linkages between the financial and real sectors: an...

28
Linkages Between the Financial and Real Sectors: An Overview Simon Gilchrist * Egon Zakrajˇ sek September 24, 2008 Prepared for the Academic Consultants Meeting, “Financial Stability and Linkages Between Financial and Real Sectors,” Federal Reserve Board, October 3, 2008. We thank Dan Sichel for helpful conversations. * Department of Economics Boston University and NBER. E-mail: [email protected] Division of Monetary Affairs, Federal Reserve Board. E-mail: [email protected] 1

Upload: others

Post on 02-Apr-2020

5 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Linkages Between the Financial and Real Sectors:

An Overview

Simon Gilchrist∗ Egon Zakrajsek†

September 24, 2008

Prepared for the Academic Consultants Meeting, “Financial Stability and Linkages Between Financialand Real Sectors,” Federal Reserve Board, October 3, 2008. We thank Dan Sichel for helpful conversations.

∗Department of Economics Boston University and NBER. E-mail: [email protected]†Division of Monetary Affairs, Federal Reserve Board. E-mail: [email protected]

1

Page 2: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

1 Introduction

The United States is currently in the throes of an acute liquidity and credit crunch, by all

accounts, the severest financial crisis since the Great Depression. The roots of this crisis

lie in the collapse of the subprime mortgage market in the wake of an unprecedented and

unexpected fall in house prices. The financial turmoil subsequently spread to a variety of

other asset markets, causing massive liquidity problems in interbank funding markets, the

sudden collapse of several major financial institutions, and a sharp reduction in lending

activity; see Brunnermeier [2008] for a detailed account of the 2007–08 financial crisis.

In the hope of preventing the financial meltdown from engulfing the real economy, the

government in recent days announced plans to purchase a broad range of mortgage-related

assets from a variety financial institutions, an intervention in the global financial markets

at an unprecedented scale.

In this essay, we assess the likely implications of such financial disruptions for the real

economy. We begin the analysis with a brief snapshot of the current state of affairs and

compare current trends in economic activity with those experienced during the last two re-

cessions. We then discuss various linkages between the financial sector and the real economy

and identify three main channels by which disruptions in financial markets can influence

real activity: a pullback in spending owing to reductions in wealth; balance sheet mech-

anisms that lead to a widening of credit spreads, which curtail the ability of households

and businesses to obtain credit; and the direct effect of impairments in the ability of finan-

cial institutions to intermediate credit. Although these channels are well-understood from

a theoretical perspective, assessing their quantitative implications remains a considerable

challenge for macroeconomists. For example, a fall in output that follows a drop in lending

associated with a major financial disruption reflects both supply and demand considera-

tions. In addition, in a world with a rapidly changing financial landscape, it is difficult to

gauge the extent to which various financial asset market indicators provide consistent and

credible information about the relationship between the health of the financial system and

economic activity.

Our analysis consists of two parts. First, we review the recent empirical evidence on

the relationship between corporate credit spreads—the difference in yields between vari-

ous corporate debt instruments and Treasury securities of comparable maturity—and eco-

nomic activity, a link elucidated by the theoretical literature that emphasizes movements in

default-risk indicators as an important signal of disruptions in financial markets. We extend

the standard analysis by attempting to separate empirically the portion of the predictive

content of default-risk indicators for economic activity that reflects the usual countercyclical

movements in expected defaults from the part due to cyclical changes in the relationship

1

Page 3: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

between measures of expected default risk and credit spreads. According to our results,

most of the information content of corporate credit spreads can be attributed to deviations

in the pricing of corporate securities relative to the expected default risk of their issuer, a

finding that suggests that signals about impending financial disruptions embedded in prices

of corporate debt instruments may account for a significant portion of the forecasting power

of credit spreads for economic activity.

To provide further insight into the linkages between the financial sector and the real

economy, we then discuss recent work that seeks to incorporate financial market frictions

into otherwise standard dynamic stochastic general equilibrium (DSGE) models. The aim

of this vein of research is to disentangle movements in the supply and demand for credit by

imposing a structural framework on macroeconomic data. In particular, using quarterly U.S.

data over the 1985:Q1–2008:Q2 period, we estimate a DSGE model based on the financial

accelerator framework developed by Bernanke, Gertler, and Gilchrist [1999].1 Although

the estimated model is relatively simple compared with other work in this area, the results

nonetheless provide a considerable insight into the importance of financial factors in business

cycle fluctuations. In particular, the model estimates suggest that financial disruptions are

responsible for sharp declines in output growth during the last two recessions and that

the easing of financial conditions during the second half of the last decade contributed

importantly to the investment boom of the late 1990s. In addition, the model estimates

imply that the current financial crisis—through its impact on business fixed investment—

appears to be responsible for a considerable portion of the observed slowdown in economic

activity.

2 Current Economic Conditions

In this section, we offer a brief assessment of recent economic developments. To do so,

we compare the time-series evolution of key macroeconomic variables during the current

episode to their evolution during the last two NBER-dated recessions. Figure 1 considers

the behavior of real GDP and its major components; Figure 2 examines the evolution of

monthly indicators of labor market conditions and economic activity; and Figure 3 focuses

on the housing sector. All series in the three figures are plotted as indexes that equal 100

during the quarter (month) that marks the beginning of the recession as dated by the

NBER. For the current episode, all quarterly (monthly) series are benchmarked so that

they equal 100 in 2007:Q2 (October 2007).

This past summer marked a one-year anniversary of the current financial crisis. Accord-

1Other formulations of financial market frictions in general equilibrium models include, for example,Fuerst [1995], Carlstrom and Fuerst [1997], Kiyotaki and Moore [1997], and Cooley, Marimon, and Quadrini[2004].

2

Page 4: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Figure 1: Real GDP and Its Selected Components

-16 -12 -8 -4 0 4 8

90

95

100

105Index peak = 100

Quarters to or from peak

1990:Q32001:Q12007:Q4

Quarterly

Real GDP

-16 -12 -8 -4 0 4 8

85

90

95

100

105

110Index peak = 100

Quarters to or from peak

1990:Q32001:Q12007:Q4

Quarterly

Consumption

-16 -12 -8 -4 0 4 8

75

80

85

90

95

100

105Index peak = 100

Quarters to or from peak

1990:Q32001:Q12007:Q4

Quarterly

Business fixed investment

-16 -12 -8 -4 0 4 8

90

100

110

120

130

140

150Index peak = 100

Quarters to or from peak

1990:Q32001:Q12007:Q4

Quarterly

Residential investment

Note: The panels of the figure depict the behavior of real GDP, real personal consumption, realbusiness fixed investment, and real residential investment at and around cyclical peaks as dated by theNBER. The black and blue lines are indexed to equal 100 during the quarter marking the beginning ofthe NBER-dated recession. The red lines—the current episode—are indexed to equal 100 in 2007:Q4.

ing to Figure 1, however, there is very little evidence that this turmoil has had a significant

negative effect on aggregate demand, as evidenced by the behavior of real GDP, consump-

tion, and business fixed investment. In contrast, output and consumption spending fell

markedly at the onset of the 1990–91 recession, and business fixed investment peaked sev-

3

Page 5: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

eral quarters prior to the 1990:Q3 cyclical peak. Business expenditures on fixed capital

also dropped sharply several quarters prior to the onset of the 2001 recession, though the

deceleration in output and consumption was substantially less severe in 2001 relative to

that experienced during the 1990–91 recession. The cyclical downturn in 2001 was driven

in large part by the collapse of investment in high-tech equipment that followed the burst-

ing of the “tech bubble.” The recession of the early 1990s, in contrast, was considerably

more broader-based, and the slowdown in economic growth reflected, in part, the effects of

“financial headwinds” that impinged on consumer spending. A major difference between

the current episode and the previous two recessions can be seen in the behavior of residen-

tial investment, a component of aggregate demand that peaked in 2005:Q4 and has fallen

60 percent over the subsequent 10 quarters.

