Michael Woodford 2010 Financial Intermediation and Macroeconomic Analysis

Journal of Economic Perspectives—Volume 24, Number 4—Fall 2010—Pages 21–44

II ssues relating to fi nancial stability have always been part of the macroeco-ssues relating to fi nancial stability have always been part of the macroeco-nomics curriculum, but they have often been presented as mainly of historical nomics curriculum, but they have often been presented as mainly of historical interest, or primarily of relevance to emerging markets. However, the recent interest, or primarily of relevance to emerging markets. However, the recent

fi nancial crisis has made it plain that even in economies like the United States, fi nancial crisis has made it plain that even in economies like the United States, signifi cant disruptions of fi nancial intermediation remain a possibility. Under-signifi cant disruptions of fi nancial intermediation remain a possibility. Under-standing such phenomena and the possible policy responses requires the use of standing such phenomena and the possible policy responses requires the use of a macroeconomic framework in which fi nancial intermediation matters for the a macroeconomic framework in which fi nancial intermediation matters for the allocation of resources.allocation of resources.

In this paper, I fi rst discuss why neither standard macroeconomic models that In this paper, I fi rst discuss why neither standard macroeconomic models that abstract from fi nancial intermediation nor traditional models of the “bank lending abstract from fi nancial intermediation nor traditional models of the “bank lending channel” are adequate as a basis for understanding the recent crisis. I argue that channel” are adequate as a basis for understanding the recent crisis. I argue that instead we need models in which intermediation plays a crucial role, but in which instead we need models in which intermediation plays a crucial role, but in which intermediation is modeled in a way that better conforms to current institutional intermediation is modeled in a way that better conforms to current institutional realities. In particular, we need models that recognize that a market-based fi nan-realities. In particular, we need models that recognize that a market-based fi nan-cial system—one in which intermediaries fund themselves by selling securities in cial system—one in which intermediaries fund themselves by selling securities in competitive markets, rather than collecting deposits subject to reserve require-competitive markets, rather than collecting deposits subject to reserve require-ments—is not the same as a frictionless system.ments—is not the same as a frictionless system.

I then sketch the basic elements of an approach that allows fi nancial inter-I then sketch the basic elements of an approach that allows fi nancial inter-mediation and credit frictions to be integrated into macroeconomic analysis in a mediation and credit frictions to be integrated into macroeconomic analysis in a straightforward way. I show how the model can be used to analyze the macroeco-straightforward way. I show how the model can be used to analyze the macroeco-nomic consequences of the recent fi nancial crisis and conclude with a discussion of nomic consequences of the recent fi nancial crisis and conclude with a discussion of some implications of the model for the conduct of monetary policy.some implications of the model for the conduct of monetary policy.

Financial Intermediation and Macroeconomic Analysis

■ ■ Michael Woodford is the John Bates Clark Professor of Political Economy, Columbia Univer-Michael Woodford is the John Bates Clark Professor of Political Economy, Columbia Univer-sity, New York City, New York. His e-mail address is sity, New York City, New York. His e-mail address is ⟨⟨[email protected]@columbia.edu⟩⟩..

doi=10.1257/jep.24.4.21

Michael Woodford

22 Journal of Economic Perspectives

Why a New Framework for Macroeconomic Analysis is NeededWhy a New Framework for Macroeconomic Analysis is Needed

It may be useful fi rst to review why familiar macroeconomic models do not It may be useful fi rst to review why familiar macroeconomic models do not already incorporate the features needed to make sense of recent economic devel-already incorporate the features needed to make sense of recent economic devel-opments. I shall argue that it is diffi cult to understand why either the signifi cant opments. I shall argue that it is diffi cult to understand why either the signifi cant decline in house prices since 2006 or the substantial losses sustained by fi nancial decline in house prices since 2006 or the substantial losses sustained by fi nancial fi rms should have so seriously affected aggregate employment and economic fi rms should have so seriously affected aggregate employment and economic activity except in the context of a model in which fi nancial intermediaries play a activity except in the context of a model in which fi nancial intermediaries play a crucial role and in which their ability to fulfi ll that function can at some times be crucial role and in which their ability to fulfi ll that function can at some times be signifi cantly impaired.signifi cantly impaired.

Housing Prices and Aggregate DemandHousing Prices and Aggregate DemandWhile the severity of the recent fi nancial crisis has been extensively discussed, While the severity of the recent fi nancial crisis has been extensively discussed,

some have questioned whether it was really the primary cause of the Great Recession. some have questioned whether it was really the primary cause of the Great Recession. For example, Baker (2010) argues that a substantial reduction in aggregate demand For example, Baker (2010) argues that a substantial reduction in aggregate demand can be explained as a wealth effect on consumer expenditure, given the decline can be explained as a wealth effect on consumer expenditure, given the decline in U.S. households’ housing wealth by several trillion dollars. In this analysis, “the in U.S. households’ housing wealth by several trillion dollars. In this analysis, “the problem is not fi rst and foremost a fi nancial crisis.” But as Buiter (2010) points out, problem is not fi rst and foremost a fi nancial crisis.” But as Buiter (2010) points out, there is no aggregate wealth effect of a decline in housing prices, since the house-there is no aggregate wealth effect of a decline in housing prices, since the house-hold sector in aggregate is both the owner of the housing stock and the consumer hold sector in aggregate is both the owner of the housing stock and the consumer of the services supplied by it. A fall in house prices reduces the value of an asset, of the services supplied by it. A fall in house prices reduces the value of an asset, but also reduces the cost of buying the stream of housing services that people were but also reduces the cost of buying the stream of housing services that people were planning to purchase by exactly the same amount.planning to purchase by exactly the same amount.

It is possible to have a nonzero effect on aggregate expenditure on other goods It is possible to have a nonzero effect on aggregate expenditure on other goods (when other prices remain unchanged), even without fi nancial frictions, owing to (when other prices remain unchanged), even without fi nancial frictions, owing to redistribution of wealth between households with a net “long” position in housing redistribution of wealth between households with a net “long” position in housing and those with net “short” positions, if the average marginal propensities to consume and those with net “short” positions, if the average marginal propensities to consume out of wealth are different between the two types. Nonetheless, because the positive out of wealth are different between the two types. Nonetheless, because the positive and negative wealth effects will largely offset one another, the effect on aggregate and negative wealth effects will largely offset one another, the effect on aggregate demand is likely to be fairly small relative to the size of the aggregate decrease in demand is likely to be fairly small relative to the size of the aggregate decrease in housing wealth.housing wealth.

Larger effects are instead possible if one recognizes that the losses resulting Larger effects are instead possible if one recognizes that the losses resulting from the collapse of housing prices were disproportionately concentrated in certain from the collapse of housing prices were disproportionately concentrated in certain fi nancial institutions which play a role in the allocation of resources that cannot fi nancial institutions which play a role in the allocation of resources that cannot easily be replaced by those to whom wealth was redistributed. A model of this kind easily be replaced by those to whom wealth was redistributed. A model of this kind is sketched below. For a quantitative analysis of the effects of the fall in U.S. housing is sketched below. For a quantitative analysis of the effects of the fall in U.S. housing prices that stresses such effects, see Greenlaw, Hatzius, Kashyap, and Shin (2008).prices that stresses such effects, see Greenlaw, Hatzius, Kashyap, and Shin (2008).

Banking and the Money SupplyBanking and the Money SupplyIt is also diffi cult to understand why large losses by fi nancial institutions on It is also diffi cult to understand why large losses by fi nancial institutions on

housing-related bets should have such a signifi cant effect on the real economy housing-related bets should have such a signifi cant effect on the real economy without a model that takes account of credit frictions. According to the well-known without a model that takes account of credit frictions. According to the well-known monetarist view, banking crises affect the economy because they reduce the total monetarist view, banking crises affect the economy because they reduce the total supply of money in the economy, since the “money multiplier”—the factor by which supply of money in the economy, since the “money multiplier”—the factor by which

Michael Woodford 23

the economy’s money supply exceeds the “monetary base” supplied by the central the economy’s money supply exceeds the “monetary base” supplied by the central bank—falls when funds are withdrawn from commercial banks in response to bank—falls when funds are withdrawn from commercial banks in response to concerns about their stability. The lower money supply is then only consistent with concerns about their stability. The lower money supply is then only consistent with money demand to the extent that money demand is also reduced, through some money demand to the extent that money demand is also reduced, through some combination of lower economic activity and defl ation. This is the classic account by combination of lower economic activity and defl ation. This is the classic account by Friedman and Schwartz (1963) of how the widespread bank failures in the United Friedman and Schwartz (1963) of how the widespread bank failures in the United States deepened the Great Depression.States deepened the Great Depression.

However, such a model, at least as conventionally elaborated, cannot explain However, such a model, at least as conventionally elaborated, cannot explain why the recent problems of the fi nancial sector should have caused a sharp reces-why the recent problems of the fi nancial sector should have caused a sharp reces-sion, for the Friedman–Schwartz story depends on the monetary base remaining sion, for the Friedman–Schwartz story depends on the monetary base remaining fi xed despite a collapse of the money multiplier. But under contemporary insti-fi xed despite a collapse of the money multiplier. But under contemporary insti-tutional arrangements, the Fed automatically adjusts the supply of base money as tutional arrangements, the Fed automatically adjusts the supply of base money as necessary to maintain its target for the federal funds interest rate; thus, any change necessary to maintain its target for the federal funds interest rate; thus, any change in the money multiplier due to a banking crisis should automatically be offset by a in the money multiplier due to a banking crisis should automatically be offset by a corresponding increase in the monetary base, neutralizing any effect on interest corresponding increase in the monetary base, neutralizing any effect on interest rates, infl ation, or output.rates, infl ation, or output.11

Moreover, many of the institutions whose failure or near-failure appeared to Moreover, many of the institutions whose failure or near-failure appeared to do the most damage in the recent crisis, such as Lehman Brothers, did not issue do the most damage in the recent crisis, such as Lehman Brothers, did not issue liabilities that would count as part of Friedman and Schwartz’s measure of the liabilities that would count as part of Friedman and Schwartz’s measure of the money supply. Under a classic monetarist view, the failure of such institutions money supply. Under a classic monetarist view, the failure of such institutions should pose no threat to the aggregate economy. (Hence the proposals by some should pose no threat to the aggregate economy. (Hence the proposals by some that fi nance can remain only lightly regulated, as long as commercial banks are that fi nance can remain only lightly regulated, as long as commercial banks are strictly excluded from the riskier activities.) But the consequences of the failure strictly excluded from the riskier activities.) But the consequences of the failure of Lehman suggest otherwise.of Lehman suggest otherwise.

