Bank Lending and Credit Supply Shocks
Simon Gilchrist∗
Egon Zakrajšek†
October 7, 2011
Abstract
This paper analyzes the linkages between credit supply conditions and bank lending. Building on the recent work of Gilchrist and Zakrajšek [2011], we use the excess bond premium—a
component of corporate credit spreads designed to measure shifts in the risk attitudes of financial
intermediaries—to empirically identify credit supply shocks. Our results indicate that shocks to
the excess bond premium lead to a pronounced and protracted contraction in economic activity,
a decline in nominal interest rates, a sharp fall in equity valuations, and an eventual decline in
business loans outstanding. To shed light on the sluggish response of business lending to our
measure of credit supply shocks, we examine the joint responses of business loans outstanding
and corresponding unused commitments, an especially cyclically-sensitive component of bankintermediated business credit. According to our results, the lagged response of business loans to
credit supply disruptions appears to be a typical feature of business cycles, because in the initial
phase of the downturn, the capacity of businesses to borrow from the banking sector shrinks
primarily through reductions in unused commitments and only eventually through a reduction
in loans outstanding. Overall, our results indicate that once differences in responses of unused
commitments and loans outstanding to credit disruptions are accounted for, the capacity of
businesses to borrow from the commercial banking sector is highly sensitive to changes in credit
supply conditions.
JEL Classification: E44, E51, G21
Keywords: credit market disruptions, lending capacity, unused commitments, credit cycles
This paper was prepared for the International Economic Association World Congress held July 4–8, 2011, in
Beijing China. We thank Bill Bassett, Hesna Genay, and Gretchen Weinbach for helpful discussions. Michael
Levere and Ben Rump provided outstanding research assistance. The views expressed in this paper are solely the
responsibility of the authors and should not be interpreted as reflecting the views of the Board of Governors of the
Federal Reserve System or of anyone else associated with the Federal Reserve System.
∗
Department of Economics Boston University and NBER. E-mail: sgilchri@bu.edu
†
Division of Monetary Affairs, Federal Reserve Board. E-mail: egon.zakrajsek@frb.gov
1
Introduction
The turmoil that raged in the global financial markets during the 2007–09 crisis left a significant
imprint on bank lending over the past several years. The successive waves of turbulence that ripped
through the financial system during that period exerted substantial pressure on both the asset and
liability sides of banks’ balance sheet, and banks, at the height of the crisis in the latter part of
2008, faced funding markets that were largely illiquid and secondary markets that were essentially
closed to sales of certain types of loans and securities. Together with the slowdown in economic
activity that emerged at the end of 2007 and accelerated appreciably in latter part of 2008, these
financial disruptions led banks to become significantly more cautious in the extension of credit and
to take steps to bolster their capital and liquidity positions. Moreover, the persistent tightness
of credit conditions faced by many borrowers, in combination with generally weak demand for
bank-intermediated credit, have continued to affect lending during the sluggish recovery. Indeed,
two years after the official end of the recession, core loans outstanding—the sum of bank loans
to households and nonfinancial businesses—remain 13 percent below the level reached during the
cyclical peak in December 2007.1
The 2007–09 financial crisis began to unfold as rising delinquencies on subprime mortgages
in the first half of 2007, triggered by the end of the housing boom in the United States, started
to lead to large losses on related structured credit products. At that time, banks, in addition
to their mounting concerns about actual and potential credit losses, recognized that they might
need to take a large volume of assets onto their balance sheets, given their existing commitments
to customers and the heightened reluctance of investors to purchasing an increasing number of
securitized products. The recognition that the ongoing turmoil in financial markets could lead to
substantially larger-than-anticipated calls on their funding abilities and investors’ concerns about
valuation practices for opaque assets arguably led banks to start adopting a significantly moredefensive lending posture in an effort to conserve their balance sheet capacity.
The full-fledged global nature of the crisis burst to the forefront on September 15, 2008, when
Lehman Brothers—with its borrowing capacity severely curtailed by a lack of collateral—filed for
bankruptcy. Investor anxiety about financial institutions escalated sharply, and market participants
became extraordinarily skittish and pulled back from risk-taking even further. Amid cascading
effects of these financial disruptions, which included the government’s rescue of Fannie Mae, Freddie Mac, and AIG, a run on money market mutual funds, and the failure of Washington Mutual
(a large U.S. thrift), asset values dropped sharply, business and consumer confidence slumped, and
aggregate credit conditions tightened further. Bank lending to households and businesses, which
had held up reasonably well through the early stages of the crisis, slumped, and the U.S. economy
found itself in the so-called “adverse feedback loop,” a precarious situation, in which deteriorating
1
Based on July 29, 2011, Federal Reserve’s H.8 Statistical Release, “Assets and Liabilities of Commercial Banks
in the United States.”
1
economic and financial conditions become mutually reinforcing.
In this paper, we examine empirically the linkages between credit supply conditions, bank
lending, and economic activity. We begin by providing a historical context for the behavior of the
major categories of bank loans to households and businesses during the 2007–09 financial crisis.
We document that prior to the escalation of financial market turmoil in the autumn of 2008, the
provision of credit to businesses and households by commercial banks—with the exception of home
mortgage loans—appeared to have been well maintained relative to the stage of the business cycle.
However, bank lending exhibited a sharp slowdown at the end of 2008, and loans on banks’ balance
sheets have continued to run off throughout the nascent recovery. From a pure timing perspective,
therefore, the contraction in bank lending is a coincident, if not lagging, indicator of business cycle
dynamics. As such, one may be tempted to conclude that the contraction in lending in the latter
stages of the downturn was a symptom of rather than a causal mechanism during the crisis.
