Monetary policy and bank lending.
Bernanke, Ben
Over the last decade or so, NBER researchers have analyzed the effect
of financial conditions, particularly the characteristics of
borrowers' and lenders' balance sheets in macroeconomic performance.(1) The result, I believe, is a much richer understanding of
the sources of aggregate fluctuations. For example, ten years ago the
leading theory of the determination of business fixed investment implied
that the main factors affecting investment were the marginal product of
capital and the real interest rate. Absent large unexplained
fluctuations in capital's marginal product, this neoclassical theory was hard pressed to explain the volatility of investment over the
business cycle. Recent empirical research has shown instead that balance
sheet conditions, such as the firm's indebtedness or its access to
internal finance, have a major influence on its investment decisions.(2)
It seems likely that cyclical changes in balance sheets or cash flows,
perhaps in interaction with the illiquid nature of capital investments,
ultimately will provide an explanation of the cyclical volatility of
investment spending that is superior to what the neoclassical model can
offer.
The research I report on here, on the link between monetary policy
and commercial bank lending, is a relatively small subtopic of this
larger research program on financial-real interactions. It shares with
the larger program the idea that balance sheets matter, as well as the
broader conviction that--because the structure of financial arrangements
affects the transmission of information and the incentives of market
participants--financial conditions do have effects on the real economy.
The Channels of Monetary Transmission: Money Versus Credit
The question I take up here is a very old one: how does monetary
policy affect aggregate demand? The standard answer is that the Federal
Reserve works its magic by changing the supply of the medium of exchange
relative to the demand. According to this story, to slow down the growth
of aggregate demand (for example), the Fed uses open market sales to
drain reserves from the banking system, reducing the money supply. This
shortage of liquidity is presumed to drive up short-term--and possibly,
through expectational effects, longer-term--interest rates. Higher
interest rates are then presumed to depress aggregate demand by raising
the cost of funds relative to the returns to capital (including housing
and consumer durables). I will refer to this standard channel as the
"money channel" of monetary transmission.
Without necessarily denying the existence of this conventional money
channel, recent research has addressed the possibility that there is an
additional channel of monetary policy transmission, which I will refer
to as the "credit channel." In contrast to the money channel,
which operates through the liabilities side of bank balance sheets
(deposits), the credit channel (if it exists) operates through the asset
side of the bank balance sheet (loans and securities). This credit
channel relies on two assumptions.
The first is that banks do not treat loans (for example, to
commercial and industrial firms) and securities (such as Treasury bills)
as perfect substitutes in their portfolios. This assumption is quite
realistic: banks hold securities primarily for liquidity, for
collateral, and to satisfy various legal requirements, while loans are
held primarily for their expected return.
The second assumption is that potential borrowers, such as business
firms, are not indifferent between bank loans and the issuance of
open-market securities, such as equities or corporate bonds, as a means
of raising funds. Again, this assumption is realistic: many firms,
especially smaller ones, have essentially no access to open-market
credit and must rely entirely on banks or other intermediaries for
funds. In part, this is because of the large fixed costs of open-market
issues, as well as because of the comparative advantage that banks have
developed in assessing the quality of business loans, which reduces the
net cost of borrowing through a bank. Even large firms do not appear to
be indifferent about their sources of funds: for large firms, bank loans
may provide short-run liquidity not available from public issues.
Studies of stock market data also have shown that the announcement of
large new bank loans raises equity values, suggesting that bank loans
are a useful method of signaling to the market that the firm's
prospects are good.
Adding these two plausible assumptions to the standard macro model
leads to a new channel of monetary policy transmission: the credit
channel.(3) The credit channel works as follows: suppose again that the
Federal Reserve wants to depress aggregate spending, and therefore has
drained reserves from the banking system. To the extent that the loss of
reserves forces a contraction in bank liabilities (deposits), it must
lead simultaneously to a parallel contraction in bank assets (loans and
securities). If loans and securities are not perfect substitutes for
banks, then in general, as banks lose deposits, they will try to reduce
both categories of assets. In particular, banks may cut down new
lending, fail to renew old loans, and in extreme cases even call
outstanding loans.