Figure 2 examines the evolution of labor market conditions and high-frequency indica-

tors of production and spending. Compared with the GDP and its selected components, the

dynamics of labor market indicators (the unemployment rate and private payroll employ-

ment) in the current situation match fairly closely the changes in labor market conditions

during the previous two recessions. Similarly, output in the factory sector—as measured

by the index of manufacturing industrial production—is beginning to show declines that

appear consistent with those of the past two cyclical downturns. By contrast, growth of

retail sales has remained relatively robust throughout the current period, though consumer

spending has shown some signs of weakness in recent months.

Figure 3 takes a look at the housing sector. As shown in the top left panel, (nominal)

house prices, as measured by the OFHEO purchase-only index, expanded briskly through

2005 before slowing noticeably in 2006. House prices were essentially flat through 2007, and

the first half of 2008 saw an outright decline in the OFHEO house price index. Consistent

with the deceleration in house prices, housing starts and sales of both new and existing

homes have plummeted from their peak levels reached early in 2006—housing starts have

dropped a whopping 100 percent, and home sales have fallen nearly 60 percent. This weak-

ness in the housing sector is also reflected in the massive buildup of new-home inventories,

as evidenced by the run-up in the months’ supply of new homes, which surged almost

60 percent over this period. In contrast, activity in the housing sector showed no signs of

slowing during the 2001 recession, and the downturn of the early 1990s was characterized

by a relatively modest slowdown in housing starts and a minor inventory buildup.

In summary, the seeds of the current economic slowdown can be traced to the sharp

slowdown in house price appreciation and the drop in residential investment that has oc-

curred since early 2006. Somewhat remarkably, both housing starts and home sales have

remained relatively stable—albeit at very weak levels—in recent months despite the tight-

ening of conditions in mortgage credit markets. Similarly, the growth of real GDP, business

4

Page 6: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Figure 2: Labor Market, Production, and Spending Indicators

-48 -40 -32 -24 -16 -8 0 8 16 24

90

100

110

120

130

140

150Index peak = 100

Months to or from peak

July 1990Mar. 2001Oct. 2007

Monthly

Unemployment rate

-48 -40 -32 -24 -16 -8 0 8 16 24

92

94

96

98

100

102Index peak = 100

Months to or from peak

July 1990Mar. 2001Oct. 2007

Monthly

Private payroll employment

-48 -40 -32 -24 -16 -8 0 8 16 24

85

90

95

100

105Index peak = 100

Months to or from peak

July 1990Mar. 2001Oct. 2007

Monthly

Manufacturing industrial production

-48 -40 -32 -24 -16 -8 0 8 16 24

80

85

90

95

100

105

110Index peak = 100

Months to or from peak*Control category.

Mar. 2001Oct. 2007

Monthly

Retail sales*

Note: The panels of the figure depict the behavior of the civilian unemployment rate, nonfarmprivate payroll employment, manufacturing industrial production, and real retail sales (control category)at and around cyclical peaks as dated by the NBER. The black and blue lines are indexed to equal 100during the quarter marking the beginning of the NBER-dated recession. The red lines—the currentepisode—are indexed to equal 100 in October 2007.

fixed investment, and consumption has been relatively well maintained during this period

of prolonged financial turmoil. In contrast, the sharp deterioration of labor market condi-

tions, along with the recent contraction in industrial output, provide compelling evidence

5

Page 7: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Figure 3: Housing Sector

-16 -12 -8 -4 0 4 8

80

90

100

110

120Index peak = 100

Quarters to or from peak

1990:Q32001:Q12007:Q4

Quarterly

OFHEO house price index

-48 -40 -32 -24 -16 -8 0 8 16 24

100

150

200

250Index peak = 100

Months to or from peak

July 1990Mar. 2001Oct. 2007

Monthly

Housing starts

-48 -40 -32 -24 -16 -8 0 8 16 24

80

100

120

140

160Index peak = 100

Months to or from peak*New and existing homes.

July 1990Mar. 2001Oct. 2007

Monthly

Home sales*

-48 -40 -32 -24 -16 -8 0 8 16 24

40

60

80

100

120

140Index peak = 100

Months to or from peak

July 1990Mar. 2001Oct. 2007

Monthly

Months supply of new homes

Note: The panels of the figure depict the behavior of the OFHEO purchase-only house price index,housing starts; home sales (new and existing homes), months’ supply of new homes at and around cyclicalpeaks as dated by the NBER. The black and blue lines are indexed to equal 100 during the quarter(month) marking the beginning of the NBER-dated recession. The red lines—the current episode—areindexed to equal 100 in 2007:Q4 (October 2007).

that economic growth is stalling. Moreover, these indicators suggest that the current de-

celeration in economic activity is comparable—both in its timing and magnitude—to the

slowdowns that occurred during the previous two recessions.

6

Page 8: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

3 Finance and the Real Economy

The benchmark macroeconomic model used to study the behavior of firms and households

is predicated on the assumption that the composition of agents’ balance sheets has no

effect on their optimal decisions. Within this Modigliani-Miller paradigm, households make

consumption decisions based solely on permanent income—the sum of their financial wealth

and the per-period income obtained from the present discounted value of future wages.

Movements in financial asset prices shape agents’ spending decisions to the extent that

they influence households’ financial wealth, whereas changes in interest rates affect spending

decisions because they alter the present discounted values and hence reflect appropriately

calculated user-costs for financing real consumption expenditures. On the business side,

firms make investment decisions by comparing the expected marginal profitability of new

investment projects with the appropriately calculated after-tax user-cost of capital. The

relevant interest rate used in such calculations reflects the maturity-adjusted risk-free rate

of return appropriate to discount the future cash flows.

Financial market imperfections—owing to asymmetric information or moral hazard on

the part of borrowers vis-a-vis lenders—provide a theoretical link between the agents’ finan-

cial health and the amount of borrowing and hence economic activity in which they are able

to engage. Although models differ on details, contracts between borrowers and lenders gen-

erally require that borrowers post collateral or maintain some stake in the project in order

to mitigate the contracting problems associated with such financial market imperfections.

For example, when the borrower’s net worth is low relative to the amount borrowed, the

borrower has a greater incentive to default on the loan. Lenders recognize these incentive

problems and, consequently, demand a premium to provide the necessary external funds.

In general, this external finance premium is increasing in the amount borrowed relative to

the borrower’s net worth. Because net worth is determined by the value of assets in place,

declines in asset values during economic downturns result in a deterioration of borrowers’

balance sheets and a rise in the premiums charged on the various forms of external finance.

The increases in external finance premiums, in turn, lead to further cuts in spending and

production. The resulting reduction in economic activity causes asset values to fall further

and amplifies the economic downturn—the so-called financial accelerator mechanism.

Although the theoretical impact of changes in financial conditions on household and

business spending decisions through the financial accelerator mechanism is well understood,

quantifying the overall strength of this mechanism remains a challenge for macroeconomists.

This task is complicated by the fact that it is very difficult to distinguish the effect of a

slowdown in economic activity on household and firm spending owing to the usual demand

channels absent financial market frictions from the effect that such a slowdown may have

7

Page 9: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

through the financial accelerator itself. Nonetheless, a careful assessment of the empirical

implications of models that allow for financial frictions relative to those that assume per-

fect capital markets have allowed researchers to make substantial progress in assessing the

empirical relevance of changes in financial conditions for real activity.

