

10.5

12.0

13.9

17.0

1980 ....................... ....................
1 9 8 1 ....................... ....................

12.3

14.3

15.5

14.2

16.5

18.1

17.3
17.8

13.2
12.4

1982 ....................... ....................

14.5

16.8

18.6

18.5

1983 .......................
1984 .......................

....................

12.2

18.8

3.8

11.9

13.9
13.7

16.7

....................

16.5

18.8

4.0

8.9
3.9

1985 ....................... ....................

11.1

12.9

15.9

18.7

3.7

.................... ....................

10.4

11.9

15.2

18.4

- 0.2

19861

‘ Rates are annual averages of monthly data. Except for home mortgages, data are for midmonth of quarter only.
fContract rate on fixed-rate mortgages for new homes.
^Before 1983 the maturity for new car loans is 36 months. Beginning in 1983 it is 48 months.
§Loans with maturities of 24 months.
//From December to December. Rate for 1986 is from December to June, annualized.
fThrough the second quarter.

Table 5

The Effects of Changes in Interest and Inflation Rates on Aggregate Debt and Debt Service Payments
In percent
Increase in the interest rate*

Increase in the inflation ratef
Percent change in
debt outstanding

Between month 0
and month:

Percent change in
debt outstanding

Percent change in
debt service
payments

S i x ........................................

0.0

1.4

T w e lv e .................................
T w e n ty -fo u r.......................
T h irty-six..............................
F o rty-e ig h t..........................
S ix ty ....................................
One-hundred twenty . . . .
One-hundred eighty . . . .
Two-hundred f o r t y .............

0.1
0.2
0.5
0.9
1.5
5.6
11.0
14.2

2.7
5.4
8.1
10.8
13.4
26.9
40.3
53.8

Fixed rate
loans
-

2.9

- 5.5
-1 0 .1
-1 4 .0
-1 7 .1
-1 9 .7
-2 5 .8
-2 5 .8
-2 5 .0

Adjustable
rate loans
-

1.9

- 3.8
- 7.2
-1 0 .2
-1 2 .8
-1 5 .0
-2 2 .2
-2 4 .6
-2 5 .0

Percent change in
debt service payments

Fixed rate
loans
-

Adjustable
rate loans

1.6

26.1

- 3.0
- 5.6
- 7.9
- 9.7
-1 1 .2
-1 4 .7
—13.8
-1 0 .4

23.7
19.4
15.4
11.8
8.6
- 3.1
- 8.8
-1 0 .4

'Prior to month 0, $100 of new loans is issued in each month, at a maturity of 240 months (20 years) and an interest rate of 0.5 percent per month.
Beginning in month 0, the interest rate on new loans doubles to 1 percent per month. No loans are prepaid.
fP rior to month 0, $100 of new loans is issued in each month, at a maturity of 240 months and an interest rate of 0.5 percent per month. Beginning in
month 0, the rate of inflation increases from 0 to 0.5 percent per month. The real value of new loans issued and the real interest rate on new loans
remain the same. Debt outstanding and debt service payments are measured in real terms. No loans are prepaid.

FRBNY Quarterly Review/Summer 1986

Table 6
Interest Rates on Consumer Loans Minus
Increases in the Consumer Price Index (CPI),
1975-86
In percent
Interest rate minus average annual increase in CPI*
Year
1975
1976
1977
1978
1979
1980
1981
1982
1983
1984
1985
1986

Auto loansf
.
.
.
.
.
.
.
.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

. .
. .
. .
. .
. .

. .
. .

5.3
3.8
1.0
- 1.0
0.6
5.8
10.6
13.2
11.1
10.3
10.2
8.9

Personal loanst

Bank credit cards§

7.1
6.8
4.8
0.9
0.8
5.2
11.6
14.7
13.5
12.8
13.5
12.4

10.9
11.2
10.5
7.0
2.9
5.8
9.9
14.7
14.3
15.2
15.7
16.2

'Interest rates are as of February. CPI increase is from February to
February.
fN ew auto loans at commercial banks. Prior to 1983, maturity on loan
equals 36 months. Beginning in 1983, it equals 48 months. Increase
in CPI is averaged over first three years of the loan.
^Twenty-four month personal loans at commercial banks. Increase in
CPI is averaged over the two-year loan period.
§lnterest rate on credit cards at commercial banks minus increase in
CPI over the first year of the loan.

increased an additional 1 percent between 1975-1 and
1986-1. The corresponding debt service payments ratio
would have been virtually unaffected by the change in
inflation.
Fluctuations in the rate of inflation also explain why
the home mortgage debt ratio is currently at an all-time
high even though mortgage originations relative to
income are not. Although home mortgage originations
have been lower relative to income during the current
expansion than during the 1975-79 expansion, the rate
of inflation has been much lower as well. If, for example,
inflation had been constant at 6 percent after 1975, with
actual borrowing patterns left unchanged in real terms,
the home mortgage debt ratio would have been 3 per
cent lower in 1985-IV than at its previous cyclical peak,
rather than 6 percent higher.15 Therefore, the high levels
of borrowing during the late 1970s may have been, in
part, a response to high levels of inflation, but the high
inflation rates offset the impact of the borrowing on the
debt ratio.
1sln this sim ulation, the spreads between interest rates and inflation
w ere held con stant at the ir 1976 levels, w hile prepaym ent rates were
kept at the ir actual values.

FRBNY Quarterly Review/Summer 1986

Effects of changes in real interest rates
Some of the differences between the debt service pay
ments and debt outstanding series may be attributed to
the effects of changes in real interest rates, i.e.,
movements in nominal interest rates independent of
movements in inflation or expectations of inflation. An
increase in the interest rate at which debt is issued,
holding the maturity, amounts borrowed, and inflation
rate constant, will increase debt service payments as
well as debt outstanding. Debt outstanding will increase
because, at the higher interest rate, principal is paid
back more slowly. That is, the higher the interest rate,
the lower the first required principal payment on a given
loan, and the higher the final principal payment. How
ever, the effect of a change in interest rates on debt
service payments is much larger than the effect on debt
outstanding. For example, an increase in the interest
rate from 0.5 to 1 percent per month— holding the
maturity constant at 20 years, the amounts borrowed
each period constant, and assuming no prepayments—
will increase the debt service payments ratio 54 percent,
and the debt outstanding ratio 14 percent, in the long
run (Table 5).
Determining how real interest rates on new loans have
changed over the 1975-86 period is problematic, even
on an ex post basis, since the real interest rate on a
given loan depends on when the loan is repaid. How
ever, very rough proxies for ex post real interest rates
on certain types of consumer loans were calculated by
subtracting the average increase in the Consumer Price
Index (CPI) over the approximate term of the loan from
the interest rate on the loan.16 The results of these
calculations suggest that real interest rates on consumer
loans have been rising in recent years (Table 6). These
relatively high rates have contributed to the recent rise
in the debt service payments ratio.
Interest rates have a much larger impact on the home
mortgage debt service payments ratio than on the
consumer installm ent debt service payments ratio.
Because home mortgages have longer maturities, prin
cipal repayments make up a small part of mortgage debt
service, and interest payments make up a large part.
Lowering interest rates on all loans issued after 1974
by 200 basis points would reduce the home mortgage
debt service payments ratio by 13 percent in 1986-1,
while decreasing the consumer installment ratio by only
3 percent.
Effect of the tax system on debt service payments
Interest payments accounted for an estimated 45 per
cent of aggregate debt service payments in 1986-1,
compared with 31 percent in 1975-1. This growth reflects
16The inflation rate was assum ed to equal 3.2 percent (annualized)
during the second half of 1986 and 4 percent during 1987.

increases in nominal interest rates as well as maturities.
Because interest payments are tax deductible, the
increase in debt service payments overstates the
increase in households’ debt burden. For example, if all
households with debt itemized their deductions and
faced a marginal income tax rate of 25 percent, the debt
service payments ratio on a post-tax basis would have
increased 3 percent between 1975-1 and 1986-1 rather
than 8 percent.
The effects of changes in maturity, inflation, and
interest rates are different on a post-tax basis. A
lengthening of maturity increases the proportion of
aggregate debt service payments that represent interest,
as does an increase in inflation and nominal interest
rates. Therefore, in the examples given previously,
assuming no prepayments, an increase in maturity
would result in a larger percentage decrease in post
tax debt service payments than in pre-tax payments in
the short run, and, in the long run, would lead to a
smaller percentage increase. Similarly, the percentage
decrease in real debt service payments that results from
an increase in inflation and nominal interest rates would
be even larger on a post-tax than on a pre-tax basis,
while the short-run increase in real payments on
adjustable rate debt would be less.

Conclusion
Because changes in maturities, inflation, and interest
rates affected debt and debt service payments differ
ently, the household debt service payments ratio grew
much less quickly than the debt ratio over the past

decade. Increases in maturities on consumer loans kept
debt service payments relatively low, and at the same
time increased debt outstanding. Inflation increased the
divergence between the home mortgage debt and debt
service payments ratios, causing debt service payments
to grow faster than debt.
The debt and debt service payments ratios may
change quite differently in the future as well. For
example, an increase in nominal interest rates and
inflation would reduce real debt but increase real debt
service payments on adjustable-rate debt in the short
run, assuming real borrowing remains constant. If, on
the other hand, inflation remains at a low rate, and
interest rates keep declining, the debt ratio will probably
continue to grow much more quickly than the debt
service payments ratio.
The recent rise in the debt ratio to an all-time high
is not simply due to unusually high levels of new loan
extensions relative to income. Rather, longer maturities
and lower inflation have contributed substantially to the
increase. During the 1970s, consumers may have
demanded longer maturities partly to offset the impact
of higher inflation and nominal interest rates, since
these factors increase the rate at which debt is amor
tized, in real terms. In recent years, consumers may
have used long maturities partly to offset the impact of
high real interest rates on monthly debt service pay
ments. In the future, if consumers choose shorter
maturities in response to lower inflation and interest
rates, the debt ratio may decrease as well, with the debt
service payments ratio increasing slightly.

Lynn Paquette

FRBNY Quarterly Review/Summer 1986

Financial transactions and the
Demand for Ml
Over the past few years, trading volume in the financial
markets has increased at a very rapid rate.1 At the same
time, M1’s growth has been considerably stronger than
would have been expected given the performance of
GNP, creating doubt about the adequacy of GNP as a
measure of the total dollar volume of transactions. In
other words, GNP may understate the overall transac
tions demand for M1 when trading in financial instru
ments is increasing at a considerably faster rate than
GNP. If this is the case, some economists suggest
using the dollar volume of debits to checking accounts
to approximate the transactions demand for money
instead of GNP, b ecause debits to checking accounts
occur for all types of transactions, financial and nonfinancial, and not just for sales of goods and services
to final purchasers.2
To evaluate whether financial trading has significantly
increased the demand for M1 by causing debits to
checking accounts to grow much more rapidly than GNP,
two links should be established. First, a relationship
needs to be found between the volume of financial
transactions and debits to transactions accounts; and

1This has been true not only for the established stock and bond
markets, but also in relatively new markets such as options, swaps,
and futures. For more detail, see the 1985 Annual Report, Federal
Reserve Bank of New York, page 18, and “ Demystifying Money’s
Explosive Growth,” Morgan Economic Quarterly (March 1986),
pages 10-13.
2See, for example, John Wenninger, “ Reserves Against Debits," this
Quarterly Review (Winter 1982-83). Also see, Ralph C. Kimball,
“ Wire Transfers and the Demand for Money,” New England Economic
Review, Federal Reserve Bank of Boston (March-April 1980); Charles
Lieberman, "The Transactions Demand for Money and Technological
Change,” Review of Economics and Statistics (August 1977); and
Alexander J. Field, “Asset Exchanges and the Demand for Money,
1919-29,” American Economic Review (March 1984).
24

FRBNY Quarterly Review/Summer 1986

second, debits should be a better proxy than GNP for
the transactions that affect money demand. If these
points cannot be established, then it is somewhat less
clear that financial transactions are affecting the growth
of M1 to a large degree.
In this article, we examine the data available on these
two linkages. By and large, there has been little quarterto-quarter correlation between trading volume and debits
to checking accounts. In addition, the data suggest that
in the longer run, debits are not a significantly better
measure than GNP of those transactions that matter for
money demand. Finally, even in the case of 1985, when
debits did track M1 growth b e tte r than GNP, it is
uncertain whether financial transactions were the pri
mary reason debits predicted M1 growth more accu
rately. The growth rates of debits and GNP can diverge
for reasons other than financial transactions.

Trading volume, debits, and GNP
Chart 1 shows the explosive growth of two readily
available data series that are sometimes taken as indi
cating the general growth of financial transactions: the
dollar volume of transactions on the New York Stock
Exchange and the dollar volume of trading by dealers
in U.S. Government securities.3 Mirroring this explosive
growth in financial trading volume has been the growth
of debits to checking accounts. And the growth of M1
has generally been faster than expected since the early
1980s, when financial transactions and debits began to

3The trends in the trading volume of these segments of the financial
markets, of course, may or may not parallel the growth in the
volume of financial transactions in all markets—but the data on other
financial transactions are rather limited and analysts have been
forced to use these two series as an indication of what is happening
more generally.

accelerate sharply relative to GNP. At first glance, the
similarity in these longer-run trends implies that the
more rapid growth of trading volume in financial instru
ments relative to GNP has contributed to more rapid
growth in debits, and this in turn has increased the
demand for M1. In theory, of course, an increase in
financial transactions should add to the demand for M1,
all other factors equal; but how important are financial
transactions in practice?
Charts 2 and 3 show that financial transactions may
not explain much of M1’s growth. Chart 2 compares the
long-run trend in the velocity of M1 measured two ways,
using debits and GNR Since the late 1950s, velocity
measured with debits has increased by a factor of 11,
whereas velocity measured with GNP is only about two

C hart 1

Growth of Financial Transactions,
Debits, and M1
Index (1971-1=100)
800

times greater. Indeed, the extremely rapid (and accel
erating) growth of velocity measured with debits sug
gests that many financial transactions— such as
arranging an overnight repurchase agreement— are
undertaken specifically to reduce checking account
balances. Since these transactions directly decrease the
volume of M1, rather than adding to the demand for M1
as conventional transactions would, they cause velocity
(as measured with debits) to rise. That is, these cash
management transactions raise velocity by increasing
the numerator in the debits/M1 ratio and decreasing the
denominator at the same time.4
Another reason the increased volume of debits may
not be increasing the demand for M1 can be seen from
the components of M1. In recent years, the growth of
total debits primarily reflected debits to demand deposit
accounts, whereas M1’s growth has been dominated by
increases in negotiable order of withdrawal (NOW)
accounts (Chart 3). If a larger volume of financial
transactions was increasing the demand for M1, we
would expect the growth of debits and the greater
demand for M1 to show up in the same component of
M1, but this generally has not been the case for the
1982-85 period.5 In 1985, however, the demand deposits
4ln addition, increased em phasis on cash managem ent in general and
tech nolog ical advances in m onitoring money balances have reduced
the level of M1 relative to both the level of GNP and the volum e of
debits over time.
5lt could be argued, of course, that the key issue is not w hich
com ponent of M 1 con tributed the most to M1’s grow th, but rather
w hether the larger volum e of debits to dem and deposits increased
the dem and for these deposits beyond what it otherw ise w ould have
been.