To explore this issue, we consider the effects of credit supply shocks on bank lending and
economic activity. The notion that the banking system may serve as a propagation mechanism
for, or a source of, macroeconomic shocks is, of course, not new. In the 1960s, for example,
Brunner and Meltzer [1963] criticized small-scale macroeconomic models for not including multiple
measures of credit prices and quantities (including those of bank loans), while in his seminal work
on the Great Depression, Bernanke [1983] argued that the widespread bank failures during the
early 1930s helped exacerbate the depth and length of the ensuing economic contraction.2 The
issue, however, remains contentious, because any empirical investigation of the role that banks—
or financial intermediaries more generally—play in economic fluctuations is complicated by severe
endogeneity problems, as a plethora of financial shocks that can affect the supply of bank loans is
also likely to have independent effects on economic activity.3
To tackle this difficult identification problem, we employ the approach used recently by
Gilchrist and Zakrajšek [2011], who use a large panel of secondary market prices of bonds issued
by U.S. nonfinancial corporations to decompose the associated credit spreads into two components: (1) a component capturing the usual countercyclical movements in expected defaults; and
(2) a component representing the cyclical changes in the relationship between default risk and
credit spreads—the so-called excess bond premium. As shown by Gilchrist and Zakrajšek [2011],
2
Empirical studies documenting the macroeconomic effects of adverse shocks to bank loan supply include, among
others, Bernanke and Lown [1991], Peek and Rosengren [1995, 1997, 2000], Ashcraft [2005], Lown and Morgan [2006],
and Bassett et al. [2010].
3
An example of such a shock is an unanticipated change in the stance of monetary policy. A large body of research
has focused on whether monetary policy might have effects on real activity through the market for bank loans: If
banks were not able to readily substitute other sources of funding for deposits, then changes in the federal funds
rate—which affect banks’ opportunity cost of issuing certain kinds of deposits—would influence the price and supply
of bank loans. In turn, this change in credit market conditions would affect investment and consumption decisions of
bank-dependent borrowers. However, monetary policy shocks also affect consumption and investment through their
influence on other interest rates. Thus, parsing the marginal effect of monetary policy shocks on economic activity
through the market for bank loans requires additional identifying assumptions, which, in practice, are difficult to
come by; see, for example, Bernanke and Blinder [1988] and Kashyap and Stein [1994, 2000].
2
movements in the excess bond premium appear to reflect shifts in the risk attitudes of financial
intermediaries, the marginal investors pricing corporate debt claims. As such, fluctuations in the
excess bond premium may provide an especially timely indicator of cyclical changes in credit supply
conditions, both within the corporate cash market and in the market for bank-intermediated credit.
Indeed, our results indicate that shocks to the excess bond premium—a measure of disruptions to
the credit intermediation process—cause a pronounced contraction in economic activity, a decline
in nominal interest rates, a sharp fall in equity valuations, and an eventual decline in bank lending.
Thus, the fact that a runoff in bank loans during the 2007–09 recession did not materialize until
the latter stages of the recession appears to be a general feature of credit market disruptions.
To analyze in more detail the apparently sluggish response of bank lending to credit supply
shocks, we examine the joint dynamics of business loans outstanding and unused business loan
commitments, an especially cyclically-sensitive component of bank-intermediated credit. Our focus
on unused commitments is motivated by the fact that the banking system provides credit to businesses (and households) in two important ways: By originating new loans (on balance sheet) and
by providing lines of credit (off balance sheet). This distinction is crucial for understanding the
cyclical dynamics of bank lending, because unused loan commitments, which represent a significant
source of off-balance-sheet credit risk, started to contract immediately with the onset of the crisis
in mid-2007, while business loans outstanding on banks’ balance sheet expanded briskly during the
first year of the recession.
A part of the contraction in unused commitment occurred as banks, in response to mounting
capital and liquidity pressures, started to reduce their off-balance-sheet credit exposures by cutting
their customers’ existing lines of credit, a move that was entirely coincident with the overall deterioration in financial market conditions as measured by a rise in the excess bond premium.4 Indeed,
unused business loan commitments plummeted at a 30 percent annual rate during the height of the
crisis in the fourth quarter of 2008, precisely when the excess bond premium reached its historical
peak. Business loans on banks’ balance sheet, by contrast, were still expanding at this point of the
cycle; see, for example, Chari et al. [2008] and Ivashina and Scharfstein [2010].
While both the extent and severity of financial market disruptions during the 2007–09 crisis
were unprecedented by post-war standards, our results indicate that such behavior represents a
typical response of bank lending to credit supply shocks—that is, in the initial phase of the cyclical
downturn, the capacity of businesses to borrow from the banking sector shrinks primarily through
reductions in unused commitments and only eventually through a reduction in loans outstanding.
All told, the empirical evidence presented in this paper indicates that loan quantities are highly
sensitive to changes in credit supply conditions once one takes into account the differences in cyclical
4
A portion of the decline in unused commitments, of course, reflected drawdowns on the existing lines by businesses,
which, at the same time, boosts the amount of loans outstanding on banks’ balance sheets. The existing data on
credit flows through the banking sector, however, are inadequate to parse out these two effects; for a more thorough
discussion of the difficulties surrounding the measurement of the underlying credit intermediated by banks during
cyclical downturns, see Bassett et al. [2011].
3
dynamics of unused loan commitments versus loans outstanding.
2
Credit Aggregates During the 2007–09 Financial Crisis
In this section, we place the behavior of credit aggregates during the recent crisis into a historical
context. We begin by documenting the cyclical properties of total private nonfinancial debt and
its main components. We then focus on credit intermediated by commercial banks and provide an
explicit comparison of lending patterns at banks with the aggregate credit flows. A common story
that emerges from these different cuts of the data is that, with the exception of home mortgage
lending, other forms of credit outstanding to households and businesses increased appreciably during
the early stages of the recession. However, the extension of credit came to an abrupt end in late
2008, following the sharp escalation of financial market turmoil sparked by the collapse of Lehman
Brothers.
Our data on debt stocks outstanding come from the Federal Reserve’s Z.1 Statistical Release,
“Flow of Funds Accounts of the United States” and cover the period from 1952:Q1 to 2010:Q4. As
noted above, we begin by analyzing the behavior of total private nonfinancial debt and its main
components: home mortgages, consumer credit (i.e., credit card, auto, student, and other consumer
loans), and debt extended to nonfinancial businesses (i.e., bank and non-bank loans, commercial
paper, and corporate bonds). We normalized each credit category by nominal GDP, a transformation that facilitates the comparison of credit cycles across recessions that differ appreciably in
their severity and duration. The resulting credit ratios, however, still contain pronounced secular
trends, reflecting the myriad of structural changes that took place in credit markets over the past
six decades. To abstract from these developments, we regressed the logarithm of each credit-toGDP ratio on a constant and linear and quadratic time trends; for each NBER-dated recession
since 1952, we then normalized the detrended series to equal zero at its respective business cycle
peak.