On the household side, the permanent income model of consumption has stark impli-

cations for the responsiveness of consumption to both income and asset values. Transitory

changes in income have very little effect on permanent income and hence consumption.

Reasonably calibrated versions of such models imply that households are relatively insensi-

tive to changes in asset values, suggesting that households should increase consumption by

three to four cents for every dollar increase in their financial wealth. More importantly, to

a first approximation, the value of housing does not represent net wealth for the household

sector, because an increase in house values is also an increase in the implicit rental cost of

housing. As a result, the household sector is no better or worse off when housing values

rise; see Buiter [2008] for a thorough discussion.

Empirical research provides compelling evidence against the permanent income model

of consumption in favor of models in which the quality of household balance sheets plays an

important role in determining their consumption decisions. A variety of studies shows that

household consumption is excessively sensitive to movements in transitory income. Whereas

the exact cause of this excess sensitivity is subject to a considerable debate, the excess sensi-

tivity is generally attributed to the fact that at least a subset of households faces significant

borrowing constraints or engages in precautionary-savings behavior because of imperfect

insurance. Similarly, in contrast to the predictions of the permanent income model, both

microeconomic and macroeconomic studies suggest an important link between house prices

and household consumption (see, for example, Case, Quigley, and Shiller [2005]; Carroll, Ot-

suka, and Slacalek [2006]; and Campbell and Cocco [2008]). Estimates of the housing wealth

effect vary but generally imply that household consumption increases by an amount ranging

from 3 to 10 cents for every dollar increase in housing wealth. This response is generally

attributed to the fact that at higher equity levels, households can obtain larger home mort-

gage loans and thus maintain high consumption levels while financing a home. Similarly,

existing home owners may engage in mortgage equity withdrawals to finance high levels of

consumption relative to their income. Although the estimated sensitivity of consumption

to housing values appears small, the significant decline in U.S. house prices experienced

during the last two years would imply a substantial drag on household consumption.

Empirical research also provides evidence that supports the notion that corporate bal-

ance sheets influence investment spending, though this evidence is more contentious. It is

well known that business investment spending is strongly correlated with corporate cash

flow. Earlier research, initiated by Fazzari, Hubbard, and Petersen [1988], has argued that

8

Page 10: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

cash flows stimulate investment because internal funds are a cheaper source of finance than

external funds. Critics, however, point out that current cash flows may also provide signals

about future profits, which, in turn, determine the firm’s net worth and hence the strength

of its balance sheet. That said, the available evidence suggests that the cash flow mechanism

is quite strong for smaller firms, firms with a limited access to corporate credit and equity

markets, or firms with weak balance sheets (see, for example, Gilchrist and Himmelberg

[1995]). Some of more recent research has questioned the macroeconomic relevance of this

effect by arguing that for large firms that account for the bulk of investment spending,

current cash flows serve mainly as signals about future profit opportunities rather than

indicators of the strength of their balance sheets (see, for example, Cummins, Hassett, and

Oliner [2006] and Rebelo, Eberly, and Vincent [2008]). Nonetheless, studies that analyze

investment spending during financial crises show that large negative shocks to firms’ bal-

ance sheets can have important adverse consequences for the investment decisions of large

firms, at least during periods of acute financial distress (see, for example, Aguiar [2005] and

Gilchrist and Sim [2007]).

According to the currently available data, corporate balance sheets appear to be in

relatively good conditions. The amount of liquid assets on the balance sheets of nonfinancial

firms is high by historical standards, and corporate profits have been surprisingly well

maintained in light of the persistent strains in financial markets. At the same time, credit

spreads on a wide variety of corporate debt instruments have widened significantly since

the middle of last year, a development that is consistent with a deterioration in the overall

financial condition of the corporate sector or a worsening of conditions within the financial

sector that serves as an originator and guarantor of corporate debt instruments. Although

macroeconomic evidence offers a mixed guidance on the importance of interest rates for

investment spending, recent work by Gilchrist and Zakrajsek [2007] using firm-level data

shows that investment is highly responsive to changes in corporate credit spreads. Thus,

although the corporate sector has maintained relatively strong balance sheets, it is still the

case that rising credit spreads may be reducing current investment spending. (We return

to this issue below.)

The financial mechanism linking balance sheet conditions of borrowers to real activity is

often described as the “broad credit channel.” Financial institutions are also likely to suffer

from asymmetric information and moral hazard problems when raising funds to finance

their lending activities. The focus of this so-called “narrow credit channel” is the health of

financial intermediaries and its impact on the ability of financial institution to extend credit.

In a fractional reserve banking system, deposits provide a source of funds for lending with

only a small fraction of total deposits held as reserves. Because a tightening of monetary

policy drains reserves from the banking system, poorly capitalized banks that are unable

9

Page 11: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

to raise external funds cut back on their lending. Bank-dependent borrowers, in particular

small firms and households that have few alternative sources of credit, reduce spending.

Kashyap and Stein [2000] document the empirical validity of this mechanism by showing

that small U.S. commercial banks that are poorly capitalized are especially sensitive to

changes in the stance of monetary policy. Although this bank lending channel appears to

have important effects on the lending behavior of smaller banks, such banks account for only

a small fraction of total bank lending in the United States, which suggests that the bank

lending channel may not be a quantitatively important channel through which monetary

policy affects the real economy. In a recent paper, however, Cetorelli and Goldberg [2008]

argue that this lending channel may also be at work at large commercial banks operating

primarily in domestic markets. In contrast, commercial banks with global operations are

able to offset declines in domestic deposits through internal funds obtained from their

global subsidiaries. In times of a worldwide financial distress, however, the ability of global

subsidiaries to provide internal funds to U.S. financial institutions is also likely to be limited

in scope, a development that would further strengthen the bank lending channel.

Although monetary policy may not have a large direct impact through the bank-lending

channel, reductions in bank capital during economic downturns can also reduce lending

activity. As economic activity slows and defaults rise, the quality of bank loan portfo-

lios deteriorates. Banks seeking to shore up their capital or to meet regulatory capital

requirements tighten their credit standards and cut back on lending, an inward shift in

loan supply that curtails spending of bank-dependent borrowers (see, for example, Van den

Heuvel [2007].) The strength of this mechanism depends on the overall health of the bank-

ing sector and on the extent to which firms and households are bank dependent. In the

United States, the bulk of investment spending is financed by relatively large firms that

rely primarily on corporate bond and equity markets to finance their capital expenditures.

Nonetheless, certain corporate debt instruments—most notably commercial paper—are typ-

ically backed by lines of credit at commercial banks. In addition, a substantial portion of

business financing through commercial and industrial loans relies on such credit lines. In

times of financial turmoil, even large nonfinancial firms may have a difficult time raising

capital in arms-length markets. As these firms tap their backup lines of credit to finance

inventories or operating expenditures in the face of falling revenues, banks may be forced

to make further cuts in lending to bank-dependent borrowers.

In summary, the recent drop in house prices is likely to bring about a reduction in

consumption spending through its impact on household borrowing and mortgage equity

withdrawals. In addition, the usual wealth channel implies that recent declines in stock

prices may also reduce household consumption, though empirical estimates suggest this

effect is likely to be relatively modest. Because nonfinancial corporate balance sheets remain

10

Page 12: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

relatively strong, investment spending is likely to remain relatively solid but will slow to the

extent that corporate borrowing rates remain elevated because of persistent and intensifying

strains in financial markets.