C hart 2

A lternative Measures of Velocity
Index 1946 = 1

400

---------------------------------------------------------------------------------------------------------------------------------

2 5

20
/
D ebits/M 1

------------ /-

------------- X ------

10-

G N P /M 1
50-

o L llll ll. l- L ll. lll 1 ll 1111II ll 11 111 I ll II 1 1I I 11111111111111 l l l l l - L
1971 72 73 74 75

S o u rce :

76 77 78 79 80 81 82 8 3 84 85

F e d e ra l R eserve B ulletin.

pi I I I l l I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I
1946

S ource:

F e d e ra l R eserve B ulletin.

FRBNY Quarterly Review/Summer 1986

category did contribute a larger amount to M1 growth
than in recent years, and financial transactions could
have played a significant role in M1’s very rapid growth
last year even though their importance in the longer run
is uncertain. In the remainder of this article, we will
show that financial transactions added about one per
centage point to M1’s growth in 1985. But this estimate
is subject to sizable error in either direction.6
Debits, GNP, and the demand for money

In making this estimate, a conventional money demand
equation (relating real money balances to real transsFor a more tec h n ic a l and d e tailed analysis of the transactions
dem and for M 1 that arrives at sim ilar estim ates of the possible
effects of fina ncial tra nsactio ns on M1 for 1985, see Peter E.
Kretzm er and R ichard D. Porter, “ The Demand for the Narrow
A g g regates— Is a Transactions A pproach S ufficient?" Board of
G overnors of the Federal R eserve System (July 1986), unpublished.

actions, a short-term interest rate, and lagged real
money balances) was used. Real debits and real GNP
were included as alternative measures of transactions.7
The left side of Table 1 shows that when debits and
GNP are included together in the money demand
equation at most one performs well. As alternative
proxies for total transactions in the economy, they are
competing to explain the movements in M1 and one of
the two is redundant. Estimated from 1959 through
1973, the coefficient on GNP is significant and has the
7Only the coe fficients on the transactions variable are shown. These
equations had all the well known problem s with sta b ility during the
mid-1970s, and the results should be interpreted with caution as a
result. Moreover, there are problem s with using the real volum e of
debits as an alternative measure of transactions. That is, w hile it
might be ap prop riate to use the GNP de flator to calcula te real M1
balances and real GNP, it may not be a p prop riate to calcula te real
debits this way since de bits contain a com bination of GNP and
non-GNP transactions.

Chart 3

Growth of T o ta l D ebits and T ra n s a c tio n D e po s its
By so u rc e
F o u rth quarter to fourth q u a rte r
P e rce n t
3 5 ------------------------------------------------------------------------------------------------------------G r o w t h of to t a l de b it s

30-

I
I Due to de b its to
------- NOW acco unts

Due to d e b its to
demand d e posits

Total de bits

25G r o w t h of M1 t r a n s a c t io n d e p o s i ts (M1 les s c ur re nc y )
20-

| Due to demand d e posits

1982

S ource:

|j§ |f | Due to NOW accounts

1983

F e d e ra l R eserve B ulletin.

FRBNY Quarterly Review/Summer 1986

M1 tra n s a c tio n s de p o sits

1984

1985

correct sign, whereas the debits variable has a negative
coefficient not significantly different from zero. When
estimated through 1982, the coefficients retain the same
signs as in the earlier period but neither is significant.
When the sample period is extended through 1985, the
coefficient on debits becomes positive and significant,
while the coefficient on GNP turns negative and insig
nificant. In contrast, when either debits or GNP is
included by itself (Table 1, right side), the estimated
coefficient is statistically significant and has the correct
sign across all sample periods, giving no clear indication
of the better measure of transactions for money demand
purposes.
Evaluating the regression results over the longer run
is complicated by the downward shift in the demand for
M1 in the m id-1970s.8 Hence, we reestimated the
money demand equation over the 1974 to 1984 period.
During this period, both GNP and debits continued to
have a strong correlation with M1 when included indi
vidually. (Their t-statistics are over 6; Table 1, bottom
right, memo.)9
Next, we use this equation to determine whether GNP
or debits can track M1 growth more accurately in 1985.
The first column of Table 2 shows that the out-of-sample
projection based on GNP substantially underestimated
M1 growth last year (four percentage points), whereas
the projection based on debits (Table 2, far right column)
missed by only about two percentage points. Therefore,
it seems that financial transactions played a major role,
adding roughly two percentage points to M1 growth. But
there are reasons to believe that the actual contribution
to M1 growth was somewhat less.
First, there is an alternative reason why GNP under
stated the demand for money. In the United States, a
larger volume of goods and services was purchased in
1985 than was produced. If consumers are purchasing
more goods, but those goods come from imports or
inventories rather than from current production, the
demand for M1 increases to make those additional
transactions but GNP does not increase. Hence GNP
(current production) understates the transactions
demand for M1, and M1 appears unusually strong.10
When domestic final demand is substituted for GNP, the
resulting error is about three percentage points (Table
2, center column). Therefore, after this adjustment, only
about one percentage point of M1 growth remains to be

attributed to financial transactions (and perhaps other
factors as well).
Second, the debits statistics may not adequately
capture financial transactions. The link between trading
volume and debits on a quarter-to-quarter basis has
been weak. Table 3 shows the results of regressing the
growth rate of debits on the growth rates of GNP and
stock and securities trading volume. The only variable
that is statistically significant (that is, has an estimated
coefficient with a t-statistic greater than 2.0) is GNP.11
"T o investigate further why the tra ding volum e variables were
insignificant, we looked at m onthly data to see if some of the
correlation was masked by quarterly averaging. As it turns out,

Table 1

Coefficients for Alternative Measures of
Transactions in a Standard Money Demand
Equation*
Included together
Years

Included separately
GNP
Debits

GNP

Debits

1959 to 1973

0.152
(3.3)

-0 .0 2 6
(1.7)

0.115
(36)

1959 to 1982

0.067
(1 8 )

-0 .0 1 4
( 1.1)

0.027
(3 0)

0.041
(3.1)
0.008
(2.5)

1959 to 1985

-0 .0 0 5
(0 .2)

0.015
(2 0 )

0.046
(4.3)

0.013
(51)

-0 .0 3 0
(0 .8)

0.025
(3.1)

0.143
(6.4)

0.030
(8 6 )

Memo:
1974 to 1985

'Standard Goldfeld formulation where the In (real M1) is regressed
on In (real GNP), In (short-term interest rate), and the In (lagged real
money). Only the coefficients on the transactions variables are
reported.

Table 2

Money Demand Errors for 1985 Using Alternative
Proxies for Transactions*
Quarterly growth rates
Gross
domestic
final demand

Debits

1985-1.............................................. 3.0

2.1

1.4

•For more detail, see Stephen G oldfeld, “ The Case of the Missing
Money,” B rookings Papers on E conom ic A ctivity (1976-111).

1985-11........................................... 2.9

1.8

1.0

1985-111........................................... 7.3

6.0

5.6

•Equations estim ated over this period also have larg er (in absolute
value) coe fficients on the short-term interest rate. This helps them
track the rapid M1 grow th in 1985 som ewhat better than those
estim ated over the longer run.

19 8 5 -IV ........................................... 2.9

1.5

0.4

Average

2.9

2.1

10For more detail, see Law rence J. Radecki and John Wenninger,
“ Recent Instability in M 1's Velocity," this Q uarterly Review (Autumn
1985).

Quarter

GNP

..........................

4.0

‘ Equations estimated from 1974 to 1984.

FRBNY Quarterly Review/Summer 1986

This result suggests that a large volume of financial
instruments is purchased without debits to the trans
actions accounts in M1, or that many financial trades
completed during a given day by an individual firm are
netted before its demand deposit account is debited.
We also tried using trading volume directly in the
money demand equation along with GNP. However, it
probably is not appropriate to deflate financial trans
actions by the GNP deflator (Footnote 7). To avoid this
problem, the money demand equation was estimated in
nominal terms. The resulting coefficients and the 1985
errors in projecting M1 growth are shown in Table 4. All
three measures of transactions have significant coeffi-

Table 3

Correlation Between Debits and Financial
Transactions
Quarterly growth rates
Debits = 9.7 + 0.92 (GNP)
(2 .8)
(2.8)

Fedwire activity

Debits = 9.7 + 0.88 (GNP) + 0.02 (Stocks*)
(2.7)
(0.9)
(2.8)
Debits = 9.4 + 0.89 (GNP)
(2.7)
(2.6)

.01 (Stocks) + 0.006 (Securities!)
(0.8 )
(0 .2)

‘ Dollar volume of transactions on the New York Stock Exchange.
fD ollar volume of trading by Government securities dealers.
Additional regressions using lags and seasonal dummies did not
produce appreciably different results.

Table 4

Results When Financial Transactions Are
Included Directly in Money Demand Equation*
Coefficients
GNP

...............................................................................

0.100
(3.2)

Stock v o lu m e ..................................................................

0.022
(38)

Securities v o lu m e ...........................................................

-0 .0 1 7
(2.4)

1985 errors (Quarterly growth rates)
1985-1................................................................................

1.7

1985-11............................................................................

3.1

1985-111............................................................................

6.0

1 9 8 5 -IV ............................................................................

2.1

Average

3.2

.........................................................................

'Nominal rather than real values were used in this equation for
money demand. As in the previous tables, only the coefficients for
the variables being studied are reported. The sample period was
from 1974 to 1984. Stock volume and securities volume are defined
in the same way as in Table 3.

FRBNY Quarterly Review/Summer 1986

cients. The coefficients on GNP and on the volume of
trading in the stock market also have the correct (posi
tive) sign. The coefficient on the volume of trading in
Government securities is negative, again implying that
many trades are done for the purpose of managing
money balances more efficiently (repurchase agree
ments, for example). In any case, this equation does not
track M1 growth in 1985 very well. Its average error was
three percentage points, compared with an error of two
percentage points when total debits were used and four
percentage points when GNP was used (Table 2).
Roughly speaking, the results in Table 4 are consistent
with those in Table 2. That is, when a variable meas
uring financial transactions is included in the equation,
the 1985 average error is about one percentage point
less than when GNP is used by itself. Hence, nonfi
nancial transactions not captured by GNP probably
account for another percentage point (Table 2, right
column), leaving about two percentage points of the
error in 1985 unaccounted for.

Another proxy for the volume of financial transactions
is the dollar volume of funds transferred over Fedwire.12
On a daily average basis, this volume has increased
from $200 billion in 1978 to about $700 billion in 1985.
Quarterly statistics on the volume of funds transferred
over Fedwire are available only since 1977. Therefore,
annual data were used to estimate money demand
equations, and the sample periods were extended back
to 1949 (Table 5). Estimated through 1974, the coeffi
cient on the dollar volume of wire transfers is significant
but has a negative sign (Table 5, equation 2), sug
gesting once again that many financial transactions are
made to manage money balances. Estimated through
1984, the coefficient on Fedwire volume remains neg
ative, but declines in absolute value by about one-half
(Table 5, equation 4), implying that its effect on M1 has
not been stable over time. These results are difficult to
interpret, however, because money demand equations
have generally not been stable when the sample period
is extended beyond 1974. Nevertheless, the negative
coefficient on the dollar volume of Fedwire transfers
does indicate that more rapid growth of financial trans
actions over the longer run has not been associated
with an acceleration in M1 growth. For financial transFootnote 11, continued
debits were only slightly more likely to increase in any same month
that trading volume increased. Over the past ten years, de bits and
Governm ent securities trading volum e moved in the same direction
in 52 percent of the 120 months, debits and stock m arkets volum e
in 61 percent.
12For earlier work using the num ber of wire transfers (as a proxy for
technological change) to explain unusual weakness in M1 in the
mid-1970s, see Ralph Kimball, op. cit., pages 12-22.

actions to explain the rapid growth of M1 in 1985, it
would be necessary to find reasons why the historical
relationship between financial transactions and M1 might
have changed.
One reason this relationship might have changed
somewhat in recent years is increased concern on the
part of banks about their potential exposure to corporate
customers that engage in large dollar volumes of
financial transactions. On any given day, some inflows
that are expected by a corporation may not materialize
(or some unexpected outflows may occur) resulting in
an overnight overdraft. If the firm does not qualify for
a line of credit that would cover the potential overdraft,
it might be required to hold a larger balance at the
beginning of the day. This could represent an indirect
channel through which the rapidly growing volume of
financial transactions could increase the level of demand
deposits, but it is impossible to quantify the effect.
Another indirect way financial transactions might affect
M1 is through the higher level of demand deposits that
firms are holding to compensate banks for the costs of
making a larger number of transactions, i.e., higher
compensating balances. Banks have also been moving
toward more explicit pricing of transaction account
services, and the growing number of financial trans
actions may now have a more pronounced effect on
M1’s growth as a result. Again, due to lack of data on
the total number of financial transactions, this effect
cannot be quantified. And working in the opposite
direction, firms have been moving toward using fees to
compensate their banks for transaction account services
rather than holding balances. So it is not clear that the
net effect of more explicit pricing has been to increase
M1 balances.13
Conclusions

In general, it appears that the more rapid growth of
financial transactions is not having a very large effect
on M1’s growth:
• Many financial transactions are made explicitly to
manage cash balances more efficiently. Such
transactions would, of course, tend to reduce M1
13Of course, com pensating balances cou ld have increased for reasons
other than the grow ing num ber of financial transactions. As interest
rates fall, firm s must hold a higher level of balances to com pensate
banks for the same level of transaction account services. W ithout
any changes in banking practices, this effect should be picked up
by the interest rate varia ble in the money dem and equation.
However, as banks move tow ard more e xp licit pricin g of services,
they may also enforce balance requirem ents more strictly. In turn,
this could make com pensa ting balances more responsive to interest
rate changes than in the past.