Models of the Bank Lending ChannelModels of the Bank Lending ChannelModels that postulate an essential role for banks in fi nancing certain kinds of Models that postulate an essential role for banks in fi nancing certain kinds of

expenditure are better able to explain how a fi nancial crisis could have such dire conse-expenditure are better able to explain how a fi nancial crisis could have such dire conse-quences for the real economy as we have observed. However, the kinds of fi nancial quences for the real economy as we have observed. However, the kinds of fi nancial constraints that were emphasized in many past models of this kind assumed specifi c constraints that were emphasized in many past models of this kind assumed specifi c institutional forms and regulatory requirements that have become less relevant to the institutional forms and regulatory requirements that have become less relevant to the U.S. fi nancial system over time.U.S. fi nancial system over time.

Consider, for example, traditional accounts of the “bank lending channel” of Consider, for example, traditional accounts of the “bank lending channel” of the transmission of monetary policy. This argument emphasized the indispensable the transmission of monetary policy. This argument emphasized the indispensable role of commercial banks as sources of credit for certain kinds of borrowers, in role of commercial banks as sources of credit for certain kinds of borrowers, in particular those without direct access to capital markets. Deposits were in turn held particular those without direct access to capital markets. Deposits were in turn held to be an indispensable source of funding for the lending of commercial banks, and to be an indispensable source of funding for the lending of commercial banks, and these were subject to legal reserve requirements. To the extent that reserve require-these were subject to legal reserve requirements. To the extent that reserve require-ments were typically a binding constraint, a reduction in the supply of reserves by ments were typically a binding constraint, a reduction in the supply of reserves by the Federal Reserve would require the volume of deposits to be reduced, which the Federal Reserve would require the volume of deposits to be reduced, which

1 For example, in the model without credit frictions expounded in Figure 2 below, a banking panic that reduces the money multiplier will have no effect other than to increase the supply of base money required to implement the central bank reaction function represented by the schedule MP.


would in turn require less lending by commercial banks. Bernanke and Blinder would in turn require less lending by commercial banks. Bernanke and Blinder (1988) and Kashyap and Stein (1994) offer expositions of this view; Smant (2002) (1988) and Kashyap and Stein (1994) offer expositions of this view; Smant (2002) provides a critical review of the literature.provides a critical review of the literature.

Clearly, the importance of this channel for effects of monetary policy on Clearly, the importance of this channel for effects of monetary policy on economic activity depended on the validity of each of the links in the proposed economic activity depended on the validity of each of the links in the proposed mechanism: that reserve requirements were a binding constraint for many banks; mechanism: that reserve requirements were a binding constraint for many banks; that commercial banks lacked sources of funding other than deposits; that an that commercial banks lacked sources of funding other than deposits; that an important subset of borrowers lacked sources of credit other than commercial important subset of borrowers lacked sources of credit other than commercial banks; and that banks lacked opportunities to substitute between other assets and banks; and that banks lacked opportunities to substitute between other assets and lending to bank-dependent borrowers. Each of these assumptions was less obviously lending to bank-dependent borrowers. Each of these assumptions was less obviously defensible after the fi nancial innovations and regulatory changes of the 1980s and defensible after the fi nancial innovations and regulatory changes of the 1980s and 1990s. Adrian and Shin (forthcoming a, b) discuss the changing structure of the 1990s. Adrian and Shin (forthcoming a, b) discuss the changing structure of the U.S. fi nancial system in more detail.U.S. fi nancial system in more detail.

Nonbank fi nancial intermediaries became increasingly important as sources of Nonbank fi nancial intermediaries became increasingly important as sources of credit, particularly as a result of the growing popularity of securitization. Figure 1A credit, particularly as a result of the growing popularity of securitization. Figure 1A shows the contributions of several categories of fi nancial institutions to total net shows the contributions of several categories of fi nancial institutions to total net lending in the United States; while commercial banks are clearly still important, lending in the United States; while commercial banks are clearly still important, they are far from the only important source of credit. More importantly, both the they are far from the only important source of credit. More importantly, both the recent lending boom and the more recent fi nancial crisis had more to do with recent lending boom and the more recent fi nancial crisis had more to do with changes in fi nancial fl ows of several of the other types shown in the fi gure; for changes in fi nancial fl ows of several of the other types shown in the fi gure; for example, lending by issuers of asset-backed securities surged in the period up until example, lending by issuers of asset-backed securities surged in the period up until the summer of 2007 and then crashed, while lending by market-based mutual funds the summer of 2007 and then crashed, while lending by market-based mutual funds and other market-based fi nancial intermediariesand other market-based fi nancial intermediaries22 crashed after the fall of 2008. crashed after the fall of 2008.

Nor are deposits the main source of funding for the fi nancial sector, even in Nor are deposits the main source of funding for the fi nancial sector, even in the case of commercial banks. Figure 1B shows the net increase in fi nancial sector the case of commercial banks. Figure 1B shows the net increase in fi nancial sector liabilities each quarter from several sources. Checkable deposits are only a small liabilities each quarter from several sources. Checkable deposits are only a small part of the sector’s fi nancing; moreover, deposits shrank during the years of the part of the sector’s fi nancing; moreover, deposits shrank during the years of the lending boom, but have risen again during the crisis—so that neither the growth in lending boom, but have risen again during the crisis—so that neither the growth in credit during the boom nor the contraction of credit in 2008–09 can be attributed credit during the boom nor the contraction of credit in 2008–09 can be attributed to variations in the availability of deposits as a source of fi nancing. Even to the to variations in the availability of deposits as a source of fi nancing. Even to the extent that deposits do matter, one may doubt the extent to which the availability of extent that deposits do matter, one may doubt the extent to which the availability of such funding is constrained by reserve requirements, as in recent years these have such funding is constrained by reserve requirements, as in recent years these have ceased to be a binding constraint for many banks (for example, see Bennett and ceased to be a binding constraint for many banks (for example, see Bennett and Peristiani, 2002).Peristiani, 2002).

In response to skepticism about the relevance of the traditional bank lending In response to skepticism about the relevance of the traditional bank lending channel, Bernanke and Gertler (1995) have instead stressed the importance of channel, Bernanke and Gertler (1995) have instead stressed the importance of an alternative “broad credit channel,” in which the balance sheets of ultimate an alternative “broad credit channel,” in which the balance sheets of ultimate borrowers constrain the amount that they are able to borrow; models incorpo-borrowers constrain the amount that they are able to borrow; models incorpo-rating such effects include those of Kiyotaki and Moore (1997) and Bernanke, rating such effects include those of Kiyotaki and Moore (1997) and Bernanke, Gertler, and Gilchrist (1999). However, the recent crisis, at least in its initial phase, Gertler, and Gilchrist (1999). However, the recent crisis, at least in its initial phase,

2 This category includes mutual funds, the government-sponsored enterprises (GSEs), GSE-backed mort-gage pools, fi nance companies, real-estate investment trusts, broker-dealers, and funding corporations. The “market-based fi nancial intermediaries” terminology derives from Adrian and Shin (forthcoming a).

Financial Intermediation and Macroeconomic Analysis 25

Figure 1Financial Flows over the Most Recent Credit Cycle

Source: Federal Reserve Board, Flow of Funds Accounts.

A: Contributions to U.S. Total Net Lending from Several Categories of FinancialInstitutions(quarterly, in billions of dollars)

$ bi

llion

s

2,000

1,500

1,000

500

0

–500

–1,000

–1,5002003-Q1

2003-Q4

2004-Q3

2005-Q2

2006-Q1

2006-Q4

2007-Q3

2008-Q2

2009-Q1

2009-Q4

Commercial bankingMoney market mutual fundsAsset-backed securities issuersOther market-based financial institutions

B: Contributions from Several Sources of Funding to the Net Increase in the Liabilities of the U.S. Financial Sector(quarterly, in billions of dollars)

2003-Q1

2003-Q4

2004-Q3

2005-Q2

2006-Q1

2006-Q4

2007-Q3

2008-Q2

2009-Q1

2009-Q4

$ bi

llion

s

2,000

1,500

1,000

500

0

–500

–1,000

–1,500

Checkable depositsMoney market mutual fund sharesFed funds and reposCommercial paper


resulted more from obstacles to credit supply, resulting from developments in the resulted more from obstacles to credit supply, resulting from developments in the fi nancial sector itself, than from a reduction in credit demand owing to the prob-fi nancial sector itself, than from a reduction in credit demand owing to the prob-lems of ultimate borrowers.lems of ultimate borrowers.

Hence, what is needed is a framework for macroeconomic analysis in which Hence, what is needed is a framework for macroeconomic analysis in which intermediation plays a crucial role and in which frictions that can impede an intermediation plays a crucial role and in which frictions that can impede an effi cient supply of credit are allowed for, a framework which also takes account of effi cient supply of credit are allowed for, a framework which also takes account of the fact that the U.S. fi nancial sector is now largely market-based. Fortunately, the the fact that the U.S. fi nancial sector is now largely market-based. Fortunately, the development of a new generation of macroeconomic models with these features development of a new generation of macroeconomic models with these features is now well underway. Adrian and Shin (forthcoming b) and Gertler and Kiyotaki is now well underway. Adrian and Shin (forthcoming b) and Gertler and Kiyotaki (forthcoming) provide surveys of recent work in this area. Here, I sketch a basic (forthcoming) provide surveys of recent work in this area. Here, I sketch a basic version of such a model, show how it can be used to interpret the recent crisis, version of such a model, show how it can be used to interpret the recent crisis, and then discuss some implications of a model of this kind for monetary policy. A and then discuss some implications of a model of this kind for monetary policy. A complete monetary dynamic stochastic general equilibrium model based on the complete monetary dynamic stochastic general equilibrium model based on the approach sketched here is developed in Cúrdia and Woodford (2009).approach sketched here is developed in Cúrdia and Woodford (2009).

Credit and Economic Activity: A Market-Based ApproachCredit and Economic Activity: A Market-Based Approach

The theory sketched here is appropriate to a market-based fi nancial system in The theory sketched here is appropriate to a market-based fi nancial system in which the most important marginal suppliers of credit are no longer commercial which the most important marginal suppliers of credit are no longer commercial banks and in which deposits subject to reserve requirements are no longer the most banks and in which deposits subject to reserve requirements are no longer the most important marginal source of funding even for commercial banks.important marginal source of funding even for commercial banks.