The direct effect of falling values of assets held by the financial sector is more difficult

to assess. Although there is clear evidence that reductions in bank capital have important

implications for the lending behavior of small banks, there is less direct evidence to support

the claim that a capital channel has important implications for the lending behavior of

large banks and nonbank financial intermediaries. Nonetheless, a sharp pullback in lending

by large commercial banks and nonbank financial institutions during the current financial

crisis—owing to lack of liquidity in the interbank funding markets or a retrenchment in

lending as these institutions seek to replenish depleted capital—would likely cause a severe

slowdown in economic activity by constricting the supply of credit. In particular, the

usual mechanisms that allow nonfinancial firms to substitute away from bank loans and

other intermediated credit towards arms-length borrowing may become nonoperational in

times of a widespread and acute financial distress, especially given the crucial role that the

nonbank financial institutions play in originating, marketing, and guaranteeing debt issued

by the nonfinancial business sector.

4 Corporate Credit Spreads and Economic Activity

Credit spreads have long been used to gauge the degree of strains in the financial system.

Because asset prices are forward looking, movements in credit spreads have been shown

to be particularly useful for forecasting economic activity.2 Despite some success, results

from this strand of research are often sensitive to the choice of a credit spread index under

consideration. Moreover, credit spread indexes that contained useful information about

macroeconomic outcomes in the past often lose their predictive power for the subsequent

cyclical downturn. These mixed results are partly attributable to the rapid pace of financial

innovation that likely alters the forecasting power of financial asset prices over time or results

in one-off developments that may account for most of the forecasting power of a given credit

spread index.

To mitigate these problems, Gilchrist, Yankov, and Zakrajsek [2008] (GYZ hereafter)

rely on the prices of individual senior unsecured corporate debt issues traded in the sec-

ondary market to construct a broad array of corporate bond spread indexes that vary across

maturity and default risk. Compared with other corporate financial instruments, senior un-

secured bonds represent a class of securities with a long history containing a number of

2The predictive content of various corporate credit spreads for economic activity has been analyzed,among others, by Stock and Watson [1989]; Friedman and Kuttner [1998]; Gertler and Lown [1999]; Mueller[2007]; and King, Levin, and Perli [2007].

11

Page 13: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Table 1: Predictive Content of Credit Spreads for Economic Activity

(Year-Ahead Forecast Horizon)

Real GDP Employment Ind. Production

Credit Spread Est. R2 Est. R

2 Est. R2

CP1m−Treas1m -0.119 0.048 -0.258 0.398 -0.189 0.132[0.98] - [2.40] - [1.41] -

Aaa−Treas10y -0.015 0.044 -0.132 0.351 -0.194 0.132[0.07] - [0.56] - [0.89] -

Baa−Treas10y -0.081 0.039 -0.243 0.370 -0.273 0.151[0.38] - [1.05] - [1.30] -

HY−Treas10y -0.435 0.147 -0.687 0.600 -0.581 0.298[1.49] - [3.56] - [2.28] -

EDF-Q1 -0.529 0.320 -0.602 0.702 -0.646 0.511[4.66] - [5.90] - [5.14] -

EDF-Q2 -0.588 0.374 -0.656 0.763 -0.723 0.583[4.80] - [7.66] - [5.65] -

EDF-Q3 -0.585 0.359 -0.653 0.752 -0.691 0.517[4.81] - [8.41] - [4.54] -

Note: For real GDP, sample period is 1990:Q2–2008:Q2 (T = 69). For private nonfarmemployment (EMP) and industrial production (IP), sample period is 1990:II–2008:VII(T = 210). Dependent variables are ∆4 ln(GDPt+4), ∆12 ln(EMPt+12), and ∆12 ln(IPt+12).Each regression specification includes a credit spread, a 4-quarter or a 12-month lag of therespective dependent variable, and a constant term (the latter two effects are not reported)and is estimated by OLS. Estimates of parameters corresponding to credit spreads arestandardized; absolute t-statistics reported in brackets are based on a heteroscedasticity-and autocorrelation-consistent asymptotic covariance matrix computed according to Neweyand West [1987].

business cycles. In addition, the rapid pace of financial innovation has done little to alter

the basic structure of these securities. Thus, the information content of spreads constructed

from yields on senior unsecured corporate bonds is likely to provide more consistent signals

regarding economic outcomes relative to spreads based on securities with a shorter history

or securities whose structure or the relevant market has underwent a significant structural

change. In addition, GYZ rely on the firm-specific expected default frequencies (EDFs)

provided by the Moody’s/KMV corporation to construct their credit spread indexes. Be-

cause they are based primarily on observable information in equity markets, EDFs provide

a more objective and more timely assessment of firm-specific credit risk compared with the

issuer’s senior unsecured credit rating.

The results in Table 1 examine the predictive content of various corporate credit spread

12

Page 14: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

indexes for the following three measures of economic activity: real GDP, private nonfarm

payroll employment, and industrial production. Specifically, we estimate a simple fore-

casting equation in which the year-ahead growth in an indicator of economic activity is

regressed on its own value lagged one year and the current value of a credit spread. The

entries in the table correspond to the standardized coefficient estimates (with t-statistics

in brackets) on the effect of the credit spread on each measure of economic activity, along

with the explanatory power of the regression as measured by the adjusted R2.

The first four regressions employ standard credit spread indexes emphasized in

this literature. These include the one month commercial paper Treasury bill spread

(CP1mo−Treas1mo); the Aaa corporate bond spread (Aaa−Treas10y), the Baa corporate

bond spread (Baa−Treas10y); and the high-yield corporate bond spread (HY−Treas10y).3

According to the entries in the table, the standard credit spread indexes—with the excep-

tion of the high-yield bond spread, which contains substantial explanatory power for the

year-ahead growth in employment—contain very little information regarding the future di-

rection in economic activity. The next set of forecasting regressions relies on a subset of

the corporate bond spread indexes constructed by GYZ. Specifically, we focus on the credit

spread indexes for which GYZ document the highest predictive content, both in and out

of sample—that is, credit spreads constructed from very long-maturity bonds (remaining

term-to-maturity greater than 15 years) issued by firms in the low- and intermediate-risk

categories as defined by the lowest three quintiles of the EDF distribution (EDF-Q1, EDF-

Q2, and EDF-Q3). Compared with the standard default-risk indicators, the EDF-based

credit spread indexes contain significant predictive power for all three measures of eco-

nomic activity. The coefficients on the EDF-based credit spreads are always statistically

and economically significant, and the EDF-based credit spreads generate an in-sample fit

that is substantially above that obtained from regressions that rely on the standard credit

spread indexes.

Table 2 focuses on the predictive content of these credit spread indexes for total business

fixed investment spending and its major components—namely, equipment and software

(E&S, excluding high tech); high-tech equipment; and nonresidential structures. Again,

the paper-bill spread and the credit spread indexes based on Aaa- and Baa-rated long-term

corporate bonds have very little explanatory power for total investment spending or its

major components. The high-yield bond spread does forecast total investment spending,

though not nearly as well as the EDF-based credit spreads. The high-yield spread appears

3Commercial paper rates are taken from the “Commercial Paper Rates and Outstanding” Federal Reservestatistical release. The source of Treasury yields and yields on Aaa- and Baa-rated corporate bonds is“Selected Interest Rates” (H.15) Federal Reserve statistical release. To construct the high-yield spread,we use the High-Yield Master II index, a commonly used benchmark index for long-term speculative-gradecorporate bonds administered by Merrill Lynch.