The solid black lines in Figure 1 depict the average behavior of these credit aggregates around
NBER-dated business cycle peaks, calculated using data for all recessions since 1953, excluding the
2007–09 downturn; the shaded band in each panel represents the corresponding range of outcomes,
while the red line shows the behavior of each series during the period surrounding the 2007–09
financial crisis. As shown in the upper left panel, total private nonfinancial debt outstanding
continued to expand—relative to the cycle—over the first five quarters following the NBER-dated
peak in 2007:Q4. The expansion in private debt outstanding owed importantly to a significant
increase in debt extended to nonfinancial business and, to a lesser extent, a rise in consumer credit.
In contrast, home mortgage debt started to decline immediately with the onset of the recession,
with the runoff intensifying noticeably precisely at the stage of the cycle when the expansion of
consumer credit and nonfinancial business debt came to an end. On the cyclically-adjusted basis, the
contraction in home mortgage debt outstanding during the 2007–09 recession is unprecedented by
4
Figure 1: Cyclical Dynamics of Private Nonfinancial Debt and its Main Components
Private nonfinancial debt
Home mortgages
Percent change from business cycle peak
Percent change from business cycle peak
5
20
15
10
0
5
0
-5
-5
-10
2007-09 recession
Average
Range
-10
-15
-20
-15
-4
-2
0
2
4
6
8
10
-25
12
-4
Quarter to and from business cycle peak
-2
0
2
4
6
8
10
12
Quarter to and from business cycle peak
Consumer credit
Nonfinancial business debt
Percent change from business cycle peak
Percent change from business cycle peak
15
20
10
10
5
0
0
-5
-10
-10
-15
-20
-20
-25
-4
-2
0
2
4
6
8
10
-30
12
-4
Quarter to and from business cycle peak
-2
0
2
4
6
8
10
12
Quarter to and from business cycle peak
Note: The panels of the figure depict the behavior of the major categories of credit outstanding to households
and nonfinancial businesses around NBER-dated business cycle peaks. Each category of credit is normalized
by nominal GDP; the logarithm of each credit ratio was detrended using linear and quadratic time trends. For
each credit category, the average cyclical component (the black lines) and the range of cyclical components
(the shaded bands) are based on data for recessions designated by the NBER since 1953, excluding the
2007–09 downturn.
5
Figure 2: The Relative Importance Bank Lending to Households and Businesses
100
Quarterly
Home mortgages
Commercial mortgages
Consumer credit
Nonfinancial business credit
80
60
40
20
0
1952
1956
1960
1964
1968
1972
1976
1980
1984
1988
1992
1996
2000
2004
2008
Note: The figure depicts the major components of loans outstanding at commercial banks, expressed as a
share of total debt outstanding in that category. Shaded vertical bars represent NBER-dated recessions.
post-war standards, while the behavior of consumer credit and nonfinancial business debt is largely
within the historical norms.
To help understand the role of the commercial banking sector in the credit allocation process,
Figure 2 plots the four major components of bank loans outstanding, expressed as a percentage of
total debt outstanding in that category. According to this metric, the relative importance of banks
in the direct provision of credit to households and businesses has clearly changed significantly over
time.5 First, the role of the banking sector in the provision of credit to nonfinancial businesses has
diminished steadily over the past six decades, a pattern reflecting primarily the deepening of, and
firms’ greater access to, capital markets, as well as the rise of non-bank financial intermediaries.6
Second, banks appear to have offset this loss of business, in part, by significantly stepping
up commercial real estate lending, as the share of commercial mortgages outstanding on banks’
balance sheets has risen from about 20 percent to more than 50 percent over the same period. The
importance of banks in the provision of consumer credit has also declined noticeably since the late
1970s, due mainly to the increased importance of the consumer finance industry. In contrast, the
share of home mortgages on the books of commercial banks has remained fairly stable at around
5
For a more thorough and complete discussion of these issues see Boyd and Gertler [1994].
Non-bank lenders active in the commercial space include business factors (i.e., non-depository institutions that
specialize in receivable-based financing and inventory finance), credit card companies, finance and leasing groups,
and private and public syndicates. These lenders offer a range of business loans and some of them even specialize in
lending to specific industries.
6
6
18 percent over the past six decades.
Figure 3 depicts the cyclical dynamics of these four aggregate bank loan categories.7 Focusing
on the most recent patterns reveals that bank lending to households and businesses generally
conformed to the aggregate credit trends shown in Figure 1. One exception appears to be the
behavior of bank-intermediated consumer credit, which during the recent crisis diverged noticeably
from historical norms. Compared with total consumer credit, consumer loans on banks’ books
increased—on the cyclically-adjusted basis—significantly more over the first five quarters of the
downturn. Nevertheless, the subsequent contraction bank-intermediate consumer credit has been
quite severe and has persisted well into the initial recovery stage of the cycle, a pattern that is also
evident in the behavior of total consumer credit.
On the business side, the exclusion of arms-length finance associated with the commercial paper
and corporate bond markets serves only to highlight the role that the banking sector serves as the
provider of business credit in the form of credit lines. Compared with total nonfinancial business
credit, bank lending to nonfinancial firms through C&I loans expanded faster and by a greater
percentage in the early stages of the 2007–09 recession.8 Bank lending to businesses through
commercial real estate loans also increased robustly on the cyclically-adjusted basis during the first
year of the recession before contracting significantly in early 2009. In general, the runoff in business
loans outstanding during the most recent downturn was considerably more severe and occurred at
a significantly later stage of the cycle, compared with an average post-war recession.
7
Although our analysis is focused on the commercial banking sector, we note that the general cyclical patterns of
these four loan categories at banks are very similar to those at all depository institutions.