Macroeconomics with a Single Interest RateMacroeconomics with a Single Interest RateIt is useful to begin by recalling how interest-rate policy affects aggregate It is useful to begin by recalling how interest-rate policy affects aggregate

activity in a conventional model that abstracts from fi nancial frictions. In the activity in a conventional model that abstracts from fi nancial frictions. In the simplest versions of such models, fi nancial conditions can be summarized by a single simplest versions of such models, fi nancial conditions can be summarized by a single interest rate, the equilibrium value of which is determined in a market for credit. interest rate, the equilibrium value of which is determined in a market for credit. Figure 2A shows the key equilibrium condition. The loan supply schedule Figure 2A shows the key equilibrium condition. The loan supply schedule LS shows shows the amount of lending the amount of lending L that ultimate savers are willing to fi nance (by refraining that ultimate savers are willing to fi nance (by refraining from expenditure themselves) for each possible value of the interest rate from expenditure themselves) for each possible value of the interest rate i received received by savers, while the loan demand schedule by savers, while the loan demand schedule LD shows the demand for such funds shows the demand for such funds for each possible value of the interest rate that must be paid by borrowers. Note for each possible value of the interest rate that must be paid by borrowers. Note that the slopes for the curves that the slopes for the curves LS and and LD both refl ect the same principle, which is both refl ect the same principle, which is that a higher interest rate gives both savers and borrowers a reason to defer current that a higher interest rate gives both savers and borrowers a reason to defer current spending to a greater extent. Equilibrium in the credit market then determines spending to a greater extent. Equilibrium in the credit market then determines both a market-clearing interest rate and an equilibrium volume of lending, as shown both a market-clearing interest rate and an equilibrium volume of lending, as shown by by i11 and and L11 in the fi gure. in the fi gure.

In Figure 2A, the loan supply and demand curves are specifi ed while holding In Figure 2A, the loan supply and demand curves are specifi ed while holding constant a great many variables other than the current interest rate. In particular, constant a great many variables other than the current interest rate. In particular, the curves are shown assuming a particular level of current-period aggregate output the curves are shown assuming a particular level of current-period aggregate output (and hence income) (and hence income) Y. A higher level of income should increase the supply of loans . A higher level of income should increase the supply of loans at any given interest rate (as not all of the additional income should be consumed at any given interest rate (as not all of the additional income should be consumed if future income expectations are held fi xed); hence an increase in if future income expectations are held fi xed); hence an increase in Y should shift should shift

Michael Woodford 27

Figure 2Interest-Rate and Output Determination in the Standard Model

Notes: In panel A, LS is the loan supply schedule and LD is the loan demand schedule, which are specifi ed holding constant aggregate income, Y. The arrows show how the curves shift with an increase in Y.Panel B shows an IS schedule, derived by tracing out the equilibrium interest rate for any assumed level of current income Y, and a monetary policy reaction function (MP ), showing how the central bank’s interest rate target will vary with the level of economic activity. The MP curve is drawn for a given infl ation rate. The arrow shows the consequence of an exogenous shift in the policy reaction function that implies a lower interest rate for any given level of economic activity.

A: Effect of an Increase in Aggregate Income on Loan Supply and Demand

B: Effect of a Loosening of Monetary Policy on Interest Rates and Output

i1

i2

L1 L2 L

LD

LSi

i1

i2

i MP

IS

Y1 Y2 Y

Inte

rest

rat

eIn

tere

st r

ate

Volume of lending

Aggregate income


the the LS curve down and to the right, as shown by the arrow. It should also reduce curve down and to the right, as shown by the arrow. It should also reduce the demand for loans, insofar as borrowers have more current income available the demand for loans, insofar as borrowers have more current income available out of which to fi nance current spending needs or opportunities, in which case the out of which to fi nance current spending needs or opportunities, in which case the LD curve shifts down and to the left, as also shown in the fi gure. The vertical shift in curve shifts down and to the left, as also shown in the fi gure. The vertical shift in the the LD curve is likely to be smaller than the vertical shift of the curve is likely to be smaller than the vertical shift of the LS curve, as shown in curve, as shown in Figure 2A, if the expenditure of borrowers is more interest-elastic than the expendi-Figure 2A, if the expenditure of borrowers is more interest-elastic than the expendi-ture of savers. The intersection of the grey curves shows the new equilibrium values, ture of savers. The intersection of the grey curves shows the new equilibrium values, i22 and and L22..

Tracing out the equilibrium interest rate for any assumed level of current Tracing out the equilibrium interest rate for any assumed level of current income income Y, one obtains the , one obtains the IS schedule plotted in Figure 2B. (Alternatively, for schedule plotted in Figure 2B. (Alternatively, for each possible interest rate each possible interest rate i, the schedule shows the level of national income for , the schedule shows the level of national income for which investment equals savings, as this is equivalent to equality between supply which investment equals savings, as this is equivalent to equality between supply of and demand for funds.) The monetary policy reaction function of the central of and demand for funds.) The monetary policy reaction function of the central bank, indicating how the central bank’s interest-rate target will vary with the level of bank, indicating how the central bank’s interest-rate target will vary with the level of economic activity, is shown by the curve economic activity, is shown by the curve MP in this fi gure. in this fi gure.33

If we suppose that the If we suppose that the MP curve is drawn for a given infl ation rate, then the curve is drawn for a given infl ation rate, then the upward slope shown indicates a response of interest rates to the level of output (rela-upward slope shown indicates a response of interest rates to the level of output (rela-tive to trend or to potential), of a kind implied, for example, by the “Taylor rule” tive to trend or to potential), of a kind implied, for example, by the “Taylor rule” (Taylor, 1993)—that is, higher interest rates when output is high relative to trend or (Taylor, 1993)—that is, higher interest rates when output is high relative to trend or potential, and lower interest rates when output is low relative to trend or potential. potential, and lower interest rates when output is low relative to trend or potential. In this case, the equilibrium level of output determined in Figure 2B depends on In this case, the equilibrium level of output determined in Figure 2B depends on the infl ation rate; a graph showing how the equilibrium level of output would vary the infl ation rate; a graph showing how the equilibrium level of output would vary with infl ation yields an aggregate demand relation in infl ation–output space. Plot-with infl ation yields an aggregate demand relation in infl ation–output space. Plot-ting that relation along with a Phillips curve (or aggregate supply) relation between ting that relation along with a Phillips curve (or aggregate supply) relation between infl ation and output, one can then fi nally determine equilibrium output.infl ation and output, one can then fi nally determine equilibrium output.44

This kind of model provides a straightforward account of the way in which a This kind of model provides a straightforward account of the way in which a central bank’s interest-rate policy affects the level of economic activity (and also the central bank’s interest-rate policy affects the level of economic activity (and also the infl ation rate, once one adjoins a Phillips curve to the model). However, this model infl ation rate, once one adjoins a Phillips curve to the model). However, this model of the credit market—in which ultimate savers lend directly to ultimate borrowers so of the credit market—in which ultimate savers lend directly to ultimate borrowers so that the interest rate received by savers is the same as that paid by borrowers—clearly that the interest rate received by savers is the same as that paid by borrowers—clearly omits some important features of actual fi nancial systems. In actual economies, we omits some important features of actual fi nancial systems. In actual economies, we observe multiple interest rates that do not move perfectly together. Changes in observe multiple interest rates that do not move perfectly together. Changes in spreads between certain of these interest rates have been important indicators of spreads between certain of these interest rates have been important indicators of changing fi nancial conditions, both during the recent housing boom and during changing fi nancial conditions, both during the recent housing boom and during the subsequent crash, as is discussed further below.the subsequent crash, as is discussed further below.

3 In the case that monetary policy is assumed to correspond to some fi xed supply of money, then the MP curve becomes simply the Hicksian LM curve. However, an upward-sloping relation of the kind shown in the fi gure will exist under many other hypotheses, including ones more descriptive of actual central bank behavior than the Hicksian construct. On the relation between IS–MP analysis and the older IS–LM analysis, see, for example, Romer (2000) in this journal.4 Alternatively, one can substitute the infl ation rate implied by the Phillips curve (for a given level of output) into the central bank reaction function, and plot the resulting relation for i as a function of Y as the curve MP. In this case, MP slopes upward, as shown, even if the central bank’s reaction function responds only to infl ation; and the equilibrium shown in Figure 2B already takes account of the endo-geneity of the infl ation rate.


Introducing Multiple Interest RatesIntroducing Multiple Interest RatesHere I illustrate one way to introduce multiple interest rates into this model. Here I illustrate one way to introduce multiple interest rates into this model.

Suppose that instead of directly lending to ultimate borrowers themselves, savers Suppose that instead of directly lending to ultimate borrowers themselves, savers fund intermediaries, who use these funds to lend to (or acquire fi nancial claims fund intermediaries, who use these funds to lend to (or acquire fi nancial claims on) the ultimate borrowers. Then, it becomes necessary to distinguish between the on) the ultimate borrowers. Then, it becomes necessary to distinguish between the interest rate interest rate i ss (the rate paid to savers) at which intermediaries are able to fund (the rate paid to savers) at which intermediaries are able to fund themselves and the interest rate themselves and the interest rate i bb (the borrowing or loan rate) at which ultimate (the borrowing or loan rate) at which ultimate borrowers are able to fi nance additional current expenditure. We can still think in borrowers are able to fi nance additional current expenditure. We can still think in terms of the two schedules shown in Figure 2A, but now the terms of the two schedules shown in Figure 2A, but now the LS schedule represents schedule represents the supply of funding for intermediaries rather than the supply of loans to ultimate the supply of funding for intermediaries rather than the supply of loans to ultimate borrowers, and we must now recognize that the supply of funding and the demand borrowers, and we must now recognize that the supply of funding and the demand for loans are functions of two different interest rates. Hence the equilibrium level for loans are functions of two different interest rates. Hence the equilibrium level of lending of lending L can be at a point other than the one where the two schedules cross, as can be at a point other than the one where the two schedules cross, as shown in Figure 3A.shown in Figure 3A.

What determines the equilibrium relation between the two interest rates What determines the equilibrium relation between the two interest rates i ss and and i bb ? Given the funding supply and loan demand curves (which means, given the ? Given the funding supply and loan demand curves (which means, given the values of a set of variables that include the current value of income values of a set of variables that include the current value of income Y ), we can deter-), we can deter-mine the unique volume of intermediation that is consistent with any given spread mine the unique volume of intermediation that is consistent with any given spread ωω between between i bb and and i ss. If the funding supply curve . If the funding supply curve LS and the loan demand curve and the loan demand curve LD have the slopes shown, then a larger credit spread have the slopes shown, then a larger credit spread ωω implies a lower equilibrium implies a lower equilibrium volume of intermediated credit volume of intermediated credit L. This relation between the quantity of intermedi-. This relation between the quantity of intermedi-ated credit and the credit spread is graphed as the curve ated credit and the credit spread is graphed as the curve XD in Figure 3B, which we in Figure 3B, which we can think of as the “demand for intermediation.”can think of as the “demand for intermediation.”