13

Page 15: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Table 2: Predictive Content of Credit Spreads for Business Fixed Investment

(Year-Ahead Forecast Horizon)

INV-TOT INV-ES INV-HT INV-NRS

Credit Spread Est. R2 Est. R

2 Est. R2 Est. R

2

CP1m−Treas1m -0.277 0.343 -0.291 0.273 -0.049 0.337 -0.253 0.160[2.32] - [2.14] - [0.30] - [1.66] -

Aaa−Treas10y -0.286 0.343 -0.016 0.186 -0.393 0.467 -0.385 0.231[0.98] - [0.08] - [1.59] - [1.31] -

Baa−Treas10y -0.334 0.337 -0.032 0.187 -0.409 0.434 -0.455 0.243[1.00] - [0.18] - [1.33] - [1.62] -

HY−Treas10y -0.812 0.559 -0.357 0.254 -0.289 0.387 -0.870 0.598[2.94] - [1.30] - [0.94] - [6.15] -

EDF-Q1 -0.610 0.653 -0.740 0.695 -0.506 0.598 -0.288 0.182[4.85] - [6.49] - [3.01] - [1.52] -

EDF-Q2 -0.656 0.692 -0.791 0.721 -0.532 0.607 -0.351 0.226[5.08] - [8.87] - [2.95] - [1.83] -

EDF-Q3 -0.662 0.677 -0.827 0.732 -0.522 0.581 -0.347 0.224[4.71] - [11.4] - [2.55] - [1.69] -

Note: Sample period: quarterly data from 1990:Q2 to 2008:Q2 (T = 69). Dependent variables are∆4 ln(INVt+4), where INV denotes real investment spending in the following categories: INV-TOT =total business fixed investment; INV-ES = equipment and software (excluding high tech); INV-HT =high-tech; and INV-NRS = nonresidential structures. Each regression specification includes a creditspread, a 4-quarter lag of the respective dependent variable, and a constant term (the latter two ef-fects are not reported) and is estimated by OLS. Estimates of parameters corresponding to creditspreads are standardized; absolute t-statistics reported in brackets are based on a heteroscedasticity-and autocorrelation-consistent asymptotic covariance matrix computed according to Newey and West[1987].

to contain substantial predictive power for investment in nonresidential structures, whereas

the EDF-based credit spreads forecast E&S and high-tech investment, but they have limited

information content for future expenditures on nonresidential structures.

Although not reported, our results indicate that corporate credit spreads have essentially

no information content for future consumption spending on both durable and nondurable

goods as well as for residential investment. This lack of predictive power holds true for both

the standard credit spread indexes and the EDF-based default risk indicators considered by

GYZ. Thus, corporate credit spreads do well at predicting business spending but contain

little information for household spending.

According to Gertler and Lown [1999], the predictive content of credit spreads for eco-

nomic activity may be due to the presence of an operative financial accelerator mechanism

14

Page 16: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

linking balance sheet conditions to the real economy through movements in the external

finance premium. As emphasized by Philippon [2008], however, the forecasting ability of

credit spreads may also reflect the fact that asset prices contain information about future

economic fundamentals in a world without financial market imperfections. In particular, as

economic fundamentals deteriorate, default risk will rise, even if defaults impose no addi-

tional dead-weight loss on the economy. According to this view, an increase in credit spreads

may reflect a reduction in expected future profits during an impending cyclical downturn

and does not necessarily provide a causal link by which movements in the external finance

premium either amplify economic disturbances originating in the real economy or exert

an independent effect on economic activity through financial disruptions that reduce the

supply of credit.

As argued by Gilchrist, Ortiz, and Zakrajsek [2008], one potential way to parse move-

ments in credit spreads between fluctuations owing to financial market imperfections and

swings reflecting solely the changes in expected default risk absent financial frictions is to

use a pricing model for corporate debt to measure deviations between the current level of

spreads and the level of spreads that should prevail assuming the usual pricing of default

risk. For example, if default becomes more costly during an economic downturn, one would

expect credit spreads to rise relative to an increase in expected default risk. Thus, devia-

tions of credit spreads from those predicted by a standard pricing model may provide direct

information about movements in the external finance premium over the business cycle.

Table 3 reports the results of such an exercise. In particular, using the monthly panel

data set of individual senior unsecured bond issues described in GYZ, we regress the cor-

porate bond spreads in month t on the expected default risk, as measured by the issuer’s

year-ahead EDF at the end of month t − 1; the regression specification also includes the

bond’s duration and par size to control for term and liquidity premiums. As shown in

column 1 of the table, the coefficient on the EDF is economically large and highly statis-

tically significant. Moreover, this reduced-form pricing model explains about 46 percent of

the variation in corporate bond spreads. Columns 2 and 3 report regression results that

allow for a nonlinear relationship between credit spreads and expected default risk, whereas

column 4—in addition to equity-based indicators of default-risk—also includes a full set

of credit rating dummies. As evidenced by the entries in the table, allowing for nonlinear

effects of expected default risk on credit spreads provides an incremental improvement in

the explanatory power of the regression. In contrast, the inclusion credit ratings fixed ef-

fects leads to a substantial increase in the goodness of fit. The specification that allows for

both nonlinearities in the relationship between credit spreads and expected default risk and

credit ratings fixed effects explains almost 62 percent of the variation in corporate bond

spreads.

15

Page 17: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Table 3: Corporate Bond Spreads and Expected Default Risk

Regression Specification

Explanatory Variable (1) (2) (3) (4)

ln(PARVALUE) 0.088 0.080 0.077 0.032(0.016) (0.016) (0.016) (0.014)

ln(DURATION) -0.002 0.004 0.005 0.050(0.104) (0.101) (0.010) (0.008)

ln(EDF−1) 0.362 0.428 0.487 0.293(0.010) (0.016) (0.015) (0.012)

ln(EDF−1)2 - 0.035 0.012 0.004

(0.005) (0.006) (0.004)ln(EDF−1)

3 - - -0.012 -0.008(0.002) (0.001)

Adj. R2 0.459 0.472 0.477 0.618

Industry Effectsa yes yes yes yes(0.000) (0.000) (0.000) (0.000)

Ratings Effectsb no no no yes(0.000)

Note: Sample period: monthly bond-level data from February 1990 toJuly 2008 (Obs. = 281,179). Dependent variable is the log of the credit spreadin month t. All specifications are estimated by OLS. Asymptotic robust standarderrors are clustered at the firm level and are reported in parentheses.

ap-values for the test of the null hypothesis of the absence of fixed industryeffects are reported in parentheses.

bp-values for the test of the null hypothesis of the absence of fixed credit ratingeffects are reported in parentheses.

We use the reduced-form pricing model in column 4 to construct the residual or pric-

ing error for each bond/month observations in the GYZ data set. We then calculate the

average of these pricing errors for each period to obtain an index that essentially removes

the average effect of movements in expected default risk on corporate bond spreads. As in

Gilchrist, Ortiz, and Zakrajsek [2008], we label this index the external finance premium.

Figure 4 shows the estimated external finance premium along with, for comparison pur-

poses, the credit spread for very long maturity bonds in the second quintile of the EDF

distribution constructed by GYZ (EDF-Q2), and the standard high-yield credit spread in-

dex (HY−Treas10y).4 All three series show substantial variation and comovement over the

business cycle, although the high-yield spread is considerably more volatile than the other

4In order to cast the external finance premium index in units that are easily interpretable, we scaled theindex such that it is in the same units as the average corporate spread in the GYZ data set.

16

Page 18: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Figure 4: Corporate Bond Spreads and the External Finance Premium

1990 1992 1994 1996 1998 2000 2002 2004 2006 2008

200

400

600

800

1000

1200Basis points

High YieldEDF-Q2External Finance Premium

Monthly

Note: The black depicts the high-yield corporate bond spread (HY−Treas10y); the red line depictsthe credit spread associated with very long maturity corporate bonds in the second quintile of the EDFdistribution (EDF-Q2); and the red line depicts the estimated external finance premium (see text fordetails). The shaded vertical bars denote NBER-dated recessions.

two credit spread indexes. Focusing on the current period of financial turmoil, the esti-

mated external finance premium shot up in the middle of 2007 and—at just under 400 basis

points—is currently substantially above the peak reached at the end of 2000. Thus, although

measured default risk may have been significantly higher during the 2001–02 period, the

external finance premium is in fact higher in the current period than at any time in the

past 19 years.