8
As shown by Gertler and Gilchrist [1993] the expansion in C&I loans outstanding during the initial stages of an
economic downturn is driven primarily by differences in borrowing between small and large firms. Because of a strong
countercyclical demand for short-term credit to finance inventory accumulation in the first few quarters following a
business cycle turning point, firms, in general, would like to increase their borrowing to smooth the effects of declining
cash flows. However, only firms with relatively unimpeded access to credit markets—typically large firms—are able
to obtain the desired funds at prevailing market rates. In such circumstances, large “high-quality” borrowers tap the
commercial paper market (Calomiris et al. [1995]) or draw down their lines of credit (Morgan [1998]).
7
Figure 3: Cyclical Dynamics of Household and Business Loans at Commercial Banks
Home mortgages
Commercial mortgages
Percent change from business cycle peak
Percent change from business cycle peak
20
20
15
2007-09 recession
Average
Range
10
10
5
0
0
-5
-10
-10
-15
-20
-20
-25
-4
-2
0
2
4
6
8
10
-30
12
-4
Quarter to and from business cycle peak
-2
0
2
4
6
8
10
12
Quarter to and from business cycle peak
Consumer credit
Nonfinancial business credit
Percent change from business cycle peak
Percent change from business cycle peak
15
20
10
10
5
0
0
-5
-10
-10
-15
-20
-20
-25
-4
-2
0
2
4
6
8
10
-30
12
-4
Quarter to and from business cycle peak
-2
0
2
4
6
8
10
12
Quarter to and from business cycle peak
Note: The panels of the figure depict the behavior of the major categories of loans outstanding—held on the
books of commercial banks—to households and nonfinancial businesses around NBER-dated business cycle
peaks. Each loan category is normalized by nominal GDP; the logarithm of each loan ratio was detrended
using linear and quadratic time trends. For each loan category, the average cyclical component (the black
lines) and the range of cyclical components (the shaded bands) are based on data for recessions designated
by the NBER since 1953, excluding the 2007–09 downturn.
8
3
Credit Supply Shocks and Bank Lending
In this section, we analyze the effects of credit supply shocks on bank lending and economic activity.
As discussed above, we rely on the excess bond premium—an indicator of financial market stress
developed by Gilchrist and Zakrajšek [2011] (GZ hereafter)—to measure changes in credit supply
conditions. Our choice is motivated by an emergent literature that stresses the importance of
balance sheet conditions of financial intermediaries—including those of commercial banks—for the
joint determination of asset prices and macroeconomic aggregates.
3.1
The Excess Bond Premium
Before presenting our main results, we briefly outline the empirical methodology underlying the
construction of the excess bond premium. Specifically, GZ employ a large panel of corporate bonds
issued by nonfinancial firms to decompose the associated credit spreads into two components: a
default-risk component capturing the usual countercyclical movements in expected defaults, and
a non-default-risk component that captures the cyclical fluctuations in the relationship between
default risk and credit spreads. The GZ decomposition of credit spreads is based on the regression
model of the following type:
ln Sit [k] = −βDDit + θ′ Xit [k] + ǫit [k],
(1)
where Si [k] denotes the credit spread on bond k (issued by firm i); DDit is the distance-to-default
for firm i; Xit [k] is a vector of bond-specific characteristics that controls for the optionality features
embedded in most corporate securities as well as for potential term and liquidity premiums; and
ǫit [k] is a “pricing error.”
The key feature of the GZ approach is that the firm-specific credit risk is captured by the
distance-to-default (DD), a market-based indicator of default risk based on the option-theoretic
framework developed in the seminal work of Merton [1974]. In this contingent claims approach
to corporate credit risk, it is assumed that a firm has just issued a single zero-coupon bond of
face value F that matures at date T . Rational stockholders will default at date T only if the
total value of the firm VT < F ; by assumption, the rights of the bondholders are activated only at
the maturity date, as stockholders will maintain control of the firm even if the value of the firm
Vs < F for some s < T . Under the assumptions of the model, the probability of default—that is,
Pr[VT < F ]—depends on the distance-to-default, a volatility-adjusted measure of leverage inferred
from equity valuations and the firm’s observed capital structure.
Using the estimated parameters of the credit spread model (1), GZ define the excess bond
premium in month t by the following linear decomposition:
b̄ ,
EBPt = S̄t − S
t
9
Figure 4: The Excess Bond Premium
Percentage points
3
Monthly
2
1
0
-1
-2
1973
1976
1979
1982
1985
1988
1991
1994
1997
2000
2003
2006
2009
Note: The figure depicts the estimated excess bond premium (see Gilchrist and Zakrajšek [2011] for details).
The shaded vertical bars represent the NBER-dated recessions.
b̄ is its predicted counterpart. Figwhere S̄t denotes the average credit spread in month t and S
t
ure 4 shows the estimated monthly excess bond premium from January 1973 to December 2010.
According to the results reported by GZ, the majority of the well-documented information content of credit spreads for future economic activity is attributable to movements in the excess bond
premium—that is, to deviations in the pricing of corporate debt claims relative to the expected
default risk of the issuer. Moreover, GZ show that shocks to the excess bond premium premium
that are orthogonal to the current macroeconomic conditions cause economically and statistically
significant declines in economic activity and inflation, as well as in risk-free rates and broad equity
prices.
Importantly, GZ also show that fluctuations in their excess bond premium are closely related
to the financial condition of broker-dealers, highly leveraged financial intermediaries that play a
key role in most financial markets.9 Taken together, the evidence presented by GZ is consistent
9
According to Adrian and Shin [2010], broker-dealers are financial institutions that buy and sell securities for a
fee, hold an inventory of securities for resale, and differ from other types of institutional investors by their active procyclical management of leverage. As documented by Adrian and Shin [2010], expansions in broker-dealer assets are
10
with the notion that deviations in the pricing of long-term corporate bonds relative to the expected
default risk of the underlying issuer reflect shifts in the effective risk aversion of the financial sector.
Increases in risk aversion, in turn, lead to a contraction in the supply of credit, both through the
corporate bond market and the broader commercial banking sector.