The demand for intermediation schedule The demand for intermediation schedule XD indicates the degree to which indicates the degree to which borrowers are willing to pay an interest rate higher than the one required in order borrowers are willing to pay an interest rate higher than the one required in order to induce savers to supply funds to fi nance someone else’s expenditure. This repre-to induce savers to supply funds to fi nance someone else’s expenditure. This repre-sents a profi t opportunity for intermediaries, to the extent that they are able to sents a profi t opportunity for intermediaries, to the extent that they are able to arrange for the transfer of funds at suffi ciently low cost. The volume of lending that arrange for the transfer of funds at suffi ciently low cost. The volume of lending that actually occurs, though, will also depend on the capacity of the fi nancial sector to actually occurs, though, will also depend on the capacity of the fi nancial sector to supply this service at a margin low enough for the services to be demanded.supply this service at a margin low enough for the services to be demanded.

The corresponding “supply of intermediation” schedule, indicating the credit The corresponding “supply of intermediation” schedule, indicating the credit spread required to induce fi nancial institutions to intermediate a certain volume spread required to induce fi nancial institutions to intermediate a certain volume of credit between savers and ultimate borrowers, is depicted by the curve of credit between savers and ultimate borrowers, is depicted by the curve XS in in Figure 3B. This curve refl ects the consequences of profi t maximization by inter-Figure 3B. This curve refl ects the consequences of profi t maximization by inter-mediaries, where the intermediaries in question need not be understood to consist mediaries, where the intermediaries in question need not be understood to consist solely or even primarily of traditional commercial banks. Both the equilibrium credit solely or even primarily of traditional commercial banks. Both the equilibrium credit spread and the equilibrium volume of credit are then determined by the intersec-spread and the equilibrium volume of credit are then determined by the intersec-tion between the tion between the XS and and XD schedules. And given an equilibrium credit spread schedules. And given an equilibrium credit spread ωω, , determined in Figure 3A, one can use Figure 3A to determine the two interest rates.determined in Figure 3A, one can use Figure 3A to determine the two interest rates.

Determinants of the Supply of IntermediationDeterminants of the Supply of IntermediationThe structural relationship represented by the supply of intermediation The structural relationship represented by the supply of intermediation

schedule schedule XS in Figure 3B can be motivated in various ways. One model assumes in Figure 3B can be motivated in various ways. One model assumes that intermediaries have costs of originating and servicing loans, or of managing that intermediaries have costs of originating and servicing loans, or of managing


Notes: i s is the interest rate paid to savers, at which intermediaries are able to fund themselves, and i b is the interest rate (the borrowing or loan rate) at which ultimate borrowers are able to fi nance additional current expenditure. In this fi gure, LS schedule represents the supply of funding for intermediaries, the LD schedule is the loan demand schedule, and these schedules are functions of two different interest rates. Hence the equilibrium level of lending Lcan be at a point other than the one where the two schedules cross, as shown in Figure 3(a). ω is the spread between i b and i s. Given the LS and LD curves we can determine the unique volume of intermediation that is consistent with any given spread ω. This relation between the quantity of intermediated credit and the credit spread is graphed as the curve XD in panel B, which we can think of as the “demand for intermediation.” The corresponding “supply of intermediation” schedule XS indicates the credit spread required to induce fi nancial institutions to intermediate a certain volume of credit between savers and ultimate borrowers.

B: Determination of the Equilibrium Credit Spread

XS

XD

ω1

ω

Inte

rest

rat

e sp

read

(b

etw

een

sav

ers

and

borr

ower

s)

L1 LVolume of lending

Figure 3Credit Market Equilibrium with Credit Supply Frictions

A: Effect of a Credit Spread ω1 on the Equilibrium Interest Rates for Borrowers and Savers, and on the Equilibrium Volume of Credit

ib

i s

i

Inte

rest

rat

e

LS

LD


ω1e

Michael Woodford 31

their portfolios, so that in a competitive equilibrium, the rate their portfolios, so that in a competitive equilibrium, the rate i bb at which they are at which they are willing to lend (or the return that they will require on assets that they purchase) will willing to lend (or the return that they will require on assets that they purchase) will exceed their cost of funds exceed their cost of funds i ss by a spread that refl ects the marginal cost of lending. by a spread that refl ects the marginal cost of lending. This marginal cost may be increasing in the volume of lending by the intermediary This marginal cost may be increasing in the volume of lending by the intermediary if the production function for loans involves diminishing returns to increases in the if the production function for loans involves diminishing returns to increases in the variable factors, owing to the fi xity of some factors (such as specialized expertise or variable factors, owing to the fi xity of some factors (such as specialized expertise or facilities that cannot be expanded quickly).facilities that cannot be expanded quickly).55

Probably a more important determinant of the supply of intermediation derives Probably a more important determinant of the supply of intermediation derives from the limited capital of intermediaries—or, more fundamentally, the limited from the limited capital of intermediaries—or, more fundamentally, the limited capital of the “natural buyers” of the debt of the ultimate borrowers—together with capital of the “natural buyers” of the debt of the ultimate borrowers—together with limits on the degree to which these natural buyers are able to leverage their posi-limits on the degree to which these natural buyers are able to leverage their posi-tions. The market for the debt of the ultimate borrowers may be limited to a narrow tions. The market for the debt of the ultimate borrowers may be limited to a narrow class of “natural buyers” for any of a variety of reasons: special expertise may be class of “natural buyers” for any of a variety of reasons: special expertise may be required to evaluate such assets; other costs of market participation may be lower required to evaluate such assets; other costs of market participation may be lower for certain investors; or the natural buyers may be less risk averse, or less uncertainty for certain investors; or the natural buyers may be less risk averse, or less uncertainty averse, or more optimistic about returns on the particular assets.averse, or more optimistic about returns on the particular assets.

Leverage may also be constrained for any of a variety of reasons. The recent Leverage may also be constrained for any of a variety of reasons. The recent literature has emphasized two broad types of constraints. On one hand, there may literature has emphasized two broad types of constraints. On one hand, there may be a limit on the size of the losses that the intermediary would be subject to in be a limit on the size of the losses that the intermediary would be subject to in bad states of the world, relative to its capital; such limits may result from regula-bad states of the world, relative to its capital; such limits may result from regula-tory capital requirements, or (the case of greatest relevance in the recent crisis) tory capital requirements, or (the case of greatest relevance in the recent crisis) such limits may be imposed by the intermediary’s creditors, who are unwilling to such limits may be imposed by the intermediary’s creditors, who are unwilling to supply additional funding if the leverage constraint is exceeded (as in Zigrand, supply additional funding if the leverage constraint is exceeded (as in Zigrand, Shin, and Danielsson, 2010; Adrian, Moench, and Shin, 2010b; Adrian and Shin, Shin, and Danielsson, 2010; Adrian, Moench, and Shin, 2010b; Adrian and Shin, forthcoming b; Beaudry and Lahiri, 2010).forthcoming b; Beaudry and Lahiri, 2010).66

Alternatively, intermediaries may raise funds by pledging particular assets as Alternatively, intermediaries may raise funds by pledging particular assets as collateral for individual loans, and the amount that they can borrow may be limited collateral for individual loans, and the amount that they can borrow may be limited by the value of available collateral. Gârleanu and Pedersen (2010) and Ashcraft, by the value of available collateral. Gârleanu and Pedersen (2010) and Ashcraft, Gârleanu, and Pedersen (forthcoming) consider the consequences of collateral Gârleanu, and Pedersen (forthcoming) consider the consequences of collateral constraints in a model where the fraction of each asset’s value that can be borrowed constraints in a model where the fraction of each asset’s value that can be borrowed using that asset as collateral is among the defi ning characteristics of the asset. using that asset as collateral is among the defi ning characteristics of the asset. Geanakoplos (1997, 2003, 2010) instead proposes a theory in which margin require-Geanakoplos (1997, 2003, 2010) instead proposes a theory in which margin require-ments are endogenously determined in competitive markets.ments are endogenously determined in competitive markets.

Under these types of theories, the capital of intermediaries becomes a crucial Under these types of theories, the capital of intermediaries becomes a crucial determinant of the supply of intermediation. For a given quantity of capital, the determinant of the supply of intermediation. For a given quantity of capital, the

5 This is one of two relatively reduced-form models of endogenous credit spreads considered in the monetary dynamic stochastic general equilibrium model we present in Cúrdia and Woodford (2009). The device of a “loan production function” is also used in Goodfriend and McCallum (2007) and in Gerali, Neri, Sessa, and Signoretti (2010).6 The “value-at-risk constraint” assumed by authors such as Zigrand, Shin, and Danielsson (2010), Adrian, Moench, and Shin (2010b), and Adrian and Shin (forthcoming b) is an example of a constraint of this form. Beaudry and Lahiri (2010) impose a similar constraint by simply assuming that intermediaries can sell only riskless debt. The constraint assumed by Adrian and Shin (forthcoming b) is formally equivalent to the one assumed by Beaudry and Lahiri (2010), though the former authors prefer to interpret the constraint as one on value at risk.


supply schedule supply schedule XS will be upward sloping, as shown in Figure 3B, if the accept- will be upward sloping, as shown in Figure 3B, if the accept-able leverage ratio is higher when the spread between the expected return on able leverage ratio is higher when the spread between the expected return on the assets in which intermediaries can invest and the rate they must pay on their the assets in which intermediaries can invest and the rate they must pay on their liabilities is greater. Consider, for example, a value-at-risk constraint, that requires liabilities is greater. Consider, for example, a value-at-risk constraint, that requires the future value of the intermediary’s assets to be worth at least some fraction the future value of the intermediary’s assets to be worth at least some fraction k of the amount owed on its debt, with at least some probability 1 of the amount owed on its debt, with at least some probability 1 – p; and suppose ; and suppose that the risky asset in which the intermediary invests will pay at least a fraction that the risky asset in which the intermediary invests will pay at least a fraction s of of its expected payoff with probability 1 its expected payoff with probability 1 – p. Then the value-at-risk constraint is satis-. Then the value-at-risk constraint is satis-fi ed if and only if the intermediary’s leverage ratio (debt as a fraction of the total fi ed if and only if the intermediary’s leverage ratio (debt as a fraction of the total value of its assets) is no greater than value of its assets) is no greater than s/k times the factor (1 times the factor (1 ++ i bb)/(1)/(1 ++ i ss), where ), where i bb is the expected return on the risky asset and is the expected return on the risky asset and i s s is the rate that the intermediary is the rate that the intermediary must pay on its debt. Thus the acceptable leverage ratio, and correspondingly the must pay on its debt. Thus the acceptable leverage ratio, and correspondingly the maximum value of assets that the intermediary can acquire, will be an increasing maximum value of assets that the intermediary can acquire, will be an increasing function of the credit spread.function of the credit spread.