Table 5
Volume of Wire Transfers Over Fedwire
and the Demand for Money
Annual observations
1949-84

1949-74
Coefficients
GNP

.................... . .

Volume of wire
tr a n s fe r s .............

Equation
1

Equation
2

Equation
3

Equation
4

0.46
(58)

0.58
(5.3)

0.11
(2 3)

0.33
(35)

- 0.11
(2.2)

-0 .0 5
(2.1)

‘ Not included.

balances, not increase them, as other transactions
would.
• A large part of total financial transactions is done
by investment firms and dealers. They are among
the most sophisticated checking account managers,
attempting to keep their balances at frictional levels
almost regardless of the volume of transactions
undertaken. Of course, as the volume of transac
tions increases, these frictional balances are likely
to increase at least somewhat because of unex
pected cash flows or the desire to avoid costly
overdrafts.
• Many financial transactions can be completed
without using the checking accounts in M1. Stock,
bond, and money market transactions are often
executed by using accounts at an investor’s broker.
Moreover, individuals are likely to hold liquid assets
suitable for investment in their money market
deposit accounts or money market funds, not in
M1. These accounts, with their limited transactions
features, can be used to make purchases of
financial instruments without the funds flowing
through an M1 account. Likewise, the proceeds
from sales of financial instruments would not need
to be deposited in M1 accounts.
Even though stronger-than-expected M1 growth has
occurred during a period of rapid growth in volume of
financial transactions, longer-run relationships do not
confirm that there is a strong linkage between the two.

John Wenninger and Lawrence J. Radecki

FRBNY Quarterly Review/Summer 1986

Short-Term Borrowing by
Local School Districts

Short-term borrowing by school districts has undergone
a dramatic change in New York State as well as else
where in the nation. Traditionally, the purpose of such
borrowing has been to finance temporary cash shortfalls
that occur before property taxes are received. In the last
several years, however, it has also been used to finance
gaps created by state delays in payment of school aid.
And even more importantly, many school districts have
begun to use short-term borrowing to finance aggressive
investm ent program s. These developm ents have
increased the exposure of school districts to certain
kinds of risk and have resulted in several districts’
incurring financial losses (box).
Large increases in short-term borrowing and invest
ment by school districts are readily apparent in the
national statistics. But additional data to analyze the
incentives for such aggressive financial management or
to help prescribe effective remedies are not available
at the national level. This article closely examines those
factors that have led to widespread use of short-term
borrowing in cash management by school districts in
New York State, where such data are available. Based
on this analysis of New York, the article also suggests
possible ways to reduce the role of debt in school dis
trict cash management. In particular, more flexibility to
carry over revenues from one fiscal year to the next,
changes in the schedule of state aid payments, and
safer investment opportunities, such as state managed
investment pools, would reduce the incentives for

The author thanks Julie N. Rappaport for her research assistance in
the preparation of this article.
30

FRBNY Quarterly Review/Summer 1986

aggressive borrowing and investment by the nation’s
school districts.
In the next section of this article, we analyze the cash
management problems and financial profiles of school
districts, focusing particularly on New York State.
Cashflow projection models are then developed based
on the alternative financial profiles that emerge. These
models reveal why and to what extent New York school
districts have responded to their cash management
problems by borrowing to finance investment at higher
yields. These models also help to quantify the success
of state efforts to alleviate the need for such borrowing.
Even with these efforts, however, the analysis shows
that numerous cash management difficulties still remain.

The cash management problems of school districts
School districts across the country have increased their
average investment activity and short-term borrowing.
In 1961, the U.S. Advisory Commission on Intergov
ernmental Relations (ACIR) encouraged all local gov
ernments to invest more actively to generate additional
income. In the first ten years following that recommen
dation, school district interest earnings as a share of
revenue doubled from one-half of 1 percent to almost
1 percent.1 From 1972 to 1982, this ratio tripled.
During this 20-year period, short-term borrowing for
cashflow and capital purposes remained the most rap
idly growing portion of U.S. school district debt, which

1See U.S. Advisory Commission on Intergovernmental Relations,
Investment of Idle Cash Balances by State and Local Governments
(January 1961). National data are from the U.S. Bureau of the
Census.

totaled $36 billion by the end of fiscal year 1984. While
total debt outstanding at year-end had grown at an
average rate of about 4 percent per year since 1962,
the short-term portion grew by 8 percent per year. The
true growth of short-term debt exposure has probably
been considerably greater than suggested by these
figures because year-end measures exclude the
unknown but large amount of cashflow borrowing repaid
just before the close of each fiscal year.
New York school districts have been in a sim ilar
position: short-term borrowing and investment became
widespread. During the fiscal year July 1, 1983 to June
30, 1984, over half of New York’s 732 school districts
(excluding New York City) borrowed short-term for
cashflow purposes (Table 1, column 1). On average,
these districts issued $3.1 million per district in the form
of tax or revenue anticipation notes (TRANs). For the
U.S. Internal Revenue Service (IRS) to grant any TRAN
tax-exem pt status, eligible issuers must be able to
substantiate the likelihood of a cashflow deficit for at
least one month of the fiscal year. Because tax-exempt
TRANs have been issued by so many school districts,
it must also be true that cashflow deficits occur for the

majority of districts in New York during the course of the
year.
These cashflow deficits occur because school districts
face inherent cash management problems. Their major
expense is for personnel at a dollar cost that is gen
erally fixed in advance with frequent disbursements in
roughly constant amounts. As a result, districts have a
fairly uniform monthly need for cash. At the same time,
their major source of local revenue is property taxes,
which are also generally fixed in advance but received
(in contrast to disbursements) very infrequently. Usually
paid once or twice a year after the fiscal year has
begun, property tax payments can create large swings
in school district cash balances and create the need for
short-term borrowing.2
P e rso n n e l costs also can create cash managem ent problem s when
labor contract negotiations are not coo rdina ted with the budget
cycle. In particular, uncertainty con cernin g the size and effective
date of salary increases will make it more d iffic u lt to an ticip ate cash
shortfalls. In this article, it is assum ed that all costs are known in
advance. It is hard to be precise about tim ing of tax receipts across
the nation’s schools, but general observations are possible because
state legislation usually establishes guidelines for how and when
property taxes are to be collected.

Table 1

Financial Profile of New York State School Districts
Fiscal year 1984

Number of d is tr ic ts .............................. . . . .

Average property t a x ........................... . . . .
Average state a i d ................................. . . . .

Non-city districts

All
districts*
(1)

Nassau
(2)

Suffolk
(3)

732

69%
24%

51%
39%

45%
43%

41%
43%

88%

47%

59%

$5.3

$7.1

$1.4

$4.3

$21 8

$19.9

$7.8

$28.4

(share of own budget)
48%
39%

Other
(4)
544

City districts*
(5)
61

(percent of districts in category)
District b o r r o w in g f.............................. . . . .

55%

91%

(million dollars per borrowing district)
Average b o r r o w in g f ........................... . . . .

$3.1
(million dollars per district)

Average expenditures

.......................

. . . .

$ 11.8

(average borrowing as a share of average expenses)
Average debt d e p e n d e n c e !............. . . . .

26%

24%

36%

18%

15%

1.1%
2.1%

1.5%
2.2%

0.3%
1.8%

0.3%
1.3%

(share of own budget)
Interest p a y m e n ts f.............................. . . . .
Interest receipts ................................. . . . .

0 .6%
1.8%

'Excluding New York City.
|Tax and revenue anticipation notes.
Sources: New York State Department of Education and New York State Office of the Comptroller.

FRBNY Quarterly Review/Summer 1986 31

These swings in cash balances and the resulting need
to borrow have often been magnified by the way the
state disburses aid, the second major source of school
revenues. Because aid to school districts is also a large
part of New York’s budget, delays in payment of aid
have been a common solution to state fiscal stress.3
3The im portance of state aid to school districts is even greater in states
outside New York. State aid rose to over 45 percent of total U.S.
school district revenues in the 1980s after having remained at about
38 percent for two decades. For examples of school district borrowing
in general and of delayed payments of school aid in particular, see
Joe Mysak, "Same Time Next Year,” Credit Markets (June 9, 1986),
page 10, and Allen J. Proctor, "Tax Cuts and the Fiscal Management
of New York State," this Quarterly Review (Winter 1984-85).

Over the years, New York State has stayed within its
own budget limitations by delaying payment of up to 75
percent of school aid until April, May, and June, the last
quarter of the school fiscal year.4 Because districts
cannot similarly delay their expenditures, a second, mid
year deficit emerges (after tax proceeds have been
spent but before most state aid arrives), and many
districts need to borrow a second time each year.

4D istricts receive 8V3 percent of state aid per month in September,
O ctober, and November. They receive 25 percent per month in A pril,
May, and June under the regular aid program . S pecial program s for
earlier paym ents are discusse d in a later section.

Risks from Current Trends in School District Finance

The end result of the trend toward more borrowing and
investment has been an increasingly aggressive cash
management style that seeks high net yields while
exposing schools to several risks. One risk is that school
districts that rely on short-term borrowing may have to
cease operations temporarily if local lenders become
unwilling to provide enough funds. The likelihood of this
problem occurring is increasing. For example, in Iowa,
banks have become reluctant or unable to supply dis
tricts with all the funds they need. As a result, the state
recently had to intervene to ensure school districts timely
access to short-term financing.*
At least as important is the risk that large interest rate
swings may turn the process of borrowing and invest
ment into a source of revenue losses rather than gains.
This risk can be substantial because the decision to
borrow, the actual borrowing, and the investment of the
borrowed funds usually occur several months apart. If
the school district incorrectly predicts interest rate
movements, the cost of borrowing may substantially
exceed the return on investment. This situation has
become more common since 1979. School districts that
have been locked into losses for months at a time can
have difficulty finding additional revenues to replace the
failed investment program.
The search for higher returns has also led school
districts to undertake investments that have placed their
principal at risk. In 1984, the failure of two securities
dealers, the Lion Capital Group and RTD Securities,
turned these risks into losses for many investors,
including school districts. In New York State, for example,
62 districts may face possible losses of up to $77 million
as a result of insufficiently secured collateral for repur
*See Joe Mysak, "S am e Time Next Year," C red it Markets
(June 9, 1986), page 10.

FRBNY Quarterly Review/Summer 1986

chase agreements.t They have not yet recovered all
their investment, and special state legislation has been
enacted for the past three years allowing the districts to
finance these losses until the funds are recovered.t
High levels of short-term cashflow borrowing underlie
all these risks, either directly or through encouragement
of aggressive investment behavior. A process intended
to improve the fiscal health of local government has
evolved into a pattern of high-risk fiscal management.
Recognizing the breadth of this problem, in 1985 the
U.S. Advisory Commission on Intergovernmental Rela
tions recommended for all local governments that “short
term borrowing, both for operating and capital purposes,
be strictly limited and regulated...” §
fS ee New York State Assembly, Gambling with Public Funds: The
Lion Capital Bankruptcy and Its Implications for Government
Investment Practices (March 1985), pages 141-147.
tT h e failure of professional dealers who were cau ght in an
unanticipated interest rate sw ing made many school d istrict
o fficia ls aware that aggressive cash m anagem ent could have
sizable risks. To increase this awareness further, in D ecem ber
1984, the O ffice of the State C om ptroller of New York issued
a detailed investm ent manual for localities that em phasized
safety and liquid ity over yield as crite ria for investm ent
decisions (Cash M anagem ent an d Investm ent Policies and
P rocedures for Use by Local G overnm ent O fficials).
Publication of these guidelines was follow ed by an extensive
educational outreach program by both the State C om ptroller
and the Federal Reserve Bank of New York. Inform ed state
officials believe that investment practices in New York are
now more cautious than in 1984. Similar prudential efforts are
not obvious in other states, and an informal survey of the
Southeast suggests that use of risky investments such as
repurchase agreements is extensive (see B. McCrackin et a!.,
"State and Local Governments' Use of Repos: A Southeastern
Perspective," Federal Reserve Bank of Atlanta, Economic
Review (September 1985).
§See Advisory Commission on Intergovernmental Relations,
Bankruptcies, Defaults, and Other Local Government Financial
Emergencies (March 1985).