3.2
The Excess Bond Premium and Bank Lending
Following GZ, we employ an identified vector autoregression (VAR) to analyze the near- and longerterm effects of financial disturbances—as measured by shocks to the excess bond premium—on bank
lending. Our first set of results is based on monthly data, a data frequency that allows us to be
more precise about the causal mechanism at work. In particular, we show that in a monthly VAR,
orthogonalized innovations to the excess bond premium have a negligible contemporaneous effect
on financial asset prices. This finding implies that our identification strategy is unlikely to be
associated with reverse causation, in which broad declines in asset valuation cause movements in
the excess bond premium, and we interpret these baseline results as providing strong validation to
our overall approach of identifying credit supply shocks.
To examine the macroeconomic consequences of credit market disturbances, we augment an
otherwise standard VAR with the excess bond premium and bank lending variables. Specifically,
our VAR includes the following endogenous variables: (1) the change in the unemployment rate;
(2) the log-difference in the index of manufacturing industrial production; (3) inflation as measured
by the log-difference of the price deflator for personal consumption expenditures (PCE); (4) the
excess bond premium; (5) the log-difference of real loans outstanding to households; (6) the logdifference of real loans outstanding to businesses; (7) the log-difference of the price-dividend ratio;
(8) the 10-year (nominal) Treasury yield; and (9) the effective (nominal) federal funds rate.10
The first three variables provide monthly measures of economic activity and inflation. They are
traditionally viewed as “slow moving” variables and are ordered before the excess bond premium
when performing the Cholesky decomposition used to compute orthogonalized innovations to the
excess bond premium. The block after the excess bond premium includes bank lending and asset
prices, all of which viewed are as “fast moving” variables and thus are allowed to react contemporaneously to innovations in the excess bond premium. According to our identification scheme,
therefore, shocks to the excess bond premium may engender an immediate response in stock market
valuations—as measured by the price-dividend ratio—as well in nominal interest rates and bank
lending. Importantly, this ordering also allows for the possibility that innovations to the excess
bond premium may cause an immediate shift in the stance of monetary policy—as measured by the
associated with increases in leverage as broker-dealers take advantage of greater balance sheet capacity; conversely,
contractions in their assets are associated with de-leveraging of their balance sheets.
10
Business loans include C&I loans and business loans secured by commercial real estate; household loans include
residential mortgages, credit card loans, and other consumer loans. All series were obtained from the Federal Reserve’s
H.8 Statistical Release. The household and business loans outstanding were deflated by the PCE price deflator.
11
Figure 5: Credit Supply Shocks, Economic Activity, and Bank Lending
Unemployment rate
Industrial production
Percentage points
PCE prices
Percentage points
Percentage points
0.4
0.5
0.3
0.0
0.2
-0.5
-0.05
0.1
-1.0
-0.10
0.0
-1.5
-0.1
-2.0
0.10
0.05
0.00
-0.15
0
6
12
18
24
30
36
0
Months after the shock
6
12
18
24
30
36
-0.20
-0.25
0
Months after the shock
Excess bond premium
12
18
24
30
36
Months after the shock
Real business loans
Percentage points
6
Real household loans
Percentage points
0.30
Percentage points
2
2
1
1
0
0
-1
-1
-2
-2
0.25
0.20
0.15
0.10
0.05
0.00
-0.05
-0.10
0
6
12
18
24
30
36
-3
0
Months after the shock
6
12
18
24
30
36
-3
0
Months after the shock
Price-dividend ratio
12
18
24
30
36
Months after the shock
10-year Treasury yield
Percentage points
6
Federal funds rate
Percentage points
Percentage points
0.05
0.1
0.00
0.0
-2
-0.05
-0.1
-3
-0.10
-0.2
-0.15
-0.3
1
0
-1
-4
-5
-6
0
6
12
18
24
30
Months after the shock
36
-0.20
0
6
12
18
24
30
Months after the shock
36
-0.4
0
6
12
18
24
30
36
Months after the shock
Note: The panels of the figure depict the impulse responses to a 1 standard deviation orthogonalized shock
to the excess bond premium (see text for details). The responses of the change in the unemployment rate, the
log-difference of industrial production, PCE price inflation, the log-difference of he price-dividend ratio, the
log-difference of real business loans, and the log-difference of real household loans have been accumulated.
Shaded bands denote 95-percent confidence intervals based on 1,000 bootstrap replications.
12
movements in the federal funds rate—to perceived disruptions in the credit intermediation process.
Figure 5 depicts the impulse responses of these variables to an orthogonalized shock to the
excess bond premium, based on the VAR(6) model estimated over the period from January 1986
to December 2010.11 An unanticipated increase of one standard deviation in the excess bond
premium—about 30 basis points—causes a significant and protracted slowdown in economic activity. In economic terms, the implications of this relatively modest financial disruption are substantial:
Unemployment rate increases a little less than one-quarter of a percentage point, while industrial
production falls a full percentage point relative to trend growth, with the peak decline in economic
activity occurring about two years after the initial shock.
The economic downturn is accompanied by a substantial fall in broad equity valuations, as
evidenced by the decline in the price-dividend ratio. In this sense, the results in Figure 5 are
consistent with the notion that turmoil in the corporate bond market—owing to an increase in
risk-aversion of financial intermediaries—leads to stresses in asset markets more generally, though
with a bit of a delay. Finally, the combination of growing economic slack and appreciable disinflation
in the wake of the financial shock elicits a significant easing of monetary policy and a decline in
longer-term yields. However, it is important to note that these forward-looking asset prices do
not react contemporaneously to our financial disturbance—indeed, the price-dividend ratio does
not begin to decline about until four months after the impact of the shock to the excess bond
premium.12
On the credit side, the shock to the excess bond premium has no immediate effect on bank
lending to either households or businesses. In the first six months after the shock, both household and business loans outstanding increase slightly. About one year after the initial financial
disruption, however, business loans outstanding begin to contract and bottom out more than one
percentage point below their trend growth. The effect on household lending over this period, by
contrast, is statistically and economically indistinguishable from zero. All told, a shock to the
excess bond premium leads to a substantial decline in business loans outstanding, though the contraction in business lending occurs with a significant lag. In the next section, we provide further
insight into this delayed reaction by examining the behavior of loans outstanding versus unused
loan commitments.