The The IS––MPMP Model with Credit Frictions Model with Credit FrictionsThe equilibrium credit spread and volume of credit shown in Figure 3A are The equilibrium credit spread and volume of credit shown in Figure 3A are

determined for a particular value of national income determined for a particular value of national income Y ; because the schedules ; because the schedules LS and and LD depend on depend on Y, as shown in Figure 2A, the location of the schedule , as shown in Figure 2A, the location of the schedule XD (at (at least) in Figure 3B also depends on least) in Figure 3B also depends on Y. For reasons already discussed above, a higher . For reasons already discussed above, a higher level of level of Y should shift should shift LS to the right and to the right and LD to the left, and each of these effects to the left, and each of these effects results in a lower equilibrium value of the rate results in a lower equilibrium value of the rate i ss paid to savers, for any given position paid to savers, for any given position of the schedule of the schedule XS. Hence we can once again derive an . Hence we can once again derive an IS schedule, indicating the schedule, indicating the equilibrium value of equilibrium value of i ss for any assumed level of income for any assumed level of income Y, but now the , but now the IS schedule schedule will also include a given assumption about the supply of intermediation.will also include a given assumption about the supply of intermediation.77

The resulting model makes many of the same qualitative predictions about the The resulting model makes many of the same qualitative predictions about the effects of economic disturbances or policy changes as the standard effects of economic disturbances or policy changes as the standard IS–MP model model (which is simply the special case in which the curve (which is simply the special case in which the curve XS is assumed to be horizontal is assumed to be horizontal at at ωω == 0). However, the dependence of the supply of intermediation on the capital 0). However, the dependence of the supply of intermediation on the capital of intermediaries provides a channel for the amplifi cation and propagation of the of intermediaries provides a channel for the amplifi cation and propagation of the effects of economic disturbances. An increase in aggregate economic activity will effects of economic disturbances. An increase in aggregate economic activity will generally increase the value of intermediaries’ assets (loans are more likely to be generally increase the value of intermediaries’ assets (loans are more likely to be repaid, land prices increase with increases in income, and so on) and hence their repaid, land prices increase with increases in income, and so on) and hence their net worth. This will allow additional borrowing by the intermediaries, and hence a net worth. This will allow additional borrowing by the intermediaries, and hence a larger volume of credit for any given credit spread. Thus, the supply of intermedia-larger volume of credit for any given credit spread. Thus, the supply of intermedia-tion schedule tion schedule XS will shift down and to the right. A reduction in the interest rate will shift down and to the right. A reduction in the interest rate

7 In fact, we can now solve for both i s and i b as functions of Y, but it is the relation between i s and Y that is relevant for the IS–MP diagram, since it is i s—the rate at which intermediaries are able to fund themselves—that corresponds to the operating target of the central bank. Plotting again the reaction of the central bank’s target for i s to changes in economic activity as a curve MP, we again have a diagram of exactly the kind shown in Figure 2B to determine simultaneously the equilibrium values of the interest rate and output; the only important difference is that now we must clarify that the interest rate on the vertical axis is the policy rate i s rather than the borrowing rate i b. Once the equilibrium values of Y and i s have been determined, they can be transferred back to Figures 3A and B to determine the implied equilibrium values of i b and L as well.


i ss at which intermediaries are able to fund themselves can also increase intermedi- at which intermediaries are able to fund themselves can also increase intermedi-aries’ net worth, if (as is often the case) they fund longer-term assets with short-term aries’ net worth, if (as is often the case) they fund longer-term assets with short-term borrowing that they must roll over, and in this case a reduction in borrowing that they must roll over, and in this case a reduction in i ss will shift the will shift the XS curve down and to the right as well.curve down and to the right as well.

Each of these effects will make the Each of these effects will make the IS curve fl atter (more interest-elastic) than curve fl atter (more interest-elastic) than it would otherwise be.it would otherwise be.88 This means that a shift in the This means that a shift in the MP curve—due either to a curve—due either to a change in monetary policy or to a supply-side disturbance that shifts the aggregate-change in monetary policy or to a supply-side disturbance that shifts the aggregate-supply curve—will have a larger effect on output as a consequence of these “fi nancial supply curve—will have a larger effect on output as a consequence of these “fi nancial accelerator” effects. Bernanke and Gertler (1995) discuss evidence for the impor-accelerator” effects. Bernanke and Gertler (1995) discuss evidence for the impor-tance of such effects in the case of monetary policy shocks. Moreover, if a disturbance tance of such effects in the case of monetary policy shocks. Moreover, if a disturbance leads to an increase or decrease in the capital of the intermediary sector, the altered leads to an increase or decrease in the capital of the intermediary sector, the altered level of capital is likely to persist for some time. This can result in effects on economic level of capital is likely to persist for some time. This can result in effects on economic activity that are more persistent than the initial disturbance.activity that are more persistent than the initial disturbance.

The presence of an upward-sloping The presence of an upward-sloping XS curve (representing credit frictions) curve (representing credit frictions) essentially makes the essentially makes the IS curve steeper, and consequently acts to dampen the effects curve steeper, and consequently acts to dampen the effects on aggregate output of disturbances that shift the on aggregate output of disturbances that shift the MP curve, to the extent that curve, to the extent that the the XS schedule is not itself shifted by the disturbances. In fact, however, the schedule is not itself shifted by the disturbances. In fact, however, the XS schedule may well shift, in which case the net effect may well be to amplify output schedule may well shift, in which case the net effect may well be to amplify output fl uctuations, rather than to dampen them.fl uctuations, rather than to dampen them.

Consequences of Shifts in the Supply of IntermediationConsequences of Shifts in the Supply of IntermediationA more important consequence of this extension of the model is the fact that A more important consequence of this extension of the model is the fact that

shifts in the shifts in the XS schedule—for either exogenous or endogenous reasons—become schedule—for either exogenous or endogenous reasons—become an additional source of variations in aggregate demand, and hence in economic an additional source of variations in aggregate demand, and hence in economic activity and infl ation.activity and infl ation.99 A disruption of the supply of intermediation will shift the A disruption of the supply of intermediation will shift the XS schedule up, so that fi nancial intermediaries supply less credit at every level schedule up, so that fi nancial intermediaries supply less credit at every level of the credit spread of the credit spread ω. ω. As shown in Figure 4A an upward shift in As shown in Figure 4A an upward shift in XS results in a results in a higher equilibrium credit spread, and a lower volume of lending, for any given level higher equilibrium credit spread, and a lower volume of lending, for any given level of economic activity (refl ected in the location of the of economic activity (refl ected in the location of the XD schedule in the fi gure). schedule in the fi gure). Transferring this larger spread back to Figure 3A, one observes that the implied Transferring this larger spread back to Figure 3A, one observes that the implied value of value of i ss will be smaller, and the implied value of will be smaller, and the implied value of i bb higher, for the given value of higher, for the given value of Y. Because this is true for each possible value of . Because this is true for each possible value of Y, the , the IS schedule is shifted down schedule is shifted down and to the left, as shown in Figure 4B. (Note that and to the left, as shown in Figure 4B. (Note that IS plots the equilibrium value of plots the equilibrium value of i s s rather than that of rather than that of i bb, because the policy reaction function specifi es a target for , because the policy reaction function specifi es a target for i ss rather than for rather than for i bb.) In the absence of any change in the monetary policy reaction .) In the absence of any change in the monetary policy reaction

8 Of course, these reasons for the IS curve to be fl atter must be balanced against the observation that a steeper XS curve for a given level of capital in the intermediary sector will imply a steeper IS curve. This is why the degree of amplifi cation from credit frictions that is found in quantitative dynamic stochastic general equilibrium models is sometimes quite modest.9 The empirical dynamic stochastic general equilibrium models of Christiano, Motto, and Rostagno (2010) and Gilchrist, Ortiz, and Zakrajsek (2009) each attribute a substantial fraction of the short-run variability of real GDP to disturbances that vary the severity of fi nancial frictions.


Figure 4Effects of a Disruption of Credit Supply

Notes: XS and XD are the supply and demand for intermediation. ω is the spread between i b, the interest rate for borrowers, and i s, the interest rate for savers. The IS schedule shows the equilibrium interest rate for any assumed level of current income Y, and MP is the monetary policy reaction function (MP).

A: Effects on the Equilibrium Credit Spread ω and Volume of Lending Lfor a Given Level of Aggregate Income Y

B: Effects on the Equilibrium Policy Rate and Aggregate Income, Taking into Account the Monetary Policy Reaction

i1

i2

is

Inte

rest

rat

e

MP

IS

Y1 YAggregate income

Y2

ω1

ω2

ω

In

tere

st r

ate

spre

ad

(bet

wee

n s

aver

s an

d bo

rrow

ers)

XS

XD


L2

Michael Woodford 35

function, the upward shift in function, the upward shift in XS should result in both a decline in the policy interest should result in both a decline in the policy interest rate and a contraction of real activity.rate and a contraction of real activity.1010

This prediction matches the consequences observed, for example, when the This prediction matches the consequences observed, for example, when the Carter administration imposed credit controls in the second quarter of 1980. This Carter administration imposed credit controls in the second quarter of 1980. This policy was followed by a contraction in real GDP at a rate of minus 8 percent per policy was followed by a contraction in real GDP at a rate of minus 8 percent per year in that quarter, while the federal funds rate also fell from a level over 17 percent year in that quarter, while the federal funds rate also fell from a level over 17 percent per annum in April to only 9 percent by July 1980. The effects of a policy tightening per annum in April to only 9 percent by July 1980. The effects of a policy tightening of this kind cannot be understood as a shift of the of this kind cannot be understood as a shift of the MP curve (or curve (or LM curve) in a curve) in a conventional conventional IS–MP (or (or IS–LM) diagram, but they are easily understood when one ) diagram, but they are easily understood when one realizes how changes in the supply of intermediation schedule should be expected realizes how changes in the supply of intermediation schedule should be expected to shift the to shift the IS curve. curve.

The dependence of the supply of intermediation on the capital of interme-The dependence of the supply of intermediation on the capital of interme-diaries also introduces an important channel through which additional types of diaries also introduces an important channel through which additional types of disturbances can affect aggregate activity. Any disturbance that impairs the capital disturbances can affect aggregate activity. Any disturbance that impairs the capital of the banking sector will shift the schedule of the banking sector will shift the schedule XS up and to the left, with the effects up and to the left, with the effects just discussed. This means that shocks that might seem of only modest signifi cance just discussed. This means that shocks that might seem of only modest signifi cance for the aggregate economy—in terms, say, of the total value of business losses that for the aggregate economy—in terms, say, of the total value of business losses that directly result from the shock—can have substantial aggregate effects if the losses directly result from the shock—can have substantial aggregate effects if the losses in question happen to be concentrated in highly leveraged intermediaries, who in question happen to be concentrated in highly leveraged intermediaries, who suffer signifi cant reductions in their capital as a result. This was an important reason suffer signifi cant reductions in their capital as a result. This was an important reason for the dramatic aggregate effects in 2008–2009 of the losses in the U.S. subprime for the dramatic aggregate effects in 2008–2009 of the losses in the U.S. subprime mortgage market.mortgage market.