We now examine the extent to which movements in the external finance premium help

predict the year-ahead growth in real GDP, private payroll employment, and industrial

production. Table 4 contains the results of this exercise; the regression specifications in

Table 4 are identical to those reported in Table 1, except that we now use the estimated

external finance premium in place of the credit spread indexes. As evidenced by the entries

in the table, the forecasting power of the external finance premium is nearly as good as that

of the EDF-based credit spread indexes. The coefficient on the external finance premium is

highly statistically and economically significant for all three measures of economic activity,

and the in-sample fit of all three specifications is almost as good as the highest adjusted R2

reported in Table 4. Overall these results suggest that a significant portion of the predictive

power of corporate bond spreads for economic activity likely reflects the information content

17

Page 19: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Table 4: Predictive Content of External Finance Premium for Economic Activity

(Year-Ahead Forecast Horizon)

Real GDP Employment Ind. Production

Explanatory Variable Est. R2 Est. R

2 Est. R2

External Finance Premium -0.560 0.294 -0.673 0.731 -0.733 0.534[2.79] - [5.64] - [3.88] -

Note: For real GDP, sample period is 1990:Q2–2008:Q2 (T = 69). For private nonfarm employ-ment (EMP) and industrial production (IP), sample period is 1990:II–2008:VII (T = 210). Dependentvariables are ∆4 ln(GDPt+4), ∆12 ln(EMPt+12), and ∆12 ln(IPt+12). Each regression specification in-cludes the estimated external finance premium (see text for details), a 4-quarter or a 12-month lag ofthe respective dependent variable, and a constant term (the latter two effects are not reported) and isestimated by OLS. Estimates of parameters corresponding to credit spreads are standardized; abso-lute t-statistics reported in brackets are based on a heteroscedasticity- and autocorrelation-consistentasymptotic covariance matrix computed according to Newey and West [1987].

of credit spreads for disruptions in financial markets or variation in the cost of default, two

factors that would cause credit spreads to widen relative to expected default risk prior to

an economic downturn.

5 DSGE Models with Financial-Real Linkages

The ability of credit spreads to predict economic activity suggests important linkages be-

tween financial conditions and macroeconomic outcomes. Quantifying these linkages, how-

ever, requires structural models of the macroeconomy that can distinguish between move-

ments in credit supply and demand and that can account for general equilibrium feedback

effects between developments in the financial and real sectors of the economy. Recent work

by Christiano, Motto, and Rostagno [2007], Queijo von Heideken [2008], Graeve [2008],

and Christensen and Dib [2008] seeks to quantify these mechanisms by estimating dynamic

stochastic equilibrium models that incorporate credit market imperfections through the

financial accelerator mechanism described in Carlstrom and Fuerst [1997] and Bernanke,

Gertler, and Gilchrist [1999] (BGG hereafter).5

Although details differ in terms of model estimation and shock specification, all of

these papers document an important role for financial factors in business cycle fluctua-

tions. Queijo von Heideken [2008], for example, shows that the ability of a model with a

rich array of real and nominal rigidities to fit both the U.S. and the Euro-area data im-

5In an alternative approach, Levin, Natalucci, and Zakrajsek [2006] employ firm-level data on creditspreads, EDFs, and leverage to estimate directly the structural parameters of the debt-contracting problemunderlying the financial accelerator model of BGG.

18

Page 20: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

proves significantly if one allows for the presence of a financial accelerator mechanism; and

Christiano, Motto, and Rostagno [2007] demonstrate that shocks to the financial sector

have played an important role in economic fluctuations over the past two decades, both in

the United States and in Europe. Queijo von Heideken [2008], however, estimates a struc-

tural model that is identified without reliance on financial data and that does not allow

for shocks to the financial sector, whereas Christiano, Motto, and Rostagno [2007], though

allowing for a wide variety of shocks to the financial sector, do not estimate the parameters

governing the strength of the financial accelerator mechanism. To date, we are aware of no

empirical work that seeks to estimate simultaneously the key parameters of the financial

accelerator mechanism along with the shocks to the financial sector.

In this section, we briefly summarize the ongoing work by Gilchrist, Ortiz, and Zakrajsek

[2008] (GOZ hereafter) that attempts to fill this gap. In particular, GOZ use Bayesian max-

imum likelihood methods to estimates a dynamic New Keynesian model that incorporates

the financial accelerator discussed in BGG. The main innovation of their approach is that

they incorporate explicitly estimates of the external finance premium constructed from the

reduced-form pricing models of corporate debt. These proxies for the unobservable external

finance premium are used to identify the strength of the financial accelerator mechanism

and to measure the extent to which disruptions in financial markets have contributed to

fluctuations in the real economy during the last two decades.

For tractability, the model is kept purposefully simple. As in BGG, it allows for a house-

hold sector that consumes, saves, and makes labor-supply decisions; an investment goods

sector that transforms current output into capital via an adjustment cost mechanism; and

a retail sector that faces Calvo-style price rigidities that result in a standard New Keyne-

sian Phillips curve. The model also allows for both habit formation in consumption and

for higher-order adjustment costs in investment. These adjustment costs imply that asset

prices—the value of capital in place—increase during economic expansions. Monetary pol-

icy is conducted by a modified Taylor rule that assumes that the monetary authority, given

interest-rate smoothing, adjusts nominal short-term interest rates in response to changes in

current inflation and output growth.

As in BGG, the model also allows for an entrepreneurial sector that faces significant

credit market frictions in the process of owning and operating the existing capital stock.

These frictions give rise to an external finance premium that creates a wedge between the

required return on capital—the rate at which entrepreneurs can borrow to finance capital

accumulation—and the risk-free rate of return received by the household sector for its

savings. In this environment, an expansion in output causes an increase in the value of

assets in place and a rise in the entrepreneurial net worth. As entrepreneurs’ net worth

expands relative to their borrowing, the external finance premium falls, causing a further

19

Page 21: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

increase in both asset values and investment demand. These general equilibrium feedback

effects, in turn, further amplify the financial accelerator mechanism.

The model is estimated using Bayesian maximum likelihood techniques over the period

1985:Q1–2008:Q2, using data on real GDP, business fixed investment, CPI inflation, the

nominal federal funds rate, and a risk spread derived from the large panel of issuer-level

credit spreads discussed above.6

The estimated model parameters include the degree of habit formation in consumption;

adjustment costs to investment; the response of inflation to the output gap in the Phillips

curve; and the coefficients that determine the monetary policy rule. GOZ also estimate

the elasticity of the external finance premium to changes in net worth, the key parameter

governing the strength of the financial accelerator. In addition to the standard set of

shocks to household preferences, technology, and monetary policy, the model allows for an

exogenous serially-correlated shock to the external finance premium. GOZ interpret this

shock as a disturbance to the financial sector that boosts the external finance premium

beyond the level warranted by the current economic conditions and the current stance of

monetary policy. Consistent with the recent work this area, the estimates of the model

parameters indicate an important macroeconomic role for financial market frictions, which

act as an amplification mechanism for real and nominal disturbances in the economy. The

results also suggests that disturbances that originate in the financial sector have significant

real consequences.