3.3
Loans Outstanding vs. Loan Commitments
A key issue in the analysis of credit cycles is the degree to which firms and households are financing
their activities by borrowing through newly issued loans versus drawing on their existing lines of
11
The starting date of the estimation period is dictated by the availability of the monthly bank lending data.
As a robustness check, we also included macroeconomic uncertainty—as proxied by either the option implied
(i.e., the VXO index) or realized stock market volatility—into the “fast moving” block of our VAR specification. The
inclusion of this variable had no effect on the impulse responses reported in Figure 5.
12
13
credit.13 Similarly, from the credit-supply perspective, banks may curtail their credit exposures in
the initial phases of a cyclical downturn primarily by reducing the amount of unused commitments.
To the extent that banks’ initial reaction to financial disturbances is to cut back on unused commitments, this may help explain the sluggish response of loans outstanding to credit supply shocks.
In fact, this hypothesis is consistent with the evidence presented by Morgan [1998], who shows that
changes in loans outstanding not made under commitment are more sensitive to changes in the
stance of monetary policy than changes in loans made under a previous commitment.
The banks’ unique role as a provider of credit in the form credit lines is illustrated in Figure 6.
According to the Call Report data, core loans outstanding exceeded the corresponding unused
commitments by a substantial margin during the early 1990s.14 Over time, however, banks’ offbalance-sheet credit exposures have grown more rapidly, and by the most recent business cycle peak
at the end of 2007, core unused commitments totaled close to seven trillion dollars, substantially
more than about five trillion dollars of core loans outstanding. As shown in the bottom panel,
credit card commitments accounted for the majority of this off-balance-sheet exposure, followed
closely by business credit lines.15
In the latter half of 2007, this sizable off-balance-sheet credit exposure presented banks with
a major risk in light escalating financial market strains and an emerging slowdown in economic
activity. The bulk of the contraction in core unused commitments during that period was accounted for by a reduction in business credit lines, the most cyclically-sensitive component of
bank-intermediated credit. Given the relative importance of banks’ commitments to fund business
loans, we define a broader measure of credit intermediation by commercial banks in this segment of
the market: Business lending capacity, which attempts to capture the full potential of businesses to
borrow from the banking sector over time, as measured by the sum of business loans outstanding
and corresponding commitments to fund such loans. The black line in Figure 7 depicts the (annualized) quarterly growth rate of business lending capacity, while the shaded portions of the vertical
bars represent the quarterly growth contributions of business loans outstanding and corresponding
unused commitments.
According to Figure 7, cyclical fluctuations in business lending capacity are driven importantly
by changes in unused commitments, a pattern that was especially pronounced during the most
recent crisis. Indeed, the data shown in the figure reveal a more general pattern: A reduction in
business lending capacity at an early stage of a downturn is due entirely to a decline in unused
commitments rather than business loans outstanding. During the recent crisis, in particular, the
13
Another option, of course, is to rely on internal liquidity. As shown by Acharya et al. [2009], firms with greater
exposure to aggregate risk find it more costly to obtain credit lines from banks and, as a result, tend to rely more
heavily on cash reserves to manage their future liquidity needs.
14
Date on unused commitment were added to Call Reports in 1990:Q2.
15
It is important to note that what we label as “business lines” is recorded in Call Reports prior to 2010 as “other”
unused commitments. More detailed data available since 2010 suggest that credit lines to businesses—both financial
and nonfinancial—account for the vast majority of this category, which indicates that these data provide a useful
proxy for unused credit lines to businesses.
14
Figure 6: Core Loans and Unused Commitments at Commercial Banks
$ Trillions
7
Quarterly
Core unused commitments
Core loans outstanding
6
5
4
3
2
1
0
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
(a) Core Loans Outstanding and Unused Commitments
$ Trillions
7
Quarterly
HELOCs
Credit cards
Business
6
5
4
3
2
1
0
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
(b) Composition of Unused Commitments
Note: The black line in the top panel depicts the dollar amount of core unused commitments, and the
dotted red line depicts the dollar amount of core loans outstanding at U.S. commercial banks. Core loan
categories include C&I, real estate, and consumer loans. The bottom panel depicts the composition of
unused commitments. All series are deflated by the GDP price deflator (2005 = 100). Shaded vertical bars
represent NBER-dated recessions.
growth of unused commitments stepped down immediately with the emergence of financial market
turmoil in the summer of 2007, and unused commitment continued to contract well after the official
end of the recession in the spring of 2009. Business loans outstanding, in contrast, began to decline
four quarters after the onset of the recession. Lastly, note that unused commitments typically begin
to expand several quarters before a resumption in the growth of business loans outstanding.
An obvious question to consider next is the extent to which the contraction in unused commitments in the early stages of the business cycle is due to demand or supply factors. On the supply
side, banks may reduce unused commitments by canceling (or reducing) their customers’ existing
lines of credit and by restricting the supply of new commitments. On the demand side, firms may
15
Figure 7: Growth in Business Lending Capacity at Commercial Banks
Percent
25
Quarterly
20
15
10
5
0
-5
-10
Business loans
Business unused commitments
Business lending capacity
-15
-20
-25
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010
Note: The black line depicts the seasonally-adjusted (annualized) quarterly growth business lending capacity at U.S. commercial banks; lending capacity is defined as the sum of business loans outstanding and
corresponding unused commitments. All series are deflated by the GDP price deflator (2005 = 100). Shaded
vertical bars represent NBER-dated recessions.
draw on their existing commitments to satisfy countercyclical liquidity needs associated with the
financing of inventories and trade payables. As we have argued above, we view movements in the
excess bond premium as a timely and reliable indicator of changes in risk attitudes in the financial
intermediary sector, with an increase in risk aversion leading to a tightening of lending standards
and an eventual reduction in credit availability. The link between the excess bond premium and
banks’ willingness to lend is made explicit in Figure 8, which plots the excess bond premium against
the change in lending standards on C&I loans obtained from the Federal Reserve Board’s Senior
Loan Officer Opinion Survey on Bank Lending Practices (SLOOS).16 According to the figure, these
two indicators contain essentially the same information regarding the availability of business credit,
16
The SLOOS is usually conducted four times per year by the Federal Reserve Board, and up to 60 banks participate
in each survey. Banks are asked to report whether they have changed their credit standards over the past three months
on the major categories of loans to businesses and households. The series plotted is the net percentage of banks that
reported tightening their credit standards on C&I loans to large and middle-market firms. Reported net percent
equals the percent of banks that reported tightening their standards minus the percent that reported easing their
standards. The SLOOS data are plotted in such a way that the change in C&I credit standards over the survey
period is aligned with the quarter in which the reported change took place. For the full text of the questions and
more information on the survey, see http://www.federalreserve.gov/boarddocs/SnLoanSurvey/.