The supply of intermediation can also shift as a result of factors other than The supply of intermediation can also shift as a result of factors other than a change in the capital of intermediaries; in particular, leverage constraints can a change in the capital of intermediaries; in particular, leverage constraints can tighten or loosen, as a result of changes in the attitudes of intermediaries’ creditors tighten or loosen, as a result of changes in the attitudes of intermediaries’ creditors regarding the acceptable degree of leverage, or in the margin requirements associ-regarding the acceptable degree of leverage, or in the margin requirements associ-ated with borrowing against the securities that intermediaries hold. Gorton and ated with borrowing against the securities that intermediaries hold. Gorton and Metrick (2009), Adrian and Shin (2009), and Geanakoplos (2010) have all stressed Metrick (2009), Adrian and Shin (2009), and Geanakoplos (2010) have all stressed the importance of increases in margin requirements in the overnight repurchase the importance of increases in margin requirements in the overnight repurchase (or “repo”) market as a factor that contracted the supply of credit in 2008 and 2009.(or “repo”) market as a factor that contracted the supply of credit in 2008 and 2009.

Even when shocks to the supply of intermediation originate in a tightening of Even when shocks to the supply of intermediation originate in a tightening of leverage constraints and/or margin constraints owing to an increased assessment leverage constraints and/or margin constraints owing to an increased assessment of the risk associated with intermediaries’ assets, the effects of the shocks will be of the risk associated with intermediaries’ assets, the effects of the shocks will be amplifi ed by the dependence of the supply of intermediation on the capital of the amplifi ed by the dependence of the supply of intermediation on the capital of the intermediary sector. Intermediaries that are forced to sell assets as a result of tight-intermediary sector. Intermediaries that are forced to sell assets as a result of tight-ened leverage constraints are likely to suffer losses, and more so to the extent that ened leverage constraints are likely to suffer losses, and more so to the extent that many of them are forced to sell similar assets at the same time, or to the extent many of them are forced to sell similar assets at the same time, or to the extent that they are the only “natural buyers” of the assets in question. These losses will that they are the only “natural buyers” of the assets in question. These losses will then further reduce their capital, further reducing the amount that they are able then further reduce their capital, further reducing the amount that they are able

10 In this respect the framework sketched here agrees with the one proposed by Bernanke and Blinder (1988), who refer to the relation that I call the IS curve as the “commodities and credit curve” instead, precisely because it is shifted by credit-supply shocks in addition to the usual determinants of the IS curve. The framework proposed here differs from that of Bernanke and Blinder primarily in offering a different model of the supply of intermediation.


to borrow, and hence requiring further asset sales. The result is a vicious spiral to borrow, and hence requiring further asset sales. The result is a vicious spiral that under some circumstances can substantially reduce credit supply. The resulting that under some circumstances can substantially reduce credit supply. The resulting contraction of aggregate output may result in further losses to the banks, further contraction of aggregate output may result in further losses to the banks, further reducing their capital, and hence tightening credit supply even more.reducing their capital, and hence tightening credit supply even more.

The Most Recent U.S. Credit CycleThe Most Recent U.S. Credit Cycle

Understanding variations in fi nancial conditions over the most recent credit Understanding variations in fi nancial conditions over the most recent credit cycle requires attention to the behavior of multiple interest rates, not just the federal cycle requires attention to the behavior of multiple interest rates, not just the federal funds rate that is targeted by the Federal Reserve. As shown in Figure 5, the Fed funds rate that is targeted by the Federal Reserve. As shown in Figure 5, the Fed Open Market Committee raised its target for the funds rate to a higher level during Open Market Committee raised its target for the funds rate to a higher level during the period 2006–07; but fi nancial conditions did not tighten as much as one might the period 2006–07; but fi nancial conditions did not tighten as much as one might expect from the increase in the funds rate. First of all, spending decisions depend expect from the increase in the funds rate. First of all, spending decisions depend more on the level of more on the level of long-term interest rates, which in turn depend on the expected interest rates, which in turn depend on the expected average level of short rates over the coming decade, rather than the current level average level of short rates over the coming decade, rather than the current level of short rates alone. Since there was good reason to regard the low level of the of short rates alone. Since there was good reason to regard the low level of the federal funds rate in 2003–04 as a temporary anomaly,federal funds rate in 2003–04 as a temporary anomaly,1111 the long rate implied by the the long rate implied by the expected average future level of the short rates did not greatly increase as a result of expected average future level of the short rates did not greatly increase as a result of the increase in the funds rate between 2004 and 2006.the increase in the funds rate between 2004 and 2006.

Moreover, yields on long-term Treasury bonds did not rise by even this much. Moreover, yields on long-term Treasury bonds did not rise by even this much. The term premium, whichThe term premium, which indicates the amount by which the actual yield on a indicates the amount by which the actual yield on a long-term bond exceeds the expected average level of short-term interest rates over long-term bond exceeds the expected average level of short-term interest rates over the term to maturity of the bond, declined during this period, as Figure 5 illustrates the term to maturity of the bond, declined during this period, as Figure 5 illustrates for the case of a 10-year bond. In turn, the rates at which private parties can borrow for the case of a 10-year bond. In turn, the rates at which private parties can borrow are not those applicable to the U.S. Treasury; the fi gure also shows, for example, are not those applicable to the U.S. Treasury; the fi gure also shows, for example, that the spread between the average yield on Baa-rated corporate bonds and 10-year that the spread between the average yield on Baa-rated corporate bonds and 10-year Treasuries also fell between 2004 and 2006.Treasuries also fell between 2004 and 2006.1212 Hence corporate borrowing costs actu- Hence corporate borrowing costs actu-ally fell, despite the increase in the federal funds rate, owing to the declines in the ally fell, despite the increase in the federal funds rate, owing to the declines in the two spreads! In contrast, the increases in the two spreads during the fi nancial crisis two spreads! In contrast, the increases in the two spreads during the fi nancial crisis greatly increased the cost of borrowing.greatly increased the cost of borrowing.

Even in the case of short-term borrowing, the federal funds rate alone is Even in the case of short-term borrowing, the federal funds rate alone is not always an adequate measure of money market conditions. Figure 5 also plots not always an adequate measure of money market conditions. Figure 5 also plots the spread between the three-month U.S. dollar London Interbank Offer Rate the spread between the three-month U.S. dollar London Interbank Offer Rate

11 For example, see the “fi tted” long rates implied by the forecasting model of Kim and Wright (2005). Indeed, the series plotted in Figure 5 is taken from the estimates of Kim and Wright (2005); their series is updated at ⟨http://www.federalreserve.gov/econresdata/researchdata.htm⟩.12 The spread between yields on this class of moderately risky corporate bonds and on similar-maturity Treasury bonds is a commonly watched indicator of disturbances to the market for corporate debt, which is strongly correlated with variations in economic activity. Gilchrist, Ortiz, and Zakrajsek (2009) use an index of corporate bond spreads as a measure of the time-varying fi nancial wedge in an estimated monetary dynamic stochastic general equilibrium model and fi nd that the co-movements of the bond spreads with other aggregate variables are consistent with this interpretation.


(LIBOR)(LIBOR)1313 and the overnight interest-rate rate swap (OIS) rate, which can be viewed and the overnight interest-rate rate swap (OIS) rate, which can be viewed as essentially a market forecast of the average level of the federal funds rate over that as essentially a market forecast of the average level of the federal funds rate over that three-month period. The sharp increases in this spread during the crisis indicate three-month period. The sharp increases in this spread during the crisis indicate that the short-term borrowing costs of many banks (especially late in 2008) were that the short-term borrowing costs of many banks (especially late in 2008) were considerably higher than would be indicated by the federal funds rate.considerably higher than would be indicated by the federal funds rate.

It is popular to attribute the credit boom (at least in part) to the Federal Reserve It is popular to attribute the credit boom (at least in part) to the Federal Reserve having kept the federal funds rate “too low for too long,” but comparison of the having kept the federal funds rate “too low for too long,” but comparison of the path of the funds rate in Figure 5 with the measures of credit growth in Figure 1A path of the funds rate in Figure 5 with the measures of credit growth in Figure 1A shows that the increase in lending was greatest in 2006 and the fi rst half of 2007, shows that the increase in lending was greatest in 2006 and the fi rst half of 2007, after the federal funds rate had already returned to a level consistent with normal after the federal funds rate had already returned to a level consistent with normal benchmarks. Instead, the fact that spreads were unusually low precisely during the benchmarks. Instead, the fact that spreads were unusually low precisely during the period of strongest growth in lending—as can be seen by comparing the spreads period of strongest growth in lending—as can be seen by comparing the spreads

13 The LIBOR rate is an average of quoted rates at which banks are able to borrow funds for a short term (3 months, in the case of the series plotted here) on an uncollateralized basis. It is important not only because it is the cost of additional funds for some banks, but because other lending rates—such as the interest rate at which commercial and industrial loans are available to fi rms under existing loan commit-ments—are often tied to the LIBOR rate. For alternative interpretations of variations in the LIBOR–OIS spread, see Giavazzi (2008), Sarkar (2009), and Taylor and Williams (2009).

Figure 5The Federal Funds Rate Target and Some Interest-rate Spreads

Sources: The FF target is from the Federal Reserve Board; the 10-year term premium was calculated by Don H. Kim and Jonathan H. Wright (available at the Federal Reserve Board website); the Baa–Treasury spread is from the Federal Reserve Bank of St. Louis; the LIBOR-OIS spread is from Bloomberg.Notes: The “FF target” is the Federal Funds rate target. The “10-year term premium” is the amount by which the yield on a 10-year bond exceeds the expected average level of short-term interest rates over the term to maturity of the bond. The “Baa–Treasury spread” is the spread between Baa-rated corporate bonds and 10-year Treasuries. The “LIBOR–OIS spread” is the spread between the three-month U.S. dollar London Interbank Offer Rate (LIBOR) and the overnight interest-rate rate swap (OIS) rate.

2003 2004 2005 2006 2007 2008 2009 20100.0

1.0

2.0

3.0

4.0

5.0

6.0FF target10-year term premiumBaa-Treasury spreadLIBOR-OIS spread

Perc

enta

ge p

oin

ts


shown in Figure 5 with the quantities in Figure 1—indicates that an outward shift shown in Figure 5 with the quantities in Figure 1—indicates that an outward shift of the supply of intermediation schedule of the supply of intermediation schedule XS was responsible, rather than a move- was responsible, rather than a move-ment along this schedule in response to a loosening of monetary policy. The reason ment along this schedule in response to a loosening of monetary policy. The reason for the shift seems to have been an increased appetite of investors for purportedly for the shift seems to have been an increased appetite of investors for purportedly low-risk short-term liabilities of very highly leveraged fi nancial intermediaries; in low-risk short-term liabilities of very highly leveraged fi nancial intermediaries; in this journal, Brunnermeier (2009) details the changes in fi nancing patterns during this journal, Brunnermeier (2009) details the changes in fi nancing patterns during this period.this period.