The left column of Figure 5 depicts the model dynamics in response to a one standard

deviation (negative) shock to monetary policy rule, whereas the right column shows the

impulse responses to a one standard deviation (negative) shock to the external finance

premium. Unanticipated expansionary monetary policy causes an increase in output and

investment and a rise in inflation (not shown). It also causes a reduction in the external

finance premium, which serves to strengthen the monetary transmission channel through

the amplification mechanism described above. Similarly, a reduction in the external finance

premium also causes an expansion in investment and output and a fall in inflation (not

shown) through the supply-side benefits of increased capital accumulation. The real effect

of this mechanism is quantitatively large—a 20 basis point decline in the external finance

premium causes an 80 basis point increase in output.

To understand the implications of the model for the conduct of monetary policy and to

6The estimation period uses data prior to 1990, the first year for which monthly estimates of expecteddefault frequencies from Moody’s/KMV are available. Consequently, we are unable to use our proxy for theexternal finance premium derived from the reduced-form pricing model discussed above. We are currently inthe process of estimating the Merton [1974] distance-to-default bond pricing model for the entire nonfinancialcorporate sector going back to 1980, which will allows us to construct a similar estimate of the external financepremium over a longer period. For the purposes of this exercise, we constructed the risk spread by extractingthe first principal component from credit spreads in a large number of credit-risk portfolios.

20

Page 22: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Figure 5: Model Responses to Selected Shocks

Monetary Policy Shock Financial Shock

0 4 8 12 16 20 24 28 32 36 40

−0.5

−0.4

−0.3

−0.2

−0.1

0.0

0.1Percentage points

Quarters after shock

Federal funds rate

0 4 8 12 16 20 24 28 32 36 40

−0.5

−0.4

−0.3

−0.2

−0.1

0.0

0.1Percentage points

Quarters after shock

0 4 8 12 16 20 24 28 32 36 40

−0.3

−0.2

−0.1

0.0

0.1

0.2Percentage points

Quarters after shock

Risk spread

0 4 8 12 16 20 24 28 32 36 40

−0.3

−0.2

−0.1

0.0

0.1

0.2Percentage points

Quarters after shock

0 4 8 12 16 20 24 28 32 36 40

−0.1

0.0

0.1

0.2

0.3

0.4

0.5Percentage points

Quarters after shock

Output growth

0 4 8 12 16 20 24 28 32 36 40

−0.1

0.0

0.1

0.2

0.3

0.4

0.5Percentage points

Quarters after shock

0 4 8 12 16 20 24 28 32 36 40

−0.2

0.0

0.2

0.4

0.6

0.8

1.0Percentage points

Quarters after shock

Output growth

0 4 8 12 16 20 24 28 32 36 40

−0.2

0.0

0.2

0.4

0.6

0.8

1.0Percentage points

Quarters after shock

0 4 8 12 16 20 24 28 32 36 40

−0.5

0.0

0.5

1.0

1.5

2.0Percentage points

Quarters after shock

Investment growth

0 4 8 12 16 20 24 28 32 36 40

−0.5

0.0

0.5

1.0

1.5

2.0Percentage points

Quarters after shock

0 4 8 12 16 20 24 28 32 36 40

−1

0

1

2

3

4Percentage points

Quarters after shock

Investment growth

0 4 8 12 16 20 24 28 32 36 40

−1

0

1

2

3

4Percentage points

Quarters after shock

Note: The red lines in each panel depicts the estimated impulse responses of selected variablesto monetary policy shock (the left column) and to the financial shock (the right column). The shadedbands denote the 80 percent confidence intervals.

evaluate the importance of financial market frictions in determining business cycle outcomes,

we calculate the portion of the movements in the actual growth of output and investment,

nominal federal funds rate, and the risk spread that can be accounted for by monetary

policy innovations and shocks to the supply of credit. Figures 6–7 contain the result of

this exercise. Each panel of these two figures, shows the actual series—in percentage point

deviations from steady state—along with the estimated contribution of the two shocks.

Figure 6 summarizes the effects of a monetary policy shock, whereas Figure 7 focuses on

21

Page 23: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Figure 6: Historical Decomposition of Monetary Policy Shocks

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

−2

−1

0

1

2

3Percentage points

ActualContribution of monetary policy shock

Output growth

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

−5

0

5

10

15Percentage points

Investment growth

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

−3

−2

−1

0

1

2

3

4Percentage points

Federal funds rate

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007−1.0

−0.5

0.0

0.5

1.0Percentage points

Risk spread

Note: The solid black lines in each panel depicts the behavior of actual variables expressed inpercentage point deviations from the steady state. The dotted red line in each panel depict the estimatedeffect of monetary policy shocks (see text for details). The shaded vertical bars denote NBER-datedrecessions.

22

Page 24: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Figure 7: Historical Decomposition of Financial Shocks

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

−2

−1

0

1

2

3

4Percentage points

ActualContribution of financial shock

Output growth

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

−10

−5

0

5

10

15Percentage points

Investment growth

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

−3

−2

−1

0

1

2

3

4Percentage points

Federal funds rate

1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

−1.0

−0.5

0.0

0.5

1.0

1.5

2.0Percentage points

Risk spread

Note: The solid black lines in each panel depicts the behavior of actual variables expressed inpercentage point deviations from the steady state. The dotted red line in each panel depict the estimatedeffect of financial shocks (see text for details). The shaded vertical bars denote NBER-dated recessions.

23

Page 25: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

the financial shock.

As shown in the four panels of Figure 6, the effect of monetary policy shocks on the

economy accord well with the historical record regarding the conduct of monetary policy

since the mid-1980s. Monetary policy was tight in the late 1980s prior to the onset of 1990–

91 recession but was eased substantially during the economic downturn of the early 1990s.

According to our estimates, tight monetary policy also contributed to the slowdown in

business investment and output during the 1994–95 period. The stance of monetary policy

was roughly neutral up through the collapse of the stock market in early 2000, and according

to our estimates, policy was eased significantly during the 2001 recession. Monetary policy

was again relatively tight during the housing boom of the 2005–07. The rapid sequence of

cuts in the federal funds rate during 2007 also appears as a significant easing of monetary

conditions that has supported expansion in investment and output during that period. An

appealing feature of this model is that the monetary transmission mechanism works in part

through its impact on balance sheet conditions—that is, the external finance premium is

strongly countercyclical in response to monetary policy shocks.

The estimated effects of financial disturbances and their impact on the real economy

also accord well with historical perceptions of the likely effects of tight credit conditions on

economic activity. According to our estimates, the economy showed signs of financial distress

at the onset of the 1990–91 recession, and adverse financial conditions remained a drag on

the real economy throughout the “jobless” recovery of the early 1990s. Indeed, between

1989 and 1993, shocks to the financial sector caused the external finance premium to rise by

150 basis points, an increase that led to an extended period of subpar economic performance.

Credit conditions improved markedly during the second half of the 1990s, a period during

which the external finance premium fell about 250 basis points. The premium moved higher

after the bursting of the “dot-com” bubble, and financial conditions deteriorated further at

the onset of the collapse in the housing sector in 2005. The model also captures the current

financial crisis as a shock to the financial sector, manifested as a 75 basis point jump in

the external finance premium that has led to a sharp slowdown in the growth of investment

and output during the last four quarters.