The overall effect of these property tax, state aid, and
expenditure flows can be estimated using cashflow
models that separately project and then combine the
various flows into a cumulative cash balance projection.
In particular, the cashflow models constructed for this
analysis project monthly cash balances that are con
sistent with the financial profiles in Table 1. While the
exact situations of individual districts may vary consid
erably, the cashflow models can reveal some of the
types of cashflow situations that have led to the current
financial practices of school districts.
The basic differences in cashflow profiles across
districts are closely related to the timing of property
tax receipts. In contrast, the timing of conventional
forms of state aid receipts and of overall expenditures
is fundamentally alike for all districts. The schedule

of property tax receipts, therefore, can be used to
divide New York districts into four general cash man
agement profiles.5
• Nassau county districts are highly dependent on
property taxes but they receive no tax revenues
until the fourth month of the fiscal year, and half the
revenues are not received until the last quarter of
the fiscal year. Combined with the fourth quarter
receipt of most state aid, these factors substantially
heighten the likelihood of cash shortages during the
year. As a result, cashflow borrowing averages 24
percent of expenditures (Table 1, column 2).
• Suffolk county d istricts have above-average
dependence on property taxes that are received
very late in the year. They must operate through
December without tax revenues. Reflecting these
circumstances, short-term borrowing on average
finances 36 percent of their expenditures (Table 1,
column 3).
• City school districts typically receive taxes in the
first through third months of the fiscal year and
borrow relatively less than the other three types of
districts, an average of 15 percent of expenditures
(Table 1, column 5).6
• Non-city school districts (outside Nassau and Suf
folk counties) generally receive property tax reve-

5The m odels used in this study assume that d is tric t receipts and
paym ents are known with certainty. Obviously, unanticipated
changes can occu r that may raise or lower m onthly cash balances
from the projected levels. One area of uncertainty is the am ount of
delinquent property tax paym ents. H igher de lin que ncies will reduce
revenues and increase school d is tric t cash deficits. The effect of
delinquencies on school revenues, however, is lim ited in duration in
New York. For most school districts, the county governm ent assumes
all delinquent school d is tric t property taxes by A pril (the beginning
of the fourth quarter of the school fiscal year). At that point, the
school d istrict receives all its levied taxes and the county
undertakes collectio n efforts.
6The New York City school d is tric t is excluded in this study. The
necessary data are unavailable because school cash m anagem ent is
so intertw ined with the rest of the c ity ’s finances. We must
necessarily generalize across other city school districts because
their situations are governed by the separate charters of their
respective cities. For exam ple, the average reliance on property
taxes rises to 44 percent if we exclude Rochester, Buffalo, and
Syracuse— fiscally dependent districts that also rely on sales tax
revenues. The date of property tax receipts is the beginning of the
fiscal year (July 1, except for Syracuse) for 12 d istricts and
Septem ber for all other city districts. At least five city d istricts
receive their property taxes in two installm ents and two d istricts in
four installm ents throughout the year. In addition, some cities benefit
from a special state aid program (Hurd aid) that can disburse
assistance much earlier in the school fiscal year than the regular
school aid program .

FRBNY Quarterly Review/Summer 1986 33

C h a rt 2

S c ho o l D is tric t C u m u la tiv e Cash B a la n c e s
M inim um b o rro w in g s tra te g y fo r a d is tr ic t w ith
firs t and th ird q u a rte r d e fic its
Thousands o f d o lla rs

S ource: F e deral R eserve B ank of New Y ork
sta ff e stim ates.

C hart 3

School D is tr ic t C u m u la tiv e Cash B ala n c e s
A lte rn a tiv e b o rro w in g s tra te g ie s
T h o u sa n d s o f d o lla rs
4500
4000
3500
3000
2500

nues in the third and fourth months of the fiscal
year.7 The slightly longer delay than for city districts
increases the likelihood of insufficient cash in the
early months of the fiscal year. Therefore, non-city
districts borrow slightly more than city districts
(Table 1, column 4). Because these non-city dis
tricts account for 544 of New York’s 732 districts,
they will be the focus of much of the following
analysis.
To illu stra te the general nature of the cashflow
problem before going into more detail, Chart 1 shows
an example of a non-city district’s cashflow profile for
the July 1 to June 30 fiscal year. Property tax receipts
create one revenue bulge in September and October,
and state aid creates a smaller bulge in April through
June. Expenditures are generally uniform across
months. The net effect of these flows would be a period
of sizable cash balances available for investment from
September through January and during May. At the
same time, the school district would need to borrow
short-term in order to pay expenses in July, August,
March, and April.
By applying appropriate interest rates to these periods
of borrowing and investment, it is possible to simulate
school districts’ likely interest earnings and payments.
By comparing actual and projected interest flows, it can
be estimated how closely d is tricts’ borrowing and
investment followed projected patterns based solely on
routine cashflow needs. Any differences between the
actual and projected values might be attributed to efforts
to earn additional income through borrowing for invest
ment purposes.
Overall, projections of interest earnings and payments,
based on fiscal year 1984 interest rates, are both con
siderably less than the $157 million of total interest all
New York districts earned and the $52 million of interest
they paid in that year. This probably occurred because
districts borrowed more than they needed for routine
purposes and invested the additional proceeds at higher
yields.8

2000
7State law allows counties to pass tax acts which, am ong other
things, determ ine w hether school property taxes may be paid in up
to six installm ents. Seven counties appear to perm it such
installm ents. Moreover, a few counties are governed by spe cia l tax
acts or charters. The Nassau and Suffolk County Tax A cts establish
sp e cific tax paym ent dates for school districts in their jurisd ictions.
These dates are unusual and create special problem s for their
school districts.

1500
1000

S ource: F e d e ra l R e s e rv e B ank of New Y ork
s ta ff estim a te s.

FRBNY Quarterly Review/Summer 1986

•Since the end of fiscal year 1984, interest rates have fallen. Similar
cash m anagem ent practices today would be consistent with lower
interest earnings and payments. The large diffe ren ce betw een actual
and projected net interest earnings is consistent with efforts by
school districts to earn above-norm al net yields through aggressive,
and generally riskier, investm ent practices. In fact, the New York
State School Boards A ssociation and the New York State A ssem bly

Borrowing for the explicit purpose of investing at
higher yields is generally referred to as arbitrage, and
it is widespread among issuers of tax-exem pt debt
throughout the United States. The next section of this
article reviews the routine cashflow borrowing needs of
non-city districts and uses cashflow models to analyze
in more detail their opportunities and incentives to
borrow for arbitrage purposes. Simulations of possible
arbitrage programs are then used to estimate whether
arbitrage has increased short-term borrowing in the four
categories of New York districts summarized in Table 1.

Arbitrage in New York school districts
The average projected cashflow of a non-city district in
New York has two potential deficit periods because of
the timing of property taxes and state aid. As shown by
the solid line in Chart 2, these deficits would require a
borrowing program lasting at least several months each
year. The school would need to borrow enough at the
beginning of the fiscal year to finance July and August
expenditures, and it would generally be able to repay
that debt in September when property taxes began to
arrive. Similarly, it could finance the third quarter deficit
by borrowing again in March and repaying in May when
most of its state aid would have arrived. The strategy
of borrowing the smallest amount for the shortest period
is shown by the dashed cash balance line in Chart 2.9
The incentive for arbitrage in New York
Starting from this basic borrowing program, a school
district could make some alterations that would reduce
its costs. The school district could reduce the cost of
issuing notes twice in the same year by extending the
term of the TRAN it issues in the first quarter. Thus, in
stead of repaying the July note in September, it would hold
on to the funds until they were needed again in March.
Keeping the debt outstanding through March has other
cost advantages as well if interest rates remain rela
tively stable. In the months when the funds are not

Footnote 8, continued

found that school districts tended at that time to favor yield over
safety in pursuing their investment programs. See New York State
School Boards Association, “ Survey of Local School District
Investment Policies," draft report (April 1985), and New York State
Assembly, Gambling with Public Funds (March 1985). In 1985, the
state undertook an extensive investment education program which
reportedly has led school districts to shift their emphasis from yield
to safety. See the box for more information.
•For some school districts this procedure may not be possible or it
may incur excessive debt issuance costs. Such districts may borrow
once a year for their entire year’s needs even if other incentives for
extended-term borrowing do not exist. Also, two measures to reduce
the size of first and third quarter deficits are discussed in the next
section. Many non-city districts appear to use these measures to
lower their cashflow deficits from the levels shown in Chart 2.

needed to finance deficits, schools are able to invest the
borrowed money at interest rates that are almost always
higher than their borrowing rates. School districts can
do this because they borrow in the tax-exempt market
and invest in the taxable market, but are not subject to
income tax themselves. The more the borrowing period
is extended, the more the school district is able to take
advantage of this favorable interest rate spread. In
practice, the school district could, by borrowing early
and repaying late, have borrowed funds available for
investment for almost the entire year. This strategy
would help districts reduce the net costs of financing
recurrent cashflow deficits.
Chart 3 illustrates how investment funds can be gen
erated by extended borrowing. The solid area represents
the amount of cash a district has available for invest
ment when it borrows for as short a period as possible.
By extending its borrowing to twelve-month maturities,
the school district increases its investable cash balances
by the amount of the gray area in the chart. This fullterm borrowing provides schools with investable sur
pluses in July and April, which would otherwise be
deficit months. It also enlarges the projected surpluses
in September through February and in May.
Comprehensive data are not available on how many
of the over 400 New York districts that borrowed in 1984
used this arbitrage technique to reduce their net bor
rowing costs. Announcements published by a smaller
number of districts, nevertheless, suggest that full-term
borrowing appears widespread. Specifically, of the 186
TRANs publicly announced in 1983 and 1984, 85 per
cent were issued around the start of the fiscal year and
96 percent matured near the end of the fiscal year.10
Another way to increase arbitrage earnings is to
increase the amount of borrowing beyond what is
needed to finance cashflow deficits. For example, a
school district with a cash shortfall of $1 million could
borrow that amount at a tax-exempt rate of 6 percent
for an annualized cost of $60,000. But if the district
desired to reduce the overall cost of financing, it could
borrow $2 million and invest the extra $1 million at a
taxable yield of 8 percent. The school district’s annu
alized cost of borrowing would rise by $60,000, but its
investment income would rise by $80,000, for a net
arbitrage profit of $20,000 and a reduction of its net
annual interest expense from $60,000 to $40,000.
As this example illustrates, from the standpoint of
arbitrage earnings, the school district has an incentive
to borrow as much as possible. The most important
limits to arbitrage activity are IRS regulations that put
a ceiling on the use of tax-exempt borrowing to finance

10These data were compiled by the Office of the New York State
Comptroller from daily issues of the Bond Buyer.
FRBNY Quarterly Review/Summer 1986 35

arbitrage.11 Nonetheless, IRS regulations still allow
schools substantial leeway in arranging an arbitrage
program.
A school district that borrows up to the IRS maximum
for the full term of the fiscal year will substantially
increase its m onthly cash balances available for
investment. The projected increase is represented by
the hatched area in Chart 3. Provided the taxable/taxexempt yield spread is large enough, the school district
will be able to use its additional investment income to
pay all the interest cost of the additional arbitrage bor
rowing and part or all of the interest cost of financing
its recurrent cash deficits.
"S c h o o ls and other local go vernm ents are perm itted to issue taxexem pt TRANs up to the sum of the ir largest monthly cum ulative
cash d e fic it plus the follow in g m onth’s expenditures. When monthly
exp enditu res are larg er than m onthly deficits, this rule allows
localitie s to borrow more than tw ice their de fic it-fin a n c in g needs.
C ashflow projection s sug gest that New York school districts
ge nera lly face the latter situation. The IRS also requires that the
borrow ed funds be spent w ithin several months of issuance, but this
requirem ent is generally satisfied for schools when the July-A ugust
shortfall is funded.

C h a rt 4

E s tim a te s of S h o rt-te r m Borrowing
B eyond R outine Cashflow Needs
T h ousa nds o f d o lla rs per
ten m illion d o lla rs of b u d g e t
5 0 0 0 ----------------------------------------------------------------------------O M axim um estim ated b o rro w in g a llo w a b le
4 5 0 0 — ♦ O b s e rv e d ave ra g e b o rro w in g FY 1984
A E s tim a te d ro u tin e b o rro w in g need
4 0 0 0 ----------------------------------------------------------------------------3 5 0 0 --------- O ------------------- O ---------------------------------------3 0 0 0 ----------------------------------------------------------------------------A
A
2 5 0 0 ------------------------------- * ---------------------------------------

2 0 0 0 ----------* -------------------------------------------------------------------1 5 0 0 ------------------------------------------------------- O ---------------100 0 -------------------------------------------------------------------------------------------------------------------------- —

------------------------------------

A
5 0 0 -----------------------------------------------------------------------------

N assau
n o n -c ity

S u ffo lk
O ther
n o n -c ity
n o n -c ity
S chool d is tr ic t c a te g o ry

In d e p e n d e n t
city

A verage b o rro w in g above routine levels su g g e s ts
tha t a rb itra g e b o rro w in g is ta k in g p la c e . S choo l
d is tric ts are groupe d by p ro p e rty tax levy dates.
S ource: F e deral R eserve Bank o f New Y ork
s ta ff e stim a te s.

FRBNY Quarterly Review/Summer 1986

Estimating the use of arbitrage
A typical non-city school district in New York State
appears to be able to offset virtually all the cost of
financing a cashflow deficit by using arbitrage. The
exact benefits of arbitrage will vary from district to dis
trict, depending on various factors including the yields
at which it can borrow and invest. Estimates suggest,
however, that an interest rate spread of only 95 basis
points would generally make arbitrage worthwhile in the
sense of reducing overall costs, and larger spreads
could result in net profits.
For example, extending the maturity of debt from two
months to 12 allows a district to reduce its net interest
cost from an estimated $7,800 to only $460 based on
a borrowing cost of 6 percent and an investment yield
of 8 percent (Table 2). Full-term borrowing of an amount
equal to the largest monthly cash deficit plus the fol
lowing month’s expenditures (the maximum allowed by
the IRS) would enable the district to pay all borrowing
costs and earn an estimated profit of about $14,000.
The prevalence of arbitrage borrowing in New York
can be estimated by comparing average observed levels
of borrowing with projections of routine and maximum
amounts of borrowing. This is done by projecting
monthly cashflows in the absence of borrowing, iden
tifying the largest monthly cashflow deficit (which
determines the routine level of borrowing necessary),
and adding to this deficit the subsequent month’s
expenditures (which determines the maximum allowable
level of borrowing).
Estimated routine borrowing needs and maximum
allowable borrowing are compared with average bor
rowing in fiscal year 1984 for all four categories of dis
tricts (Chart 4). If observed average borrowing is near
the estimated routine borrowing level, then it is likely
that school districts in that group do not generally
borrow primarily for arbitrage purposes. If borrowing
exceeds the routine level, on the other hand, then some
arbitrage borrowing is taking place. Borrowing near the
estimated maximum level suggests that arbitrage is the
principal motivation for short-term borrowing by most
school districts in that group.
This procedure reveals that city districts borrowed on
average as much as IRS regulations would allow; that
is, arbitrage played a significant role in their fiscal year
1984 short-term borrowing programs. In contrast, non
city districts in Nassau and Suffolk counties seemed to
borrow only about as much as required by their severe
projected cashflow problems. Thus, while they borrow
proportionately more than any other districts in the state,
they do not appear motivated primarily by arbitrage.
Arbitrage does seem to play a modest role, however, in
the borrowing decisions of non-city districts in other
counties. These districts on average borrow more than

Table 2

The Use of Arbitrage to Reduce the Cost of Financing Cashflow Deficits*
In dollars
Without arbitrage

With arbitrage

Maturity of b o rro w in g ...............................................................................
Amount of b o rro w in g ...............................................................................