The effects of such a shift were like those shown in Figure 4, but with the The effects of such a shift were like those shown in Figure 4, but with the reverse sign; as a consequence, the Fed’s increase in the funds rate over the period reverse sign; as a consequence, the Fed’s increase in the funds rate over the period between 2004 and 2006 did less to restrain demand than would ordinarily have been between 2004 and 2006 did less to restrain demand than would ordinarily have been expected.expected.1414 The increase in the riskless short-term rate did reduce households’ and The increase in the riskless short-term rate did reduce households’ and fi rms’ willingness to hold demand deposits, as a conventional money-demand equa-fi rms’ willingness to hold demand deposits, as a conventional money-demand equa-tion would imply, and checkable deposits declined during this period, as shown in tion would imply, and checkable deposits declined during this period, as shown in Figure 1B; but this did not prevent a net increase in the overall liabilities of fi nancial Figure 1B; but this did not prevent a net increase in the overall liabilities of fi nancial intermediaries, and hence in credit supply.intermediaries, and hence in credit supply.

The fi nancial crisis that began in summer 2007 also originated in a change in The fi nancial crisis that began in summer 2007 also originated in a change in the supply of intermediation. It began when increased perceptions of risk resulted the supply of intermediation. It began when increased perceptions of risk resulted in increases in the margin requirements demanded by creditors in short-term in increases in the margin requirements demanded by creditors in short-term lending collateralized by mortgage-backed securities, creating a liquidity crisis for lending collateralized by mortgage-backed securities, creating a liquidity crisis for issuers of asset-backed commercial paper. The effect of deleveraging in this sector issuers of asset-backed commercial paper. The effect of deleveraging in this sector on the market value of mortgage-backed securities further impaired the capital of on the market value of mortgage-backed securities further impaired the capital of fi nancial intermediaries more broadly, requiring further deleveraging, in a vicious fi nancial intermediaries more broadly, requiring further deleveraging, in a vicious spiral: again, Brunnermeier (2009) describes this process in detail in this journal.spiral: again, Brunnermeier (2009) describes this process in detail in this journal.

In terms of the model, the net result of both reductions in the acceptable degree In terms of the model, the net result of both reductions in the acceptable degree of leverage and impairment of the capital of the fi nancial sector was a sharp leftward of leverage and impairment of the capital of the fi nancial sector was a sharp leftward shift of the supply of intermediation shift of the supply of intermediation XS. As illustrated by Figure 4, the result was a . As illustrated by Figure 4, the result was a simultaneous contraction of the volume of lending, as shown in Figure 1, and an simultaneous contraction of the volume of lending, as shown in Figure 1, and an increase in spreads, as shown in Figure 5. The resulting leftward shift of the increase in spreads, as shown in Figure 5. The resulting leftward shift of the IS curve curve (Figure 4) meant a contraction of aggregate demand, despite the substantial cuts (Figure 4) meant a contraction of aggregate demand, despite the substantial cuts in the federal funds rate shown in Figure 5. The reduction in the riskless short-term in the federal funds rate shown in Figure 5. The reduction in the riskless short-term rate caused an increased willingness to hold transactions deposits, and checkable rate caused an increased willingness to hold transactions deposits, and checkable deposits increased substantially, as seen in Figure 1B. But plentiful deposits were deposits increased substantially, as seen in Figure 1B. But plentiful deposits were not enough to restore the fl ow of credit, for an inability to increase the volume of not enough to restore the fl ow of credit, for an inability to increase the volume of deposits was not the relevant constraint on the supply of credit.deposits was not the relevant constraint on the supply of credit.

Once this process was underway—and given that, for a time, it appeared Once this process was underway—and given that, for a time, it appeared that the crisis might spiral out of control—uncertainty about the macroeconomic that the crisis might spiral out of control—uncertainty about the macroeconomic environment likely caused a further leftward shift of the environment likely caused a further leftward shift of the IS curve, by increasing curve, by increasing precautionary saving and increasing the option value of deferring investment. precautionary saving and increasing the option value of deferring investment. Once the Once the IS curve shifted suffi ciently far, it ceased to be possible to maintain curve shifted suffi ciently far, it ceased to be possible to maintain

14 Under this analysis, the fact that the Fed did not tighten policy even further can be said to have contrib-uted to the credit boom. But the problem was not that the Fed failed to conform to the conventional benchmark provided by the “Taylor rule,” as argued by Taylor (2009), but rather that it followed it too faithfully, rather than taking account of the change in fi nancial conditions.

Michael Woodford 39

output near potential through cuts in the federal funds rate alone, owing to the output near potential through cuts in the federal funds rate alone, owing to the zero lower bound on nominal interest rates. Of course, the fact that reduced zero lower bound on nominal interest rates. Of course, the fact that reduced aggregate demand resulted in lower economic activity and employment, rather aggregate demand resulted in lower economic activity and employment, rather than simply in reductions in wages and prices to the extent needed to maintain than simply in reductions in wages and prices to the extent needed to maintain full employment, depended on the stickiness of wages and prices, as described in full employment, depended on the stickiness of wages and prices, as described in standard textbook accounts.standard textbook accounts.

Implications for Monetary PolicyImplications for Monetary Policy

To what extent does this extension of the standard model imply changes to the To what extent does this extension of the standard model imply changes to the conventional conduct of monetary policy? conventional conduct of monetary policy?

Taking Account of Financial ConditionsTaking Account of Financial ConditionsThe model’s most obvious implication is that decisions about interest-rate policy The model’s most obvious implication is that decisions about interest-rate policy

should take account of changes in fi nancial conditions—in particular, of changes in should take account of changes in fi nancial conditions—in particular, of changes in interest-rate spreads. Suppose that one’s goal is to set a value of the policy rate at each . Suppose that one’s goal is to set a value of the policy rate at each point in time that is consistent with output equal to potential (or, more precisely, point in time that is consistent with output equal to potential (or, more precisely, the “natural rate of output” in the sense of Friedman, 1968). In the model sketched the “natural rate of output” in the sense of Friedman, 1968). In the model sketched above, this interest rate can be determined at any time given two other numbers: above, this interest rate can be determined at any time given two other numbers: 1) the current value of the “natural rate of interest”—the real interest rate required 1) the current value of the “natural rate of interest”—the real interest rate required for output equal to the natural rate, in the absence of fi nancial frictionsfor output equal to the natural rate, in the absence of fi nancial frictions1515—converted —converted into an equivalent nominal interest rate by adding the current expected infl ation into an equivalent nominal interest rate by adding the current expected infl ation rate, and 2) the current interest-rate spread rate, and 2) the current interest-rate spread ωω..1616

The model therefore suggests that changes in credit spreads should be an The model therefore suggests that changes in credit spreads should be an important indicator in setting the federal funds rate; the funds rate target should be important indicator in setting the federal funds rate; the funds rate target should be lower than would otherwise be chosen, given other conditions, when credit spreads lower than would otherwise be chosen, given other conditions, when credit spreads are larger. John Taylor (2008) has proposed, in this spirit, that his well-known are larger. John Taylor (2008) has proposed, in this spirit, that his well-known rule for setting the federal funds rate target (explained in Taylor, 1993) should rule for setting the federal funds rate target (explained in Taylor, 1993) should be modifi ed to specify a funds rate target equal to that prescribed by the standard be modifi ed to specify a funds rate target equal to that prescribed by the standard “Taylor rule” minus the current value of the LIBOR–OIS spread shown in Figure 5. “Taylor rule” minus the current value of the LIBOR–OIS spread shown in Figure 5. In Cúrdia and Woodford (2010a), my coauthor and I show, in the context of a In Cúrdia and Woodford (2010a), my coauthor and I show, in the context of a New Keynesian dynamic stochastic general equilibrium model with credit frictions, New Keynesian dynamic stochastic general equilibrium model with credit frictions, that such a modifi cation of the standard Taylor rule can improve the economy’s that such a modifi cation of the standard Taylor rule can improve the economy’s response to disturbances to the supply of intermediation.response to disturbances to the supply of intermediation.

15 This concept, derived from the ideas of Knut Wicksell, is discussed extensively in Woodford (2003, chap. 4). One might alternatively defi ne the natural rate as the real rate that would be required for output equal to the natural rate of output under the assumption of a credit spread equal to some normal (steady state) level; the important feature of the proposed defi nition is that it abstracts from the effects of variations in the size of credit frictions.16 In Cúrdia and Woodford (2009), we derive an intertemporal version of the “IS curve” in which the credit spread appears as a shift factor. Gaspar and Kashyap (2006) were perhaps the fi rst to propose such a relation.


Alternatively, a forecast-targeting approach to monetary policy of the kind I Alternatively, a forecast-targeting approach to monetary policy of the kind I recommended in this journal in Woodford (2007)—in which the central bank’s recommended in this journal in Woodford (2007)—in which the central bank’s target for the policy rate should be adjusted as necessary in order for its projections target for the policy rate should be adjusted as necessary in order for its projections of infl ation and real activity to satisfy a quantitative target criterion—will automati-of infl ation and real activity to satisfy a quantitative target criterion—will automati-cally incorporate responses to changes in fi nancial conditions to the extent that cally incorporate responses to changes in fi nancial conditions to the extent that these shift the these shift the IS curve, as in the model sketched above. In addition, this alternative curve, as in the model sketched above. In addition, this alternative approach has the advantage of not requiring the central bank to focus on a single approach has the advantage of not requiring the central bank to focus on a single interest rate spread when multiple aspects of fi nancial conditions are each relevant interest rate spread when multiple aspects of fi nancial conditions are each relevant to aggregate demand and supply determination.to aggregate demand and supply determination.

“Unconventional” Monetary Policies“Unconventional” Monetary PoliciesThe model also implies that traditional interest-rate policy alone will not, in The model also implies that traditional interest-rate policy alone will not, in

general, provide a fully adequate response to a disturbance to credit supply, no general, provide a fully adequate response to a disturbance to credit supply, no matter how large the cut in the policy rate that may be engineered. The reason is matter how large the cut in the policy rate that may be engineered. The reason is that even if a suffi cient reduction in the policy rate can offset the decline in aggre-that even if a suffi cient reduction in the policy rate can offset the decline in aggre-gate demand that would otherwise result from the shift in the gate demand that would otherwise result from the shift in the IS curve, this does not curve, this does not fully undo the distortions created by the increase in credit spreads. To the extent fully undo the distortions created by the increase in credit spreads. To the extent that savers would be willing to supply additional funds at an interest rate lower than that savers would be willing to supply additional funds at an interest rate lower than the rate at which borrowers would be willing to borrow additional funds, then there the rate at which borrowers would be willing to borrow additional funds, then there remains a misallocation of expenditure, even if the aggregate level of expenditure is remains a misallocation of expenditure, even if the aggregate level of expenditure is optimal: on this point, in Cúrdia and Woodford (2009), my coauthor and I provide optimal: on this point, in Cúrdia and Woodford (2009), my coauthor and I provide an explicit welfare analysis. Thus, to the extent that it is possible for policy to reduce an explicit welfare analysis. Thus, to the extent that it is possible for policy to reduce the size of the credit spread, this is desirable, even when interest-rate policy is able the size of the credit spread, this is desirable, even when interest-rate policy is able to maintain output at potential.to maintain output at potential.