If firms were indifferent about their source of finance, then a
cutback in bank lending would not affect their behavior; they simply
would switch to alternative credit sources. However, if alternative
forms of credit are more expensive to the firm, or simply are not
available,(4) then a drying up of bank lending may force firms to cancel
or delay capital projects, reduce inventories, or even cut payrolls. In
short, the credit channel story says that contractionary monetary policy can force banks to cut loans, and that reduced bank lending in turn
impels firms that are wholly or partially dependent on banks for credit
to reduce their spending. Similar effects could operate in the consumer
sector, to the extent that households are directly or indirectly
dependent on banks for certain types of credit.(5)
The existence of a credit channel does not rule out the simultaneous
existence of a money channel. Indeed, there could be other modes of
monetary transmission operating as well: for example, the research on
investment alluded to in my introduction suggests that changes in
nominal interest rates may affect investment spending via their impact
on balance sheets or cash flows, as well as through the familiar
"cost-of-capital" effect. Putting these various channels
together yields a picture of the operation of monetary policy that is
potentially much richer than the simple textbook analysis.
Evidence for the Credit Channel
Does a tightening of policy by the Federal Reserve lead to a
reduction in bank lending, as required by the credit channel story? Does
a fall in bank lending lead potential borrowers to reduce their
spending?
A number of researchers have investigated the timing between monetary
tightening and bank lending. Blinder and I found that, during the
pre-1980 sample period, a tightening of monetary policy (as indicated by
a rise in the federal funds rate) was followed in subsequent months by a
decline in bank deposits and a matching decline in bank holdings of
securities.(6) Bank loans did not fall during the first months after a
tightening; indeed, initially, loans rose slightly. However, within six
to nine months, banks began to rebuild their security holdings and to
reduce lending substantially. The timing of the fall in lending
corresponded closely to a rise in the unemployment rate. Blinder and I
interpreted this pattern as being consistent with the credit channel
story, arguing that the relatively slow reaction of lending was the
result of costs that banks faced in adjusting their loan portfolios in
the short run.
A potential problem with our interpretation is that a similar pattern
might arise if only the money channel was operative. Suppose for example
that a Fed tightening raised interest rates and induced firms to delay
investment projects. Then we would expect again to see a decline in bank
lending follow the tightening of policy, except that in this case the
decline in lending would be the result of a fall in borrowers'
demand for loans, rather than a reduced willingness of banks to lend.
In an interesting attempt to resolve the identification problem, Anil
K. Kashyap, Jeremy C. Stein, and David W. Wilcox examined the behavior
of alternatives to bank lending following episodes of monetary
tightening.(7) They argued that, if the source of the lending slowdown
was a reduction in loan supply, as implied by the credit channel theory,
then nonbank sources of credit should rise following policy tightening,
as firms looked to alternative lenders. In contrast, if the reason for
the slowdown in lending was a decline in credit demand, then all forms
of credit extension should fall after a tightening of policy. These
authors' empirical results favored the credit channel view, as they
found that issuance of commercial paper in particular has tended to rise
sharply following a tightening of policy. Mark Gertler and Simon
Gilchrist found that much of the Kashyap-Stein-Wilcox result was driven
by the relatively more severe impact of monetary tightening on small,
bank-dependent firms (as opposed to larger firms with access to
commercial paper markets).(8) Gertler and Gilchrist interpreted their
results as being consistent with a combination of the existence of a
credit channel for monetary policy and a tendency for small firms to be
financially weaker than larger firms.