Minimum-term
Routine

Full-term
Routine

Full-term
Maximum

Cost of borrow ing......................................................................................
Income from investment of borrowing .................................................
Net cost of bo rrow in g...............................................................................

13,340
5,540
7,800

35,350
34,890
460

79,500
93,760
-1 4 ,2 6 0

*Based on simulations of a non-city school district receiving 50 percent state aid with an annual budget of $10 million. Interest rates are 8 percent for
investments and 6 percent for borrowing applied to balances at the beginning of each month. The principal invested is the same as in Chart 5. Total school
district investment income also includes $58,910 of interest earned from investment of unborrowed funds, principally unexpended property tax receipts.
Full-term, maximum borrowing has estimated arbitrage gains exceeding costs for taxable/tax-exempt spreads exceeding 95 basis points in this simulation.

their routine cashflow needs but not generally as much
as IRS regulations would allow.12
Up to this point, the evidence shows that school dis
tricts in New York borrow short-term both for routine
cashflow and for arbitrage purposes. Estimates of
probable cashflows suggest that this borrowing finances
one or more temporary deficits that result from untimely
receipt of property taxes and state aid. Once a school
district does any borrowing, it has a strong incentive to
borrow even more for longer periods in order to reduce
its net cost of borrow ing. Since debt seems to
encourage even more debt, the obvious solution is to
reduce, if not eliminate, routine short-term borrowing
needs. (Also see the box for additional reasons for
reducing debt, including the risks and losses some
districts have incurred.)
State programs to reduce districts’ need to borrow

New York has two laws that attempt to reduce or elim
inate the cashflow deficits that many schools encounter.
One provides schools with some scope to self-finance
their first quarter deficits. The second provides earlier
payment of state aid that may reduce third quarter
cashflow financing needs for some school districts.
First quarter deficits
The majority of school districts are often short of cash
12These com parisons are based on averages for each group of school
districts and, therefore, only suggest the motives underlying
individual dis tric t borrow ing. The am ount of borrowing and the role
of arbitrag e vary across districts. For exam ple, fiscal year 1984
borrow ing by the 61 city dis tric ts (excluding New York City)
averaged 15 percent of expenditures but ranged from zero to 73
percent. Twenty-five districts did not borrow that year, twelve
borrow ed about 10 percent of expenditures, twelve borrowed about
20 percent, nine borrow ed up to 30 percent, and three borrowed
more than half their budgets.

at the beginning of the fiscal year in July and August
because they receive no property tax revenues or state
aid before September. At the same time, the closing
months of the fiscal year bring an estimated cash sur
plus. If schools were able to carry some fourth quarter
surplus over into the initial months of the next fiscal
year, the need for first quarter borrowing might be
reduced.
In general, school districts are required to return to
taxpayers any funds that are left over at the close of
the fiscal year.13 Since 1977, however, New York state
law has allowed schools routinely to retain up to 2
percent of their budgets as unreserved balances to
finance the initial months of the subsequent fiscal year.
Aggregate data provided by the State Department of
Education suggest that school districts on average have
taken full advantage of this program. But it is not
enough. Cashflow projections suggest that even if the
amount of permitted carryover were doubled, enough
cash would be on hand to finance only July cash needs
for the average district. About half of the first quarter
borrowing would still be necessary to cover shortages
in August.
Third quarter deficits
The projected deficits that many school districts
encounter in the third quarter are caused by payment
of most state aid in the closing months of the fiscal year.
13ln New York, the general fund balances of districts are called
unreserved balances. The prohibition against accum ulating
unreserved balances is intended to keep property tax rates as low
as possible. Ironically, using tax rate changes instead of
accum ulated balances to balance budgets over a business cycle
may actually raise the overall burden on taxpayers. For additional
discussion of this effect, see Allen J. Proctor, “ Tax Cuts and the
Fiscal M anagement of New York State,” this Quarterly Review (Winter
1984-85).

FRBNY Quarterly Review/Summer 1986 37

As the state increases its support (in percentage terms)
of local education, the unfavorable effects on schools’
cash management problems (not necessarily their
overall budgets) of this aid disbursem ent schedule
become more pronounced. In particular, increased reli
ance on state aid shifts a larger share of total school
revenues into the last quarter of the school year. This
shift has two adverse consequences for school districts’
cashflow profiles. Increased aid dependence reduces
the projected amount of cash that schools have avail
able for investm ent in the second quarter, and it
increases the projected size of third quarter cash deficits
(Chart 5).14
Lower balances in the second quarter will reduce
investment income. The loss of investment income off
sets some of the value of state aid to a district’s budget.
As a result, a district’s taxes may be higher, expendi
tures lower, or investments more aggressive than they
would be if state aid were paid earlier. All other things
equal, a school district financed 60 percent by state aid
may have as much as two-thirds less cash available for
investment than a school district financed only 40 per
cent by state aid.
14D ependence on state aid has also cre ated cashflow uncertainty for
d istricts in A pril and May because the state has not always made its
paym ents on tim e (due to delays in approval of the state budget,
w hich must be approve d before any paym ents can be made after
A pril 1, the start of the state fiscal year). Schools, therefore, must
take steps to ensure that they can continue operations if aid is late.
The state has begun to fund an escrow account to ensure spring
paym ents. At some point in the future, the amount in escrow may be
sufficie nt to elim inate the risk of late A pril or May payments.

The second effect of greater state support is that third
quarter cash shortfalls may increase and schools may
need to incur more short-term debt. For example, model
projections suggest that the average-sized non-city
school district will have cashflow deficits of about $1
million in March and April if it is 60 percent funded by
state aid, whereas it would have cashflow surpluses if
state aid were only 40 percent of its budget (Chart 5).
In fiscal year 1984, state aid to non-city districts aver
aged around 60 percent for districts in 20 counties,
around 50 percent for districts in 19 counties, and 40
percent or less for districts in the remaining 18 counties.
Recognizing the adverse impact on some school dis
tricts of paying state aid in the spring, the state has a
special payments program that moves some state aid
payments from April and May into December, January,
February, or March. The formula determining the size
of the special payments is essentially based on the
share of state aid in a school district’s budget.
A large number of districts receive these more rapid
payments. In the school year ending June 1984, 405
districts received March payments, 228 received Feb
ruary payments, and 43 received January payments. No
school district received aid payments in December of
that year. The amount of special aid payments totaled
about 6 percent of the state school aid budget with
about 5 percent paid in March.15
15The am ount and tim ing of the special paym ents varies each year
acco rding to the form ula. In fiscal year 1987, for exam ple, some aid
is scheduled to be paid in D ecem ber and the total am ount of
special aid represents 7 pe rcent of the state school aid budget.

Table 3

Projections suggest that districts needed this share of aid each month...
Projected average percentage of aid needed

Independent c i t y ..................................................... . . . .
Non-city
O th e r ..................................................................... . . . .
Nassau .................................................................. . . . .
S u f f o lk .................................................................. . . . .

December

January

February

March

12.5

0
12.5
50

0
12.5
0

11.2
12.5
0

12.5
12.5
0

While the state paid early aid to this share of districts in fiscal year 1984...
Percentage of districts In category
Independent c i t y .....................................................
Non-city
O th e r ......................................................................
Nassau ..................................................................
S u f f o lk ..................................................................

....

....
....
....

0
0
0

8
0
0

39
2
3

63
2
19

These estimates of needed aid are based on school districts which are heavily dependent on state aid in their annual budgets. The need for early aid
would be lower in school districts where state aid accounts for less than half of their budgets.

FRBNY Quarterly Review/Summer 1986

The overall success of this more rapid payment pro
gram can be assessed by comparing model projections
of the monthly need for special aid (as measured by
periods of cash deficits) to the special payments that
were actually made. In fiscal year 1984, a total of 25
percent of aid was paid from Septem ber through
November, 6 percent from January through March, and
69 percent from April through June. Table 3 provides
estimates of a projected distribution of special aid that
would have precluded the need for school districts to
borrow from December through March. Comparisons of
these estimates with aggregate data on the state special
aid program suggest how successfully the program may
be meeting individual district needs.
• Independent city districts, on average, required up
to 12.5 percent of their aid in December and again
in March in order to avoid cash shortfalls. By and
large, the bulk of city districts received March
payments that year. None received December
payments, which probably left a number of city
schools with deficits to finance.
• Most non-city districts (outside Nassau and Suffolk
counties) have projected deficits in February and
March that could be eliminated by special aid pay
ments of up to 12.5 percent in each month. Districts
receiving the average state aid allotment would
need few if any special payments, but districts
heavily dependent on state support are likely to run
short of cash without a sizable amount of special
aid. By and large, the state program addressed
their February and March needs that year: the
majority of non-city districts (outside Nassau and
Suffolk) that were heavily dependent on state aid
received February or March payments.
• Non-city districts in Nassau and Suffolk counties
have unusual cashflow problems that are not well
addressed by the state program for earlier payment
of aid. In particular, districts in these counties
receive most of their state aid and property tax
revenues in the last quarter of the fiscal year. As
a result, their need for special aid payments is
projected to be on average the most severe in the
state, particularly in December and January.16 In
1984 (and currently), Nassau and Suffolk districts
have not received December or January aid pay

ments and only a small percentage have received
payments in February or March.
In sum, the New York program seems to address third
quarter cashflow problems for the majority of school
districts. Most city and non-city districts receive special
aid payments that generally correspond to their probable
needs. This program, however, does not serve many
districts in Nassau or Suffolk counties, despite the
likelihood of severe cashflow problems. Even if the
program were applied to more districts in those two
counties, however, it would be of limited value. In par
ticular, the February and March payments generally
provided by the program would not address their sizable
projected cash shortages in earlier months. As a result,
these districts would probably continue to require sub
stantial third quarter borrowing.
The combined effect of state programs
Even with state efforts to reduce the need for short-term

C hart 5

The R e la tio n s h ip B e tw e e n P ro je c te d Cash
B ala n c e s and the Use of S ta te Aid
By sh a re of sta te aid in scho ol d is tr ic t bu dgets
T h o u sa n d s of d o lla rs pe r
ten m illion d o lla rs of budget
3000A id as sha re of
d is tric t revenue

2500-

40 p e rc e n t
50 p e rce n t
60 p e rc e n t

S e p te m b e r-F e b ru a ry

M a rch -A p ril

P ro je c te d a ve ra g e m onthly ba la nce
S e p te m b e r-F e b ru a ry b a la nces are a s o u rc e of

16State o fficia ls were aware of this problem when the spe cia l aid
form ula was set up. The state form ula was con structe d to help
d istricts whose third qu arte r needs were p rin cip a lly due to the state
aid schedule. In contrast, the third qu arter needs of districts in
Nassau and Suffolk counties are prin c ip a lly due to the late property
tax schedules esta blished by their respective County Tax Acts.

in ve stm e n t incom e. M a rc h -A p ril d e fic its re fle c t the
n eed fo r e a rlie r state aid or s h o rt-te rm bo rrow in g.
S ource: F e d e ra l R e se rve Bank of New Y ork
s ta ff e stim a te s .

FRBNY Quarterly Review/Summer 1986

borrowing by New York school districts, about 60 per
cent of all city districts, 90 percent of non-city districts
in Nassau and Suffolk counties, and about 50 percent
of other non-city districts issued TRANs in fiscal year
1984. Some of this borrowing is unavoidable, and some
appears to be for the purpose of generating arbitrage
profits.
Overall, it appears that the cashflow problems faced
by most school districts in New York are primarily in the
first quarter when one or two months’ operations need
to be financed until property taxes are received.
Addressing this problem would go far in reducing the
need for borrowing by schools and, hence, would also
reduce some of their arbitrage activity. The cashflow
problems of districts in Nassau and Suffolk counties are
different from those in the rest of the state. They are
more severe and require a different approach than has
been taken thus far.