16
Figure 8: The Excess Bond Premium and Banks’ Willingness to Lend
Percentage points
4
Net percent
100
Monthly
Excess bond premium (left scale)
Change in C&I lending standards (right scale)
•
3
2
75
•
•
•
••
•
•
1
••
0
•• •
•• •
-1
•
••
•
• ••• •••• •••••• •
•• •
•
••
••
• ••
•••••
•
•
•
••••
•
•
••
• •• •
• ••
••••
•••
50
•
25
•
•
•••••
0
-25
-2
-50
-3
-75
1990
1992
1994
1996
1998
2000
2002
2004
2006
2008
2010
Note: The black line depicts the excess bond premium. The overlayed dots depict the net percent of SLOOS
respondents that reported tightening their credit standards on C&I loans over the past three months. The
timing of the plotted SLOOS data is such that the change in C&I credit standards over the survey period
is aligned with the quarter in which the reported change took place. The shaded vertical bars denote the
NBER-dated recessions.
though there is some evidence that changes in lending standards precede movements in the excess
bond premium by a few months.
In Figure 9, we focus on the recent events by plotting the excess bond premium against the
growth of unused commitments and business loans outstanding over the 2005–10 period. This figure
highlights the fact that the contraction in unused commitments is entirely coincident with the rise
in the excess bond premium that began in the second half of 2007. Moreover, unused commitments
plummeted as the excess bond premium shot up during the height of the financial market turmoil
in late 2008. To the extent that fluctuations in the excess bond premium reflect changes in credit
supply conditions, this coincident behavior strongly suggests that adverse credit supply dynamics
played an important role during the most recent economic downturn.
To investigate this issue more systematically, we estimate a VAR in which shocks to the excess
bond premium have a separate effect on the two components of business lending capacity. Because data on unused commitments are available only from 1990 onwards, we consider a relatively
parsimonious specification that includes the log-difference of real GDP, the excess bond premium,
17
Figure 9: The Excess Bond Premium and Business Lending, 2005–2010
Percentage points
3
2
1
Percent
30
Monthly
•
• •
• •• •
•
•
• • • • ••
• • ••
•
•
•
•
0
•
-1
20
10
• •
•
•
•
•
•
-2
•
•
Excess bond premium (left scale)
Change in business unused commitments (right scale)
Change in business loans (right scale)
-3
• • •
•
•
• •
•
•• •
-30
•
2005
2006
2007
2008
-10
-20
• •
-4
0
2009
-40
2010
Note: The black line depicts the excess bond premium. The overlayed red dots depict the (annualized)
quarterly growth of unused business commitments, while the green dots depict the (annualized) quarterly
growth of business loans outstanding. The shaded vertical bar denotes the 2007–09 NBER-dated recession.
the log-difference of real business loans outstanding, the log-difference of real unused business loan
commitments, the real 10-year Treasury yield, and the real federal funds rate.17 As credit supply
shocks, we again consider orthogonalized innovations to the excess bond premium, using a recursive
ordering scheme in which the excess bond premium is ordered after output growth but before all
other endogenous variables of the system.
The impulse responses of the endogenous variables to the excess bond premium shock based on
the VAR(2) model are shown in Figure 10. The effects of a financial shock on the real economy and
interest rates are substantial and similar to those shown in Figure 5 and are entirely in line with the
quarterly VAR results reported by Gilchrist and Zakrajšek [2011]. Also consistent with the results
reported in Figure 5 is the fact that business loans outstanding exhibit no immediate response to a
shock to the excess bond premium, though business loans eventually fall about 3 percentage points
relative to trend growth. This is an appreciably larger decline than the one documented earlier,
and it primarily reflects the limited time series range of the data, a time frame in which financial
17
By specifying the VAR in “real” terms, we abstract from inflationary dynamics, which are not the focus of our
analysis. Both the real federal funds rate and the real 10-year Treasury yield are defined as the corresponding nominal
rate less average CPI inflation over the next ten years, as reported by the Survey of Professional Forecasters.
18
Figure 10: Credit Supply Shocks and Business Lending Capacity
Real GDP
Excess bond premium
Percentage points
Percentage points
0.2
0.4
0.0
0.3
-0.2
0.2
-0.4
-0.6
0.1
-0.8
0.0
-1.0
-0.1
-1.2
-1.4
0
2
4
6
8
10
-0.2
12
0
2
Quarters after the shock
4
6
8
10
12
Quarters after the shock
Real business loans
Real business unused commitments
Percentage points
Percentage points
2
2
0
0
-2
-2
-4
-4
-6
-6
-8
-8
-10
0
2
4
6
8
10
-10
12
0
2
Quarters after the shock
4
6
8
10
12
Quarters after the shock
Real 10-year Treasury yield
Real federal funds rate
Percentage points
Percentage points
0.2
0.05
0.00
0.0
-0.05
-0.10
-0.2
-0.15
-0.4
-0.20
-0.25
-0.6
-0.30
-0.35
0
2
4
6
8
10
-0.8
12
0
Quarters after the shock
2
4
6
8
10
12
Quarters after the shock
Note: The panels of the figure depict the impulse responses to a 1 standard deviation orthogonalized shock
to the excess bond premium (see text for details). The responses of the log-difference of real GDP, the logdifference of real business loans, and the log-difference of real unused commitments have been accumulated.
Shaded bands denote 95-percent confidence intervals based on 1,000 bootstrap replications.
19
disturbances associated with the past three recessions played an especially prominent role. In
contrast to the response of business loans outstanding, unused commitments fall almost a full percentage point upon the impact of the financial shock and eventually decline more than 5 percentage
points relative to trend growth.