Another type of evidence favoring the existence of a credit channel
follows from the cyclical behavior of various interest rate spreads. For
example, several researchers have found that the spread between the
commercial paper (CP) rate and the interest rate on Treasury bills of
similar maturity has remarkably strong forecasting power for the real
economy; in particular, an increase in the CP rate relative to the
T-bill rate signals an impending recession.(9) While several factors may
explain the predictive power of this interest rate spread, one
possibility is the operation of the credit channel of monetary
transmission: a tightening of policy that reduces bank lending should
lead both to a wider spread between the CP rate and the T-bill rate, as
a constricted loan supply forces borrowers to rely more heavily on
commercial paper issuance, and to a subsequent economic downturn.
Indeed, I show that this particular interest rate spread is closely
linked to indicators of monetary policy.(10) Similar results, with a
similar interpretation, obtain for other spreads such as the spread
between the rate on bank CDs and the T-bill rate.
If we accept that a tightening of monetary policy reduces the supply
of bank loans, there still remains the question of whether potential
borrowers are forced to reduce spending, or whether they can switch
without significant costs to alternative credit sources. The evidence
here is more limited but suggests that there do exist
"bank-dependent" borrowers who face significant costs of
finding alternative sources of credit. Gertler and Gilchrist's
finding that small firms suffer disproportionately from episodes of
monetary stringency is consistent with this view. Similarly, Kashyap,
Owen A. Lamont, and Stein found that bank-dependent firms, defined to be
firms without bond ratings and with low reserves of liquidity, were much
more likely than other firms to shed inventories during a period of
tight money.(11)
The evidence that I have described here briefly is, of course, drawn
from historical episodes. It is possible that, whatever the relevance of
the credit channel in the past, institutional changes that have occurred
over the last 20 years or so may have rendered the credit channel
inoperative by now. For example, Christina D. Romer and David H. Romer
have stressed the significance of the elimination of reserve
requirements on banks' managed liabilities; now, if an open market
sale drains reserves and reduces banks' core deposits, banks that
wish to make loans still can obtain funds by issuing large-denomination
CDs. Other institutional changes that may have weakened the credit
channel are the increasing ability of banks to securitize assets and the
proliferation of alternative sources of commercial credit, including,
among others, junk bonds, finance companies, and asset-backed commercial
paper.
While the strength of the credit channel no doubt has changed, I
suspect that it remains a nontrivial part of the monetary transmission
mechanism. The ability of banks to issue large CDs at a given interest
rate, for example, likely is limited by the depth of the market and
concerns about bank risk, while economic theory suggests strongly that
there is a core of information-intensive loans that will be difficult to
securitize. On the borrower aide, small firms still do not have access
to junk bonds or asset-backed commercial paper; and finance companies,
which do service small firms, have focused on more standardized lending
rather than customized, information-intensive loans. Also, it should be
remembered that institutional changes, particularly the offering by
nonbanks of close substitutes for bank deposits, likewise have
potentially affected the strength of the conventional money channel, so
that it may be that the relative importance of the credit channel has
not declined. In any case, we can hope that research on the alternative
channels of monetary transmission will be of some help to policymakers
as they try to cope with a rapidly changing financial environment.
The Credit Channel and the 1990 Recession
One of the striking features of the 1990 recession, documented by
Cara S. Lown and me, among many others, is the sharp decline in bank
lending.(12) Does this decline reflect the operation of the credit
channel?
Interestingly, the answer appears to be no. Indeed, the 1990
recession may be one of the few postwar recessions in which a decline in
bank lending engendered by tight monetary policy did not play a part.
Evidence for the claim that the credit channel was not in operation
during this recession includes: the decline of the federal funds rate
well in advance of the recession; an apparent reluctance of banks to
issue managed liabilities (contributing to slow M2 growth); weak growth
of commercial paper and other nonbank forms of credit; and an unusual
prerecession decline in the spreads between the CP and CD rates on the
one hand and the T-bill rate on the other. Rather than the credit
channel of monetary policy, Lown and I argue that two other types of
financial developments contributed to the recession that began in 1990:
first, a shortage of bank capital that constrained bank lending,
particularly in the Northeast; second and more importantly, weakness in
borrowers' balance sheets arising from high levels of indebtedness.