Potential improvements to school district finance
While this analysis has focused on New York, cash
management problems are an integral part of local
school finance everyw here. The severity of these
problems obviously varies across school districts and
across states, yet one or more common sources of dif
ficulty emerge:
• Property tax schedules may not be coordinated with
spending requirements.
• State aid schedules may have adverse effects on
the cashflow of schools.
• Strong incentives may exist for schools to invest
aggressively and to borrow more than necessary.
Property tax payments ideally should begin at the start
of the fiscal year. If rescheduling tax payments is not
practical, an alternative would be to allow schools to
accumulate sizable surpluses that could be carried from
one fiscal year to the next. Permitting self-financing of
cashflow deficits that occur at the start of the fiscal year
would make a change in property tax schedules less
critical. Under present conditions, existing limits on
carryover may create an artificial need for schools to go
into debt, with accompanying pressures to reduce the
cost of borrowing through arbitrage. For school districts
to carry over amounts large enough to eliminate firstquarter borrowing, it might be necessary to raise taxes
temporarily to accumulate sufficient surpluses. This
additional cost to the taxpayer is worthwhile only if it
is offset by the benefits of lower debt.
States should also reexamine the impact of school aid
payment schedules on local cash management. New

FRBNY Quarterly Review/Summer 1986

York’s special rapid payment program meets the needs
of a majority of school districts, but even this program
seems insufficiently focused on the overall cashflow
condition of individual districts. As with property tax
payment schedules, aid payments ideally should closely
parallel districts’ spending needs.
States, of course, have their own cashflow problems,
and the appropriate solution would strike a balance
between the timing of local needs and the state’s ability
to pay on an earlier schedule. For example, because of
cash management problems, New York State currently
finances the bulk of school aid with its own short-term
borrowing, and, hence, it has an incentive to make the
payments as late in the year as possible. This approach,
in effect, means that both the state and the school dis
tricts are often borrowing against the same aid dollar.
New York State is borrowing in anticipation of tax rev
enues, and the school districts are borrowing until state
aid arrives. The overall expense of double borrowing to
the taxpayer is obvious, and the state is already taking
steps to reduce its reliance on short-term debt. But it
would also be expensive for the state to pay aid even
earlier in the year so that borrowing by school districts
could be reduced or eliminated.
Finally, there appears to be a need for increased
prudential oversight of short-term borrowing by school
districts. At a minimum, improved monitoring and
reporting would enable public officials to anticipate
financial problems before they became unmanageable.
Some states might prefer more direct involvement in the
cash management of schools. Three specific improve
ments could be considered:
• More information is needed on individual district’s
cashflow problems and their borrowing practices.
Presently the cashflow situation of school districts
must be simulated using models such as those
used in this study. With more complete data, state
officials would be better equipped to develop local
property tax laws and school aid programs that
provide schools with adequate cashflow throughout
the year. And, on the debt side, short-term bor
rowing is not necessarily a good indicator of school
district cashflow problems. It would be easier to use
debt statistics to assess the financial condition of
districts if enough information were available to
separate, for example, arbitrage borrowing from
other short-term borrowing.
• Increased restrictions on how borrowed funds could
be invested would help avoid some potential prob
lems (box). Arbitrage provides strong incentives for
aggressive investing to raise yields. This behavior
exposes public funds to levels of risk that have

already resulted in financial losses. One policy
response would be to allow investment of borrowed
funds only in instruments with the lowest credit and
market risk. Such a requirement would sharply
reduce the incentive for arbitrage borrowing. New
York’s recent guidelines on permissible investments
are a worthwhile move in this direction.17

17See Office of the State Comptroller, Cash Management and
Investment Policies and Procedures for Use by Local Government
Officials (Albany, New York, December 1984).

• Requirements that state or regional short-term
investment pools manage school districts’ invest*
ments would insulate districts from pressures to
maximize investment income. Without direct control
over the yield from investing their borrowed funds,
districts may be less inclined to borrow for invest
ment purposes and, of course, would not be able
to take unacceptable risks. Large pools also would
provide other advantages, including lower man
agement costs, greater portfolio diversification, and
lower transaction fees.

Allen J. Proctor

FRBNY Quarterly Review/Summer 1986 41

In Brief

Economic Capsules
Are Large U.S. Banks
Moving International
Activity Off Their Balance
Sheets?
Recently market observers have drawn attention to the
increased off-balance-sheet activity of banks.1 This
article finds evidence that the top nine banks are doing
more international business off their balance sheets.
This shift by big U.S. banks is centered in the G-10
countries.2 Also, there is some evidence for the asso
ciation of the shift with a new financial instrument.
This financial innovation, the note issuance facility
(NIF), is transforming international lending. Instead of
arranging a medium-term credit facility with a syndicate
of banks, many borrowers now arrange a NIF. Under this
facility, the borrower can repeatedly issue short-term
paper through an agent or by an auction. In the event
that the paper does not sell below a contracted interest
rate, expressed as a spread over an interbank reference
rate, the underwriting or managing banks undertake to
buy the paper or to make advances. Some NIFs permit
borrowing from the committed banks with no prior
attempt to sell notes. The contract, which typically has
a life of three to seven years, limits the underwriters’
obligation to extend credit under specified circumstances
that may include a deterioration in the borrower’s
creditworthiness. Thus, the borrower may enjoy cheaper
short-term funding with some assurance of its availability
over the medium term.

’For example, see Bank for International Settlements, Recent
Innovations in International Banking (1986).
2G-10, as used here, includes Belgium, Canada, France, Italy, Japan,
the Netherlands, Sweden, Switzerland, West Germany, and the
United Kingdom.
42

FRBNY Quarterly Review/Summer 1986

The innovation of the NIF has an ambiguous effect
on U.S. banks’ commitments to lend. On the one hand,
borrowers who have switched from medium-term syn
dicated credits to NIFs may rely on U.S. banks only for
the commitment to lend while depending on other
sources— including nonbanks—for actual funding. On the
other hand, borrowers who previously relied exclusively
or mainly on large U.S. banks for lines of dollar credit
may be broadening the nationality and increasing the
number of their committed banks and so lowering the
cost by arranging NIFs. Thus, the market for contrac
tually committed lines of credit is growing, but large U.S.
banks may be losing market share.
U.S. banks do not report international commitments
to lend per se, but these can be approximated using
data from the Country Exposure Lending Survey
(CELS). Loan commitments are taken to be all contin
gent (off-balance-sheet) claims adjusted for guarantees,
less letters of credit.3 An increase in the ratio of loan
commitments to adjusted assets (on-balance-sheet
claims) is taken to indicate a shift to increased offbalance-sheet banking by U.S. banks.
The change in this ratio over three years suggests
that large U.S. banks are moving their international
activity off their balance sheets. The top nine U.S.
banks4 have raised the ratio of loan commitments to
adjusted claims from 35 percent to 43 percent between
end-1982 and end-1985 for 22 OECD countries and five
selected Asian borrowers— India, Indonesia, Malaysia,
South Korea, and Thailand (table).5
The rise in commitments to lend in relation to assets
is not uniform, however. The shift from on- to off-

3lt would be better if letters of credit adjusted for guarantees were
available.
4The top nine banks’ aggregate reported in the CELS continues to
include Continental Illinois.
5The most recent CELS data, from end-March 1986, show the same
ratio, 43 percent.
IN BRIEF—ECONOMIC CAPSULES

International A ctivity of the Top Nine U.S. Banks: Ratio of Loan Commitments to Adjusted Assets
(2 )

( 1)

End-1982
ratio

R anking/C ountries
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

New Zealand . . . .
Sweden ....................
A u s tra lia ....................
F in la nd .......................
D enm ark....................
T h a ila n d ....................
N o r w a y ....................
S w itz e rla n d .............
G reece ....................
United K ingdom . . .
In d ia ...........................
N e th e rla n d s .............
I c e l a n d ....................
F ra n c e .......................
M a la y s ia ....................
I t a l y ...........................
In d o n e s ia .................
C anada ....................
B elg iu m -L u x.............
West G erm any . . .
A u s tria .......................
Ire la n d .......................
S p a i n .......................
Japan .......................
P o rtu g a l....................
Korea .......................
T u rk e y .......................

.
.
.
.
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.
.
.
.
.

W eighted averages:
O v e r a ll.............................. . .
G-10 + Sw itzerland . . . . .
Other O E C D .................... . .
East A s i a t .......................

(3)

End-1985
ratio

Ranking/C ountries

1.11
0.74
0.71
0.58
0.51
0.50
0.49
0.49
0.48
0.45
0.45
0.41
0.39
0.38
0.37
0.33
0.32
0.32
0.28
0.26
0.24
0.21
0.18
0.13
0.12
0.11
0.04

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27

Sweden ....................
F ra n c e .......................
N o r w a y ....................
United Kingdom . . .
C anada ....................
A u s tra lia ....................
S w itz e rla n d .............
A u s tria .......................
N e th e rla n d s .............
B elgium -Lu x.............
S p a i n .......................
Ire la n d .......................
F i n la n d ....................
D enm ark....................
New Zealand . . . .
I c e l a n d ....................
I t a l y ..........................
West G erm any . . .
In d o n e s ia ................
In d ia ..........................
Greece ....................
T h a ila n d ....................
K o r e a .......................
Japan .......................
T u rk e y .......................
P o rtu g a l....................
M a la y s ia ....................

0.35
0.33
0.45
0.25

W eighted averages:
O v e r a ll.............................. . .
G-10 + Switzerland . . . . .
Other O E C D .................... . .
East A s ia :):.......................

.
.
.
.
.

. .
.
.
.
.

.
.
.
.

.
.
.
.
.
.
.

Percent
change,
1982-85*

(4)
Total NIFs
arranged,
1983-85(b illion s of dollars)

1.69
0.66
0.63
0.60
0.55
0.54
0.53
0.45
0.45
0.43
0.43
0.38
0.36
0.34
0.26
0.26
0.26
0.26
0.24
0.21
0.20
0.18
0.16
0.13
0.12
0.11
0.08

129
72
30
33
75
-2 4
10
90
9
56
137
81
-3 9
-3 4
-7 6
-3 2
-2 0
0
-2 6
-5 4
-5 8
-6 5
51
4
249
-1 4
-7 8

9.6
5.0
1.4
4.5
1.1
11.6
1.2
0.6
1.3
0.5
0.6
0.3
1.3
1.4
2.9
0.2
1.7
0.3
0.5
0.4
0.0
0.2
0.9
1.4
0.0
0.6
0.3

0.43
0.47
0.40
0.18

25
40
-1 0
-2 8

Totals:
49.8
26.6
20.9
2.3

'F o r countries in colum n (2).
fA rra n g e m e n ts with all banks, U.S. and non-U.S.
^In clu d e s India, Indonesia, Korea, Malaysia, and Thailand.
Sources: Federal Financial Institutions Examination C ouncil, Country Exposure Lending Survey; Bank of England; and Interna tiona l Financing
Review Data on NIFs include some arrangem ents with partial or without con tractu ally associated bank com m itm ents to lend; renegotiated
fa cilities (for exam ple, New Z e aland's) are double-counted.

balance-sheet exposure is concentrated in the G-10
countries, where the ratio of total commitments to total
assets rose from 0.33 to 0.47. In the G-10, the overall
growth of the market for loan commitments appears to
be more than offsetting any loss of share in that market
by U.S. banks.
By contrast, the balance is shifting toward on-balancesheet financing in Asia. Whereas commitments to the
five Asian borrowers stood at 25 percent of assets in
1982, they fell to 18 percent of assets in 1985. Since
Japanese banks are well represented as underwriters
of NIFs in Asia, U.S. banks may be losing market share
there. The same loss of market share may lie behind
the decline in off-balance-sheet exposure in the OECD

outside of the G-10, especially in Australia and New
Zealand.
In general, the countries whose borrowers have been
most active in the NIF market tend to rank higher in the
ratio of commitments to assets (table). Swedish, French,
Norwegian, and British borrowers have been active in
the NIF market and hold the top four positions;6 Japa
nese, German, and Italian borrowers have been rela
tively less active. The relative rise of commitments to
Canadian borrowers, even though Canadian withholding
tax makes NIFs unattractive, shows that NIFs are not
A u s tra lia n borrow ers have arranged a large sum of NIFs but have
drawn more heavily on them than most NIF custom ers. As a result,
the A ustralian ratio de clin ed over the three years.

FRBNY Quarterly Review/Summer 1986

a necessary vehicle for the shift to off-balance-sheet
banking.
In summary, CELS data provide evidence that large
U.S. banks are increasing their off-balance-sheet com
mitments to lend cross-border in relation to their actual
international assets. Further, these data locate the shift
in the G-10 countries. Finally, the rise in off-balancesheet exposure appears to be associated with activity
in the NIF market.

Robert N. McCauley

One important difference between U.S. travel and
merchandise trade is their regional compositions. The
Western Hemisphere’s share of U.S. travel is 53 per
cent, considerably more than its 37 percent share of our
merchandise trade (Table 2). Mexico alone accounts for
20 percent of overall travel trade, but only 6 percent of
merchandise trade. In contrast, Japan accounts for a
hefty 16 percent of U.S. merchandise trade, but only 7
percent of total travel expenditures. As explained later,
these regional differences suggest that the recent
decline in the dollar is effectively much less for U.S.
travel than for merchandise trade.
Historical perspective

Prospects for the
U.S. International
Travel Deficit
Travel is an important part of U.S. international trade
that has posted record deficits in recent years. From a
position of near balance in 1981, the travel and tourism
account approached a $10 billion deficit in 1985 (Table
1). Analysis of past experience confirms that travel flows
are heavily influenced by the same factors that deter
mine merchandise trade—in particular, the deterioration
in the U.S. travel balance during the 1980s is largely
attributable to the strong dollar and rapid real growth
relative to abroad. Though the dollar’s recent decline
against the Japanese yen and major European curren
cies should significantly lower the merchandise trade
deficit over the next several years, it is likely to lead to
only a modest improvement in the travel balance—
probably no more than $3 billion from economic factors
alone. This is because the dollar has not fallen against
the currencies of our primary travel partners, Canada
and Mexico.
The travel account measures spending by foreign
visitors to America and by U.S. travelers abroad on
items such as food, lodging, transportation, and enter
tainment. In 1985, foreign visitors spent a total of $14
billion here while U.S. travelers spent $24 billion abroad,
amounts that represent about 7 percent of U.S. mer
chandise exports and imports respectively. The travel
deficit was about 8 percent of the 1985 U.S. current
account deficit and accounted for almost half of the $20
billion deterioration in the U.S. services balance from
1981 to 1985.