In summary, our results indicate that both business loans outstanding and corresponding unused
loan commitments are highly responsive to changes in credit supply conditions, both during the
most recent financial crisis and, as evidenced by the VAR results, more generally in response to
credit supply shocks, as measured by the innovations to the excess bond premium. According to
our results, the initial effect of a credit supply shock manifests itself in a sharp reduction in unused
loan commitments and only with an appreciable lag leads to a reduction in loans outstanding. The
immediate decline in unused commitments in response to an adverse credit supply shock implies a
substantial reduction in the capacity of businesses to borrow from the commercial banking sector,
a development that contributes significantly to an ensuing economic downturn.
4
Conclusion
The 2007–09 financial crisis highlighted the importance of the health of financial intermediaries
for macroeconomic outcomes. This paper examined one facet of this relationship, namely the
link between financial market stress and bank lending. Our results indicate that financial market
strains—as measured by a rise in the excess bond premium—preceded the decline in bank lending
by a significant margin and that fluctuations in the excess bond premium provided a reliable
and timely gauge of changes in credit supply conditions during the crisis. In spite of the early
deterioration in credit supply conditions, core loans outstanding increased noticeably during the
initial phase of the crisis.
To help elucidate the link between financial distress and bank lending—both during the financial
crisis and as a general response to adverse credit supply shocks—we analyzed the joint dynamics of
business loans outstanding and the corresponding unused commitments. According to our results,
banks’ initial reaction to disruptions in financial markets is to reduce their off-balance-sheet credit
exposures—by cutting their customers’ unused loan commitments—while the decline in business
loans outstanding occurs with a substantial delay. These differential dynamics between on- and
off-balance-sheet credit exposures have important implications for policymakers who must rely on
the available information on credit prices and quantities to gauge the severity of credit market
disruptions and their potential effects on the macroeconomy. Our results also suggest that macroprudential policy should take into account banks’ off-balance-sheet exposures created by unused
commitments when assessing the overall risk of the financial sector.
20
References
Acharya, V. V., H. Almeida, and M. Campello (2009): “Aggregate Risk and the Choice
Between Cash and Credit Lines,” Working Paper, Dept. of Finance, University of Illinois at
Urbana-Champaign.
Adrian, T. and H. S. Shin (2010): “Liquidity and Leverage,” Journal of Financial Economics,
19, 418–437.
Ashcraft, A. B. (2005): “Are Banks Really Special? New Evidence from the FDIC-Induced
Failure of Healthy Banks,” American Economic Review, 95, 1712–1730.
Bassett, W. F., M. B. Chosak, J. C. Driscoll, and E. Zakrajšek (2010):
“Changes in Bank Lending Standards and the Macroeconomy,” Available at SSRN:
http//ssrn.co/abstract=1758832.
Bassett, W. F., S. Gilchrist, G. C. Weinbach, and E. Zakrajšek (2011): “Improving Our
Ability to Monitor Bank Lending During a Financial Crisis,” Working Paper, Dept. of Economics,
Boston University.
Bernanke, B. S. (1983): “Nonmonetary Effects of the Financial Crisis in Propagation of the
Great Depression,” American Economic Review, 73, 257–276.
Bernanke, B. S. and A. S. Blinder (1988): “Credit, Money, and Aggregate Demand,” American Economic Review, 78, 435–439.
Bernanke, B. S. and C. S. Lown (1991): “The Credit Crunch,” Brookings Papers on Economic
Activity, 2, 205–239.
Boyd, J. H. and M. Gertler (1994): “Are Banks Dead? Or are the Reports Greatly Exagerrated?” Federal Reserve Bank of Minneapolis Quarterly Review, 18, 2–23.
Brunner, K. and A. H. Meltzer (1963): “The Place of Financial Intermediaries in the Transmission of Monetary Policy,” American Economic Review, 53, 372–382.
Calomiris, C. W., C. P. Himmelberg, and P. Wachtel (1995): “Commercial Paper, Corporate Finance, and the Business Cycle: A Microeconomic Perspective,” Carnegie-Rochester
Conference Series on Public Policy, 42, 203–250.
Chari, V. V., L. Christiano, and P. J. Kehoe (2008): “Facts and Myths About the Financial
Crisis of 2008,” Federal Reserve Bank of Minneapolis Working Paper No. 666.
Gertler, M. and S. Gilchrist (1993): “The Role of Credit Market Imperfections in the Monetary Transmission Mechanism: Arguments and Evidence,” Scandinavian Journal of Economics,
95, 43–64.
Gilchrist, S. and E. Zakrajšek (2011): “Credit Spreads and Business Cycle Fluctuations,”
NBER Working Paper No. 17021, Forthcoming American Economic Review.
Ivashina, V. and D. Scharfstein (2010): “Bank Lending During the Financial Crisis of 2008,”
Journal of Financial Economics, 97, 319–338.
21
Kashyap, A. K. and J. C. Stein (1994): “Monetary Policy and Bank Lending,” in Monetary
Policy, ed. by N. G. Mankiw, Chicago: The University of Chicago Press, 221–262.
——— (2000): “What Do a Million Observations on Banks Say About the Transmission of Monetary Policy?” American Economic Review, 90, 407–428.
Lown, C. S. and D. P. Morgan (2006): “The Credit Cycle and the Business Cycle: New
Findings from the Loan Officer Opinion Survey,” Journal of Money, Credit, and Banking, 38,
1575–1597.
Merton, R. C. (1974): “On the Pricing of Corporate Debt: The Risk Structure of Interest Rates,”
Journal of Finance, 29, 449–470.
Morgan, D. P. (1998): “The Credit Effects of Monetary Policy: Evidence From Using Loan
Commitments,” Journal of Money, Credit, and Banking, 30, 102–118.
Peek, J. and E. S. Rosengren (1995): “The Capital Crunch: Neither a Borrower nor a Lender
Be,” Journal of Money, Credit, and Banking, 27, 625–638.
——— (1997): “The International Transmission of Financial Shocks,” American Economic Review,
87, 625–638.
——— (2000): “Collateral Damage: Effects of the Japanese Bank Crisis on Real Activity in the
United States,” American Economic Review, 90, 30–45.
22