To the extent that our diagnosis is correct, the 1990 recession
illustrates the point made at the beginning of this article: that
balance sheets play a potentially important role in business cycle
dynamics.
1 Recent NBER volumes bearing on this theme include: Asymmetric
Information, Corporate Finance, and Investment (1990) and Financial
Markets and Financial Crises (1991), both edited by R. G. Hubbard and
published by the University of Chicago Press for the NBER.
2 See, for example, S. Fazzari, R. G. Hubbard, and B. C. Petersen,
"Financing Constraints and Corporate investment," Brookings
Papers on Economic Activity (1988:1), pp. 141-195.
3 For a simple formal analysis, see B. S. Bernanke and A. S. Blinder,
"Credit, Money, and Aggregate Demand," NBER Reprint No. 1205,
June 1989, and American Economic Review Papers and Proceedings, 78, 2
(May 1988), pp. 435-439. An additional necessary assumption is that the
Fed can affect the quantity or cost of funds available to the banking
system.
4 The literature on the credit channel often has been related to the
literature on credit rationing. However, credit rationing, in the sense
of strict limits on credit availability, is not needed for this channel
to operate; it is sufficient that credit obtained from alternative
sources be more expensive than the bank loan.
5 The idea that bank lending plays a role in monetary transmission is
not new; it was discussed under the rubric of the "availability
doctrine" in the 1950s and can be found even in Keynes. However,
because of developments in the economics of imperfect information and in
econometric techniques, the theoretical and empirical bases for the idea
are noticeably stronger than in the past.
6 B. S. Bernanke and A. S. Blinder, "The Federal Funds Rate and
the Channels of Monetary Transmission," NBER Working Paper No.
3487, October 1990, and in American Economic Review 82, 4 (September
1992), pp. 901-921.
7 A. K. Kashyap, J. C. Stein, and D. W. Wilcox, "Monetary Policy
and Credit Conditions: Evidence from the Composition of External
Finance," NBER Working Paper No. 4015, March 1992, and in American
Economic Review, forthcoming.
8 M. Gertler and S. Gilchrist, "Monetary Policy, Business
Cycles, and the Behavior of Small Manufacturing Firms," NBER
Working Paper No. 3892, November 1991.
9 B. M. Friedman and K. N. Kuttner, "Money, Income, and Prices
After the 1980s," NBER Reprint No. 1731, July 1992, and J. H. Stock
and M. W. Watson, "New Indexes of Coincident and Leading Economic
Indicators," in NBER Macroeconomics Annual, Volume 4, O. J.
Blanchard and S. Fischer, eds., Cambridge, MA: MIT Press, 1989.
10 B. S. Bernanke, "On the Predictive Power of Interest Rates
and Interest Rate Spreads," NBER Working Paper No. 3486, October
1990, and in New England Economic Review, Federal Reserve Bank of Boston (November-December 1990), pp. 51-68. See also B. M. Friedman and K. N.
Kuttner, "Why Does the Paper-Bill Spread Predict Real Economic
Activity?" NBER Working Paper No. 3879, October 1991, and in New
Research in Business Cycles, Indicators, and Forecasting, J. H. Stock
and M. W. Watson, eds., Chicago: University of Chicago Press,
forthcoming.
11 A. K. Kashyap, O. A. Lamont, and J. C. Stein, "Credit
Conditions and the Cyclical Behavior of Inventories: A Case Study of the
1981-2 Recession," NBER Working Paper No. 4211, November 1992; also
summarized in NBER Digest, February 1993.
12 B. S. Bernanke and C. S. Lown, "The Credit Crunch,"
Brookings Papers on Economic Activity (1991:2), pp. 205-239.