As with U.S. trade generally, changes in the travel bal
ance are largely explained by relative income growth
and exchange rate movements. Rising incomes increase
demand for most goods and services, including travel.1
Travel expenditures are also sensitive to relative cost
considerations and hence changes in exchange rates.
A decline in the dollar tends to raise the cost of over
seas travel for U.S. citizens and makes the United
States more attractive to visitors from the rest of the
world. Prior studies suggest that a 1 percent decline in
the dollar can be expected to increase travel receipts
’ See, for instance, prior work by Jacques Artus, ‘A n E conom etric
A nalysis of International Travel," IMF S taff Papers (1972), pages 579614; Jane S. Little, “ International Travel in the U.S. B alance of
Payments," New E ngland Econom ic Review (M ay/June 1980), pages
42-55; and Jeffrey Rosensweig, "The Dollar and the U.S. Travel
D eficit," Econom ic Review, Federal Reserve Bank of A tlanta
(O ctober 1985), pages 4-13. Their find in gs suggest that a 1 percent
rise in a cou ntry’s real GNP can be expected to raise its travel
expenditures by som ewhat more than 1 percent.

Table 1

International Travel and
Passenger Fare Transactions
In billions of dollars
1977

1981

1985

Total travel paym ents ( - )
Passenger fares ..............................
Payments by U.S. travelers
in foreign countries ....................

10.2
2.7

16.0
4.5

23.8
7.3

7.5

11.5

16.5

Total travel receipts
Passenger fares ..............................
R eceipts from foreign
travelers in the United States . .

7.2
1.0

15.5
2.6

14.1
2.5

6.2

12.9

11.6

Net travel and
passenger paym ent deficit

3.0

0.5

9.7

Memo:
U.S. current account balance

-1 4 .5 *

6.3 -1 1 7 .8 *

'In d ic a te s deficit.

The author thanks B ernadette Alcam o for her research assistance in
the preparation of this article.

FRBNY Quarterly Review/Summer 1986

IN BRIEF— ECONOMIC CAPSULES

from abroad by somewhat more than 1 percent and to
reduce travel expenditures by U.S. citizens by less than
1 percent.2
Examination of the experience of the past decade
illu stra te s the response of the travel balance to
exchange rate movements and real growth here and
abroad (Chart 1). From 1977 to 1981 the travel deficit
declined from $3 billion to $0.5 billion. Although the
United States grew, on average, at a rate close to that
of other industrial nations, the travel account improved,
largely due to a 15 percent fall in the value of the dollar
against the currencies of our major travel partners
during this period.3 The deterioration in the travel deficit
since 1981 can be attributed, in large measure, to the
dollar’s appreciation (over 30 percent on both a traveland trade-weighted basis through mid-1985) and to
faster real growth in the United States than abroad.
From 1981 to 1985, total receipts from foreign travelers
declined by almost 10 percent while expenditures by
U.S. travelers abroad rose 44 percent.
Developments in our bilateral travel flows since the
late 1970s provide further evidence of how sensitive the
travel balance is to exchange rate movements and real
growth. Between 1977 and 1981, the U.S. travel deficit
showed significant improvement with all major partners
2D ollar de p re cia tio n will norm ally lower the dollar value of travel
paym ents less than proportionately. The resulting de clin e in real
exp enditures by U.S. travelers is pa rtially offset by the rise in the
dollar prices paid for food, lodg ing, and other items w hile abroad.
On the other hand, the effect of increased real exp enditures by
foreign travelers to the United States as a result of the falling dollar
is reinforced by any rise in the dollar price s paid.
3The w eights for the tra vel-w eig hted do lla r index com puted for this
article are bilateral shares in U.S. travel trade of eight major travel
partners: C anada, Mexico, the United Kingdom , France, Germany,
Italy, Japan, and the Bahamas. The index is in real terms, that is
with exchange rates adjusted by indexes of relative national price
levels.

whose currencies appreciated against the dollar (Ger
many, the United Kingdom, Japan, and Mexico). At the
same time, our balance with Canada, whose currency
depreciated against the U.S. dollar, deteriorated.
The worldwide appreciation of the dollar over 198185 accompanied a deterioration in our major bilateral
travel balances, as U.S. citizens stepped up their travel
throughout the world. In contrast to the general pattern,
however, our bilateral balance with Japan improved,
partly because Japan grew somewhat more rapidly (on
average) than the U.S. over this period, and because
the yen depreciated less against the dollar than the
major European currencies did.
Our travel balance with Mexico is particularly sensitive
to exchange rate movements because residents along
the border can readily cross the boundary to shop.
(Indeed, such spending amounts to over 50 percent of
the two countries’ bilateral travel trade.) As a result, a
real appreciation of the peso combined with rapid
Mexican growth produced a $1.6 billion improvement in
our bilateral balance from 1977 to 1981. The subse
quent sharp rise in the dollar’s real value during 198283 led to a $2.6 billion decline in this balance. In the
next two years our bilateral deficit changed little as an
appreciation of the peso through mid-1985 apparently
offset the effects of a weakening Mexican economy.
Prospects

Looking ahead, prospects for improvement in the U.S.
travel deficit seem limited because the dollar’s depre
ciation has been much less uniform than its prior
appreciation. The dollar declined approxim ately 30
percent in real terms against the currencies of West
Germany and Japan between the second quarters of
1985 and 1986. However, it has maintained its average
value in relation to Western Hemisphere currencies; in

Table 2

Regional Distribution of the U.S. Travel Balance: 1985
(excludes passenger fares)
In billions of dollars

Western H e m is p h e r e ...........................
C anada ...........................................
M exico ...........................................
C aribbean and Central A m erica.
W estern E u ro p e .....................................
Japan .....................................................

Travel paym ents

Travel receipts

Net balance

8.4
2.7
3.6
1.8
5.9
0.5

6.5
3.0
2.0
0.6
2.3
1.4

-1 .9
0.3
- 1.6
- 1.2
-3 .6
0.9

Percent
Percent
share of U.S.
share of U.S.
travel trade* m erchandise tra d e f
53
20
20
9
29
7

37
23
6
2
24
16

'B ila te ra l shares of U.S. travel flows (receipts plus payments, excluding passenger fares).
tB ila te ra l shares of U.S. m erchandise trade (exports plus im ports).

FRBNY Quarterly Review/Summer 1986

Chart 1

C h a rt 2

T h e U.S. T rav e l D eficit and the Dollar
B illio n s o f d o lla rs

Index, 1980=100

T ra d e -w e ig h t e d and T ra v e l- w e ig h te d
Real V alu es of the Dollar
Index, 1980=100

T5 0 --------------------------------------------------------------------------------

1976 77

* T ra v e l-w e ig h te d inde x of the d o lla r’s re al v a lue
(e xchang e rate ad ju sted by re la tiv e national p ric e
levels) a g a in s t c u r re n c ie s of eigh t m ajor U.S. tra v e l
p a rtn e rs . W eights a re b ila te ra l s h a re s o f U.S. travel.
^S e a s o n a lly a d ju s te d annual ra te s ; minus indica tes
surplus.
S ources: U.S. D epartm ent of C om m erce and
Inte rn a tio n a l M onetary Fund, International Fin ancial
S ta tis tic s .

fact, the dollar has appreciated sharply against the
Mexican peso. As a result, the real value of the dollar
has fallen only 6 percent from last year’s average on a
travel-weighted basis. This is significantly less than its
decline during 1977-81 and reverses only about one-fifth
of its rise since 1981. In contrast, the trade-weighted
dollar has declined substantially, offsetting more than
one-half of its prior rise as measured by most published
indexes (Chart 2).
The extent to which the travel balance responds to
the dollar’s decline depends on the pace of real growth
here and abroad. Assuming that the United States and
other industrial nations grow at close to their potential
rates (which implies increases in domestic demand
growth rates abroad compared with the average of recent
years), we could expect a modest decline of $1.5 to $3.0
billion in the travel deficit over the coming two years.4 Most
U n d e rly in g this analysis is the assum ption that real travel
e xp enditu res have an ela s tic ity of roughly 1.5 relative to incom e
grow th and real exchange rate m ovements.

1976

* R e a l e ffe c tiv e tra v e l-w e ig h te d in d e x o f the d o lla r's value
using b ila te ra l sh a re s of eigh t m a jo r U.S. tra v e l p a rtn e rs.
^M o rg a n G ua ran ty real e ffe c tiv e (b ila te ra l) index o f the
d o lla r’s value a g a in s t c u rre n c ie s of eleven m a jo r U.S.
m e rchan dise trade p a rtn e rs.
S ources: M organ G uaranty T ru st C om pany and
Interna tiona l M onetary Fund, Inte rn a tio n a l Financial
S tatistics.

of this decline should result from increased visits to the
United States, particularly from Western Europe and
Japan. Travel receipts might increase by as much as 20
percent annually over this period; spending by U.S.
citizens abroad is likely to continue to increase, but at
a slower rate than in recent years.
Of course, the actual improvement in the travel bal
ance may d iffe r sig n ifica n tly from this estim ate,
depending on a variety of factors that are difficult to
predict. For example, a pickup in foreign relative to U.S.
growth would probably result in further improvement of
the travel balance. Political factors also can be important
influences on travel flows. In particular, recent events
suggest that concerns over terrorism could sig n ifi
cantly reduce U.S. travel abroad this year and next.
If so, the decline in the travel deficit may be substan
tially greater than economic factors alone would sug
gest—although modest, in any case, in relation to the
overall current account deficit (presently over $120 bil
lion annually).
Bruce Kasman

FRBNY Quarterly Review/Summer 1986

IN BRIEF— ECONOMIC CAPSULES

February-April 1986 Report
(This report was released to the Congress
and to the press on June 4, 1986)

Iteasury and Federal Reserve
Foreign Exchange Operations
The dollar moved substantially lower against all major
currencies during the three months ended April,
declining by more than 8 percent against most European
currencies and by almost 12 percent against the Jap
anese yen. The depreciation was a continuation of the
trend that emerged in early 1985 and gained momentum
after the September G-5 meeting. The U.S. authorities
did not intervene for their own account in the exchange
markets during the February-April period, and the dol
lar’s decline proceeded without any new concerted
intervention in the exchange markets.
The decline took place against a background of
downward-moving interest rates and narrowing interest
rate differentials favoring the dollar. The market’s atti
tude also seemed influenced by the implications of
continuing large current account imbalances and by
assessments of the varying inflation prospects in dif
ferent countries as a result of oil price changes. Market
participants were also influenced by perceptions of
official views about the dollar, particularly in light of the
January meeting of G-5 Finance Ministers and Central
Bank Governors where they expressed satisfaction with
the trend of exchange rates since the September
meeting.
As the period opened, the foreign exchange markets
were in a sensitive phase. The dollar had already
declined a substantial amount from its highs of the year
before. Yet there was little evidence that the U.S.

A report by Sam Y. Cross, Executive Vice President in charge of the
Foreign Group at the Federal Reserve Bank of New York and
Manager of Foreign Operations for the System Open Market
Account. Willene A. Johnson was primarily responsible for the
drafting of this report, assisted by Elisabeth S. Klebanoff.

economy or its trade position had reaped enough benefit
to allay concerns among market participants that U.S.
authorities would wish to see further dollar depreciation.
Indeed, statistics then being released indicated the U.S.
deficit on merchandise trade had widened substantially
during December, and the external sector continued to
exert a significant drag on domestic production. The
mid-February estimate of GNP growth for the final
quarter of 1985 revealed a disappointing, downward
adjustment. Statistics for January’s personal income and
consumption were also worse than expected, leading
many in the market to conclude that the outlook for
1986 was even less optimistic than they had believed
earlier. Subsequent data were interpreted as indicating
points of weakness, rather than strength, for the U.S.
economic outlook.
At the same time, concern faded about potentially
adverse effects on prices of any continuing dollar
depreciation. Inflationary expectations in the United
States were being rapidly scaled back in response to
further dramatic drops in prices for oil and some other
commodities. Thus, the risk of an accelerating decline
in dollar exchange rates was seen by many market
observers to be less than before. In addition, through
early April, interest rates in the United States continued
to ease at a faster pace than those abroad as many
market participants expected that the Federal Reserve
might ease monetary policy considerably. As a result,
interest differentials favoring the dollar continued to
narrow. By April 16, interest differentials on long-term
government securities fell to within 200 basis points visd-vis German and Japanese securities.
Beginning in February, a heavy schedule of state

FRBNY Quarterly Review/Summer 1986

ments and testimonies about U.S. economic policies,
together with the approach of a number of important
international meetings, provided opportunities for offi
cials here and abroad to be questioned about their
attitude toward exchange rate trends. Administration
officials generally expressed satisfaction with the decline
in dollar rates since the G-5 agreement in September.
They noted that the dollar’s decline had been orderly
and denied that there was any particular level or range
of rates at which they expected or desired the dollar to
trade. The President, in his annual State of the Union
address, noted the problems that previous exchangerate fluctuations had caused for many Americans and
asked Secretary of the Treasury Baker to determine
whether it would be useful to hold an international

C hart 1

T h e d o lla r c o n tin u e d to decline.
P e rc e n t *

J ap anese

i
A i

Y\
x 'H i

P <

fri \

Pound
ste rlin g

\— r f V *

S w iss

French
fra n c /\

\
G erm ar

mark

i ii i Im
M
J

lu i la i d ii ill i i L u i i l i . i i
J
A
S
O
N
D
J
1985

conference to discuss with other countries the role and
relationships of currencies. Market participants and
journalists tended to conclude from these statements
that U.S. authorities would welcome a further depre
ciation of the dollar, notwithstanding the repeated deni
als by U.S. officials that they had a target for the dollar.
In these circumstances, the market’s attitude toward the
dollar was predominantly bearish.
During February and early March, the dollar declined
across the board without interruption. The focus of
market attention initially was the yen. It was supported
by mounting monthly trade surpluses and the view that
declining oil prices would be particularly beneficial to the
Japanese economy. The dollar, which had closed at
¥191.4 0 at the end of January, declined steadily
throughout early February to breach the psychologically
important ¥180 level and to reach ¥177.40, a seven and
one-half year low, by February 19. The yen’s rise then
stalled after Japan’s Finance Ministry confirmed that it
was developing plans to ease regulations on capital
outflows which could have the effect of raising the
demand for dollars by Japanese institutional investors.
At this point market participants shifted their focus to
the dollar/mark exchange rate. Forecasts of German
economic growth in 1986 were being revised upward
toward 4 percent. Given the relatively lackluster growth
performance in the United States during the last quarter
of 1985 and the weakness in several U.S. business
indicators early in 1986, Germany emerged as a likely

y ’Sv
I m 11111111 i I ll .l
F

M
A
1986

U nited S ta te s m o n e ta ry a u th o r itie s did
not in te r v e n e F e b ru a ry th ro u gh April.