In summary, this relatively simple model of the financial accelerator—when estimated

using both real and financial market data—does remarkably well at capturing much of the

historical narrative regarding the conduct of monetary policy and developments in financial

markets that led to episodes of financial excess and distress over the last two decades. De-

spite this apparent success, it would be undoubtedly useful to expand the current model to

incorporate additional real-side and nominal frictions such as variable capacity utilization

and sticky nominal wages, combined with wage and price indexation. Another useful direc-

tion of this line of research would be to enrich the financial sector by introducing frictions

24

Page 26: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

in the intermediation process that links household borrowing to house prices as in Aoki,

Proudman, and Vlieghe [2004] and Iacoviello [2005] and by explicitly modeling the financial

sector and the role that financial capital may have on the real economy. These extensions

of the basic framework are necessary both for empirical realism and to capture what are

undoubtedly the main concerns of policymakers today, namely the ongoing collapse of house

prices and its impact on the real economy through its effect on the value of assets held by

financial institutions.

References

Aguiar, M. (2005): “Investment, Devaluation, and Foreign Currency Exposure: A Caseof Mexico,” Journal of Development Economics, 78, 95–113.

Aoki, K., J. Proudman, and G. Vlieghe (2004): “House Prices, Consumption, andMonetary Policy: A Financial Accelerator Approach,” Journal of Financial Economics,13, 413–435.

Bernanke, B. S., M. Gertler, and S. Gilchrist (1999): “The Financial Acceleratorin a Quantitative Business Cycle Framework,” in The Handbook of Macroeconomics, ed.by J. B. Taylor, and M. Woodford, pp. 1341–1393. Elsevier Science B.V., Amsterdam.

Brunnermeier, M. K. (2008): “Deciphering the 2007–08 Liquidity and Credit Crunch,”Forthcoming, Journal of Economic Perspectives.

Buiter, W. H. (2008): “Central Banks and Financial Crises,” Paper presented at the Fed-eral Reserve Bank of Kansas City’s symposium on ”Maintaining Stability in a ChangingFinancial System”, at Jackson Hole, Wyoming.

Campbell, J. Y., and J. Cocco (2008): “How Do House Prices Affect Consumption?Evidence From Micro Data,” Forthcoming, Journal of Monetary Economics.

Carlstrom, C. T., and T. S. Fuerst (1997): “Agency Costs, Net Worth, and BusinessCycle Fluctuations: A Computable General Equilibrium Analysis,” American Economic

Review, 87, 893–910.

Carroll, C. D., M. Otsuka, and J. Slacalek (2006): “How Large is the HousingWealth Effect? A New Approach,” Mimeo, Dept. of Economics, Johns Hopkins Univer-sity.

Case, K. E., J. M. Quigley, and R. J. Shiller (2005): “Comparing Wealth Effects:The Stock Market versus the Housing Market,” The B.E. Journal of Macroeconomics, 5,Article 1.

Cetorelli, N., and L. S. Goldberg (2008): “Banking Globalization, Monetary Trans-mission and the Lending Channel,” FRB of New York Staff Report No. 333.

25

Page 27: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Christensen, I., and A. Dib (2008): “Monetary Policy in an Estimated DSGE Modelwith a Financial Accelerator,” Review of Economic Dynamics, 11, 155–178.

Christiano, L. J., R. Motto, and M. Rostagno (2007): “Shocks, Structures, orMonetary Policies? The Euro Area and U.S. After 2001,” NBER Working Paper No.13521.

Cooley, T. F., R. Marimon, and V. Quadrini (2004): “Aggregate Consequences ofLimited Contract Enforceability,” Journal of Political Economy, 112, 817–847.

Cummins, J. G., K. A. Hassett, and S. D. Oliner (2006): “Investment Behavior,Observable Expectations, and Internal Funds,” American Economic Review, 96, 796–810.

Fazzari, S. M., R. G. Hubbard, and B. C. Petersen (1988): “Financing Constraintsand Corporate Investment,” Brookings Papers on Economic Activity, 1, 141–195.

Friedman, B. M., and K. N. Kuttner (1998): “Indicator Properties of the Paper-Bill Spread: Lessons From Recent Experience,” Review of Economics and Statistics, 80,34–44.

Fuerst, T. S. (1995): “Money and Financial Interactions in the Business Cycle,” Journal

of Money, Credit, and Banking, 27, 1321–1338.

Gertler, M., and C. S. Lown (1999): “The Information in the High-Yield Bond Spreadfor the Business Cycle: Evidence and Some Implications,” Oxford Review of Economic

Policy, 15, 132–150.

Gilchrist, S., and C. P. Himmelberg (1995): “The Role of Cash Flow in Reduced-FormInvestment Equations,” Journal of Monetary Economics, 36, 541–572.

Gilchrist, S., A. Ortiz, and E. Zakrajsek (2008): “Bayesian Estimation of a DSGEModel with Financial Frictions,” Mimeo in progress.

Gilchrist, S., and J. W. Sim (2007): “Investment During the Korean Financial Crisis:A Structural Econometric Analysis,” NBER Working Paper No. 13315.

Gilchrist, S., V. Yankov, and E. Zakrajsek (2008): “Credit Market Shocks andEconomic Fluctuations: Evidence From Corporate Bond and Stock Markets,” Mimeo,Dept. of Economics, Boston University.

Gilchrist, S., and E. Zakrajsek (2007): “Investment and the Cost of Capital: NewEvidence from the Corporate Bond Market,” NBER Working Paper No. 13174.

Graeve, F. D. (2008): “The External Finance Premium and the Macroeconomy: U.S.Post-WWII Evidence,” FRB of Dallas Working Paper No. 0809.

Iacoviello, M. (2005): “House Prices, Borrowing Constraints, and Monetary Policy inthe Business Cycle,” American Economic Review, 95, 739–764.

26

Page 28: Linkages Between the Financial and Real Sectors: An Overviewpeople.bu.edu/sgilchri/BOG_Gilchrist_Zakrajsek_24sep2008.pdf · 2009-04-10 · Linkages Between the Financial and Real

Kashyap, A. K., and J. C. Stein (2000): “What Do a Million Observations on Banks SayAbout the Transmission of Monetary Policy?,” American Economic Review, 90, 407–428.

King, T. B., A. T. Levin, and R. Perli (2007): “Financial Market Perceptions ofRecession Risk,” Finance and Economics Discussion Series Paper No. 57, Federal ReserveBoard.

Kiyotaki, N., and J. H. Moore (1997): “Credit Cycles,” Journal of Political Economy,105, 211–248.

Levin, A. T., F. M. Natalucci, and E. Zakrajsek (2006): “The Magnitude andCyclical Behavior of Financial Market Frictions,” Mimeo, Federal Reserve Board.

Merton, R. C. (1974): “On the Pricing of Corporate Debt: The Risk Structure of InterestRates,” Journal of Finance, 29, 449–470.

Mueller, P. (2007): “Credit Spreads and Real Activity,” Mimeo, Columbia BusinessSchool.

Newey, W. K., and K. D. West (1987): “A Simple, Positive Semi-Definite, Het-eroskedasticity and Autocorrelation Consistent Covariance Matrix,” Econometrica, 55,703–708.

Philippon, T. (2008): “The Bond Market’s Q,” Forthcoming, Quarterly Journal of Eco-

nomics.

Queijo von Heideken, V. (2008): “How Important are Financial frictions in the U.S.and the Euro Area?,” Sveriges Riksbank Working Paper No. 223.

Rebelo, S., J. C. Eberly, and N. Vincent (2008): “Investment and Value: A Neoclas-sical Benchmark,” Mimeo, Kellogg School of Management, Northwestern University.

Stock, J. H., and M. W. Watson (1989): “New Indexes of Coincident and LeadingEconomic Indicators,” in NBER Macroeconomics Annual, ed. by O. J. Blanchard, and

S. Fischer, pp. 351–394. The MIT Press, Cambridge.

Van den Heuvel, S. J. (2007): “The Bank Capital Channel of Monetary Policy,” Mimeo,The Wharton School, University of Pennsylvania.

27