Chart 2
E x c h a n g e m a r k e t p a rtic ip a n ts fo c u s e d
on th e policy im p lic a t io n s of p e rs is te n t
t r a d e im b a la n c e s .
B illio ns of U.S. d o lla rs
Me nth ly tra d e bal ances *

B illio n s of U.S. d o lla rs e q u iv a le n t

4 -------------------------------------------------------------------------------------------N et p u rc h a s e s of fo re ig n c u rre n c ie s

3 ----------------------- ------------------- --------------------------------------------2 -----------------------

---------------------------------------------

1-----------------

--------------------------------

o l_______2______ J ____________ 11.~ — - — :: •„_______Q_____
May 1985Jul 1985

Aug 1985O c t 1985

Nov 1985Jan 1986

Feb 1986Apr 1986

...

Japan
* P erce ntag e cha nge of w eekly ave ra g e ra te s fo r
d o lla rs from the a v e ra g e ra te fo r the w eek ending
A p ril 25, 1985. F ig ures c a lc u la te d from N ew York
noon qu otatio ns.

FRBNY Quarterly Review/Summer 1986

G erm an y

U n ite d S ta te s

♦ A v e ra g e o f th e se a s o n a lly a d ju s te d tra d e b a la n ce s fo r
D e c e m b e r 1985 and Jan uary and F e b ru a ry 1986.

candidate for relatively high growth. Moreover, the
German trade surplus had been increasing during the
previous few months. The continued drop in oil prices
made it appear possible that Germany’s trade surplus
for 1986 would be double the level of the previous year.
Traders also anticipated a possible realignment within
the European Monetary System (EMS) in which the
mark would be revalued. With these expectations in the
minds of market participants, the dollar declined against
the mark until early March as far as DM2.1960.
On March 6 and 7, the central banks of Germany,

C h a rt 3

The de clin e in w orld oil prices.
D o lla rs p e r b a rre l
35-

25-

20-

B re n t oil
n .
\
WestTexasN^
V

15-

\
! i i i I i.i i
10L
Nov
D ec
1985

M i l l
Jan

V "

1 1 1 l._ l...1 1 Q - L . l
Feb
M ar
Apr
1986

Japan, the United States, France, and the Netherlands
lowered their official interest rates. The central banks
of the United Kingdom, Italy, and Sweden soon followed
by cutting their official lending rates. These actions had
little immediate impact on exchange rates since the
reductions had been widely anticipated and international
differentials in market rates were expected to be largely
unchanged. But the concerted round of interest rate
reductions underscored to the market the potential for
further coordinated policy actions. For some time
thereafter, most dollar exchange rates moved narrowly
as markets reassessed the near-term outlook for the
dollar and the prospects for other coordinated actions.
Early in April, the dollar resumed its decline. Indica
tions that world oil exporters were failing to agree on
a plan to support oil prices kept alive expectations that
inflation in the major countries would continue to slow.
Thus dealers anticipated that the major central banks
would act soon to cut interest rates again. Some market
participants even thought that the Federal Reserve
might cut its discount rate more than other central
banks. In the United States, there was concern about
the impact of lower oil prices on U.S. banks with expo
sure to the oil-exporting developing countries and to the
domestic energy sector. In Germany, the central bank
had expressed doubts that a further cut in the discount
rate would be appropriate given existing conditions in
the domestic economy.
Selling pressures against the dollar then emerged,

C hart 4

showed th ro u g h in low er rates
of in fla tio n .

In te re st d iffe re n tia ls fa vo ra b le to the d o lla r
narrow ed as long-term in te re s t rates
declined in the United States more ra p id ly
than abroad.

P e rc e n t
12
------------- ------------------C o n su m e r p ric e inde x
T w e lv e -m o n th e rc e n ta g e c h a n g e

P e rce n t
14
Y ie ld s on te n -y e a r m a tu ritie s

10 -

Mi l ,

S^ ' ^ SN ^ U .S . Treasu ry bonds
F rance

o----- —0-—_

U n ite d St at es
Jap an

G erm any

G erm an p u b lic -s e c to r bonds

—
Ja p a n e se G ove rnm ent bonds

----

— 2 L

Nov

Dec

Jan

1985

1............ 1....
Feb
M ar
1986

Apr

.... J

i
F

1
A

i
M

i
J

1
i
i
1
J
A S O
1985

i
N

i
D

_L_i

I I

F M A
1986

FRBNY Quarterly Review/Summer 1986

especially against the yen. That currency had already
appreciated significantly during the past year, and Jap
anese exporters, particularly small- and medium-size
firms, were facing a substantial drop in the demand for
their products. Under these circumstances, Japanese
officials began to voice concern whenever the yen
advanced toward the level of ¥175 against the dollar.
Dealers therefore remained wary that foreign exchange
market intervention or some other official action might
be taken to curb the exchange rate move. But the
demand pressures on the yen were becoming so strong
as to bring into question the Japanese authorities’ ability
to resist its appreciation unilaterally, and the yen
strengthened relative to all major currencies.
Against the major European currencies, the dollar was
influenced early in April by strains emanating from a
realignment of exchange rates within the joint inter
vention arrangement of the EMS. Market participants
had long expected that some adjustment in exchange
rates might occur soon after the March elections in
France to adjust for inflation differentials that had
existed between the participating countries since the
general realignment in 1983. As such an event was
thought to become more imminent, the dollar’s decline
against the mark slowed. Some dealers were antici
pating that the dollar might benefit from speculative
reflows out of the mark after the realignment. But the
realignment that occurred over the first weekend in April
unleashed a strong demand for French francs against
all currencies. French residents unwound long-standing
commercial leads and lags and nonresidents sought

quickly to build up investment positions in francs and
to benefit from the relatively high interest rates in
France. As a result, the dollar again came on offer
against the European currencies after the realignment.
By late April, the dollar had declined 10 percent
against the Continental currencies and 13 percent
against the yen from end-January levels. It touched
¥166.10 against the yen, a record low for the postwar
period, and DM2.1520 against the mark, the lowest level
against the German currency since April 1981. At this
point the dollar was more than 37 percent below its
highs vis-a-vis those currencies of about a year before.
Meanwhile, the financial markets around the world
were astir with talk of a renewed drop in interest rates
in many major countries. The plunge in oil prices was
beginning to show through in reduced inflation rates for
a number of countries. Partly in sympathy, interest rates
on U.S. long-term securities had declined almost to the
level of the Federal funds rate, indicating strong market
expectations that a further cut in the Federal Reserve’s
discount rate was in the offing. Japanese bond yields
fell to postwar lows. And in Europe, a number of coun
tries that had not participated fully in the easing of
interest rates prior to the EMS realignment were seen
as having increased scope to cut rates.
In fact, the moderation of global inflationary expec
tations did provide the impetus for a lowering of official
interest rates in a number of countries. The Federal

C hart 6

O ffic ia l in te r e s t ra te s w e re lo w e re d .
C h a rt 5

P e rce n t

In th e United S ta te s , lo n g -te rm yields fell
m o re ra p id ly than s h o rt-te rm yields.

U nited K ingdom

P e rc e n t

.J "

"L
L

Fra nee

I-----------1

|___
L
U nited S tate s

I—

——■i

Japan
i_________ i
G erm any
I
Sep
Y ie ld s on a c tiv e ly -tra d e d is s u e s a d ju s te d to c o n s ta n t
m a tu ritie s .

FRBNY Quarterly Review/Summer 1986

I
O ct
1985

I
Nov

1
Dec

Jan

1
Feb
1986

1
M ar

I
Apr

Net Profits ( + ) or Losses ( - ) on United
States Treasury and Federal Reserve Current
Foreign Exchange Operations
In m illions of dollars

Period
February 2, 1986A pril 30, 1986 ....................
Valuation profits and losses
on outstanding
assets and lia b ilities
as of A pril 30, 1986* . . . .

Federal
Reserve

United States Treasury
Exchange
S tabilization Fund

- 0-

+ 962.9

+ 1,031.4

Data are on a value -da te basis.
'V aluation gains represent the increase in the do lla r value of
ou tstand ing c u rrency assets valued at en d-of-period
exchange rates, com pared with the rates prevailing at the
tim e the foreign currencie s were acquired.

Reserve and the Bank of Japan both lowered their dis
count rates by one-half of a percentage point, effective
April 21. Central banks in the United Kingdom, France,

and Italy lowered their official rates in one or two steps
by one-half of a percentage point or more at about the
same time. The German central bank, feeling more
constrained by the weakness of the mark within the
EMS, did not join in this second round of reductions of
official lending rates.
At the end of April, the dollar’s decline paused. Some
foreign exchange and bond dealers had expressed
concern that investors might be reluctant to acquire the
additional dollar-denominated assets needed to finance
the continuing U.S. current account and fiscal deficits.
Market participants were also well aware of pressures
on foreign governments to ease the pain for their own
industries of too-rapid a fall of the dollar. Accordingly,
they considered the possibility that some agreement to
support the dollar might emerge from the discussions
at the Economic Summit in Tokyo early in May. In
addition, by late April, there was evidence that foreign
participation in U.S. securities markets continued to be
strong. The dollar therefore moved up somewhat from
its lows by the close of the period under review.

The EMS Realignment

On April 6, the European Community announced a realign
ment of the central rates within the EMS. This was the first
overall realignment of EMS central rates in more than three
years. The realignment was initiated by the new French
government as part of its program to restore competitiveness
in the French economy and dismantle exchange and other
financial controls. It involved in effect a devaluation of almost
6 percent for the French franc’s bilateral central rates against
the German mark and the Dutch guilder. Other adjustments
against the mark and the guilder were much smaller,
including 3 percent effective devaluations in the central rates
of the Italian lira and the Irish pound and 2 percent for the
Danish krone and the Belgian franc.
Movements of the market exchange rates for the EMS
currencies after the realignment were relatively small. The
French franc depreciated 31/2 percent against the mark,
substantially less than the 6 percent devaluation of its bilat
eral rate with the German currency. In other respects the
configuration of exchange rates within the EMS also showed
only modest change—except for the German mark, together
with the Dutch guilder, which moved from the top to the
bottom of the narrow band. Heavy reflows of funds following
the realignment were reflected in substantial changes in the
foreign exchange reserves of several countries. German
reserves declined by almost $4 billion equivalent during the
two weeks following the realignment, although much of this
decline was recouped later in the month. At the same time,
there were large increases in the foreign-currency reserves
of France and Italy during April. These two countries, as well
as the other countries whose EMS central rates were
effectively lowered against the guilder and the mark, took
advantage of the relief from exchange market pressure late
in April to add reserves, ease exchange controls, and lower
interest rates.

C h a rt 7

T he G e rm a n m a rk m o v e d to th e b o tto m
of th e EMS a fte r th e re a lig n m e n t.
EMS
re a lig n m e n t

P e rc e n t*

Feb

Mar

A pr

1986
W ee kly a ve ra g e s of d a ily 9 a.m. ra te s.
♦ P erce ntag e d e v ia tio n o f ea ch c u rre n c y from its ECU
c e n tra l ra te. D o tte d lin e s c o rre s p o n d to th e S y s te m 's
2 % p e rc e n t lim it on m ovem ent fro m b ila te ra l ce n tra l
e x c h a n g e ra te s fo r all p a rtic ip a tin g c u rre n c ie s e x c e p t
the Italian lira. The lira may flu c tu a te 6 p e rc e n t fro m
its c e n tra l ra te s a g a in s t o th e r EMS c u rre n c ie s .

FRBNY Quarterly Review/Summer 1986

Subscriptions to the Quarterly Review (ISSN 0147-6580) are free. Multiple copies
in reasonable quantities are available to selected organizations for educational pur
poses. Write to the Public Information Department, 33 Liberty Street, New York, N.Y.
10045 (212-720-6134). Single and multiple copies for United States and for other
Western Hemisphere subscribers are sent via third- and fourth-class mail, respectively.
All copies for Eastern Hemisphere subscribers are airlifted to Amsterdam, where
they are then forwarded via surface mail. Multiple-copy subscriptions are packaged
in envelopes containing no more than ten copies each.

Quarterly Review subscribers also receive the Bank’s Annual Report.

Quarterly Review articles may be reproduced for educational or training purposes
only, providing they are reprinted in full, distributed at no profit, and include credit to
the author, the publication, and the Bank.

Library of Congress Catalog Card Number: 77-646559

FRBNY Quarterly Review/Summer 1986

A 15-chapter book entitled Recent Trends in Commercial Bank Profitability—A Staff Study was recently
completed by the Federal Reserve Bank of New York.
The book traces the performance of the banking industry over the past decade. Among the areas cov
ered are competition from foreign and domestic banking institutions and nonbank financial institutions,
as well as the effects of inflation, interest rates, loan loss provisions, and deregulation on bank earnings.
It includes a review of recent literature on the subject and commentary by leading bankers and financial
analysts on the condition of the banking industry.
The new 379-page bound publication is available for $10 a copy from the Public Information Department,
Federal Reserve Bank of New York, 33 Liberty Street, New York, N.Y. 10045. Checks or money orders
should be made payable to the Federal Reserve Bank of New York. Payment will be accepted in U.S.
dollars only; if ordering from outside the United States, please use a check or money order drawn on a
U.S. bank or its foreign branch.

Full text of Economic Policy Review (Federal Reserve Bank of New York) : Summer 1986, Vol. 11, No. 2

FRASER