But the case for acting to reduce credit spreads becomes even stronger if But the case for acting to reduce credit spreads becomes even stronger if the policy rate is constrained by the zero lower bound on nominal interest rates. the policy rate is constrained by the zero lower bound on nominal interest rates. In the case of a large enough disturbance to the supply of intermediation, the In the case of a large enough disturbance to the supply of intermediation, the IS curve may shift so far down to the left that the point on it corresponding to the curve may shift so far down to the left that the point on it corresponding to the natural rate of output may involve a negative nominal interest rate; for quanti-natural rate of output may involve a negative nominal interest rate; for quanti-tative examples, see Cúrdia and Woodford (2010b). In this case, conventional tative examples, see Cúrdia and Woodford (2010b). In this case, conventional monetary policy is unable to achieve the required level of aggregate demand, monetary policy is unable to achieve the required level of aggregate demand, because even a massive expansion of the supply of bank reserves cannot drive the because even a massive expansion of the supply of bank reserves cannot drive the policy rate below zero. (The Federal Reserve found itself in this situation after policy rate below zero. (The Federal Reserve found itself in this situation after December 2008, as shown in Figure 5.) Under such circumstances, a policy that December 2008, as shown in Figure 5.) Under such circumstances, a policy that can reduce credit spreads can further increase aggregate demand (by shifting can reduce credit spreads can further increase aggregate demand (by shifting the the IS curve to the right), despite the lack of room for any further reduction in curve to the right), despite the lack of room for any further reduction in the policy rate.the policy rate.

Broadly speaking, two types of “unconventional” central-bank policies can Broadly speaking, two types of “unconventional” central-bank policies can reduce credit spreads by shifting the supply of intermediation schedule reduce credit spreads by shifting the supply of intermediation schedule XS to the to the right. One is the extension of credit to intermediaries by the central bank on easier right. One is the extension of credit to intermediaries by the central bank on easier terms than are available from private creditors; in particular, in the case that the terms than are available from private creditors; in particular, in the case that the relevant fi nancing constraint is the existence of too-high margin requirements for relevant fi nancing constraint is the existence of too-high margin requirements for private lending using assets held by the intermediaries as collateral, the central private lending using assets held by the intermediaries as collateral, the central bank may choose to lend against that collateral with a lower margin requirement. bank may choose to lend against that collateral with a lower margin requirement.


Ashcraft, Gârleanu, and Pedersen (forthcoming) discuss the Federal Reserve’s Term Ashcraft, Gârleanu, and Pedersen (forthcoming) discuss the Federal Reserve’s Term Asset-Backed Lending Facility, which provided fi nancing for private purchases of Asset-Backed Lending Facility, which provided fi nancing for private purchases of asset-backed securities, as an example of a policy of this kind and present evidence asset-backed securities, as an example of a policy of this kind and present evidence of its success at reducing the spreads associated with asset-backed securities eligible of its success at reducing the spreads associated with asset-backed securities eligible for the program. Such a policy can relax the constraint on the size of intermediary for the program. Such a policy can relax the constraint on the size of intermediary balance sheets resulting from limited capital in the intermediary sector, by allowing balance sheets resulting from limited capital in the intermediary sector, by allowing increased leverage.increased leverage.

Alternatively, the central bank may directly purchase debt claims issued by Alternatively, the central bank may directly purchase debt claims issued by private borrowers, so that total credit extended to the private sector can exceed private borrowers, so that total credit extended to the private sector can exceed the size of intermediary balance sheets. Examples of policies of this kind during the size of intermediary balance sheets. Examples of policies of this kind during the recent crisis include the Fed’s purchases of commercial paper through its the recent crisis include the Fed’s purchases of commercial paper through its Commercial Paper Funding Facility and its purchases of mortgage-backed secu-Commercial Paper Funding Facility and its purchases of mortgage-backed secu-rities and agency debt. On the motivation for and effects of these programs, rities and agency debt. On the motivation for and effects of these programs, see, for example, Adrian, Kimbrough, and Marchioni (2010), Gagnon, Raskin, see, for example, Adrian, Kimbrough, and Marchioni (2010), Gagnon, Raskin, Remache, and Sack (2010), and in this journal Kacperczyk and Schnabl (2010). Remache, and Sack (2010), and in this journal Kacperczyk and Schnabl (2010). In this case as well, the supply of intermediation In this case as well, the supply of intermediation XS is shifted to the right even is shifted to the right even though the equilibrium relation between the credit spread and the quantity of though the equilibrium relation between the credit spread and the quantity of risky assets that can be held on the balance sheets of private intermediaries does risky assets that can be held on the balance sheets of private intermediaries does not change.not change.1717

It should not be assumed that because it is possible in principle for the central It should not be assumed that because it is possible in principle for the central bank to reduce equilibrium spreads through direct intervention in credit markets, bank to reduce equilibrium spreads through direct intervention in credit markets, it is therefore desirable for the central bank to intervene continually to maintain it is therefore desirable for the central bank to intervene continually to maintain zero spreads. In Cúrdia and Woodford (2010b), we assume costs of central-bank zero spreads. In Cúrdia and Woodford (2010b), we assume costs of central-bank lending to the private sector that imply that under normal circumstances, it lending to the private sector that imply that under normal circumstances, it will not be optimal for the central bank to hold assets other than highly liquid will not be optimal for the central bank to hold assets other than highly liquid Treasury securities on its balance sheet; but even so, central-bank lending to the Treasury securities on its balance sheet; but even so, central-bank lending to the private sector can be justifi ed on welfare grounds in the case of a large enough private sector can be justifi ed on welfare grounds in the case of a large enough disruption of credit supply. Gertler and Karadi (2010) reach a similar conclusion disruption of credit supply. Gertler and Karadi (2010) reach a similar conclusion using a related model. using a related model.

Monetary Policy and Financial StabilityMonetary Policy and Financial StabilityFinally, the fact that a reduction in the capital of intermediaries has an adverse Finally, the fact that a reduction in the capital of intermediaries has an adverse

effect on the supply of intermediation—which in turn can seriously disturb both aggre-effect on the supply of intermediation—which in turn can seriously disturb both aggre-gate demand and the composition of expenditure—implies that it is desirable to reduce gate demand and the composition of expenditure—implies that it is desirable to reduce how frequently such crises occur. The role that monetary policy can or should play in how frequently such crises occur. The role that monetary policy can or should play in this regard remains controversial. However, a crisis that sharply reduces intermediary this regard remains controversial. However, a crisis that sharply reduces intermediary capital can more easily occur, in the sense that the size of the required exogenous capital can more easily occur, in the sense that the size of the required exogenous disturbance is smaller, when intermediaries are highly leveraged. Thus, while the disturbance is smaller, when intermediaries are highly leveraged. Thus, while the increased volume of lending that a relaxation of leverage constraints makes possible increased volume of lending that a relaxation of leverage constraints makes possible

17 Note that on this analysis, the effects of targeted central bank asset purchases have nothing to do with “quantitative easing,” as the effects do not depend on the purchases being fi nanced by an increase in bank reserves, nor do conditions in the market for bank reserves play any role in our analysis. See Cúrdia and Woodford (2010b) for further discussion.


can improve the short-run allocation of resources, this benefi t must be weighed against can improve the short-run allocation of resources, this benefi t must be weighed against the increased risk of occurrence of a crisis that will (if it occurs) increase distortions in the increased risk of occurrence of a crisis that will (if it occurs) increase distortions in the future, in ways that monetary policy will not then be able to counteract fully.the future, in ways that monetary policy will not then be able to counteract fully.

The model sketched here implies that increased leverage in the fi nancial sector The model sketched here implies that increased leverage in the fi nancial sector is a natural consequence of looser monetary policy because of the effects of higher is a natural consequence of looser monetary policy because of the effects of higher incomes on loan demand and supply, shown in Figure 2A. Other, more complex incomes on loan demand and supply, shown in Figure 2A. Other, more complex mechanisms, such as the model of misperception of risk by the funders of interme-mechanisms, such as the model of misperception of risk by the funders of interme-diaries proposed by Dubecq, Mojon, and Ragot (2009) can make this effect even diaries proposed by Dubecq, Mojon, and Ragot (2009) can make this effect even stronger. Given this, the consequences of policy for fi nancial stability need to be stronger. Given this, the consequences of policy for fi nancial stability need to be considered in making interest-rate decisions, alongside the consequences of policy considered in making interest-rate decisions, alongside the consequences of policy for aggregate economic activity and infl ation.for aggregate economic activity and infl ation.

The nature of this consideration should not be completely symmetrical: The nature of this consideration should not be completely symmetrical: marginal adjustments of interest rates always have consequences for output and marginal adjustments of interest rates always have consequences for output and infl ation, while they will have nonnegligible consequences for the risk of fi nancial infl ation, while they will have nonnegligible consequences for the risk of fi nancial instability only at certain times when the leverage is extreme enough for even small instability only at certain times when the leverage is extreme enough for even small changes in asset values to have substantial effects on intermediary capital. Improved changes in asset values to have substantial effects on intermediary capital. Improved regulation and/or macroprudential supervision could further reduce the range of regulation and/or macroprudential supervision could further reduce the range of circumstances in which this consideration would matter for monetary policy deci-circumstances in which this consideration would matter for monetary policy deci-sions; and this would be desirable, if possible, as freeing monetary policy to focus sions; and this would be desirable, if possible, as freeing monetary policy to focus solely on output and infl ation stabilization should allow those goals to be more solely on output and infl ation stabilization should allow those goals to be more effectively achieved. But in the absence of a complete solution of that kind, it is effectively achieved. But in the absence of a complete solution of that kind, it is diffi cult to defend the view that fi nancial stability can be ignored in monetary policy diffi cult to defend the view that fi nancial stability can be ignored in monetary policy decisions; and the development of practical real-time indicators of risks to fi nancial decisions; and the development of practical real-time indicators of risks to fi nancial stability is accordingly an important challenge of the present moment.stability is accordingly an important challenge of the present moment.

■ I would like to thank Tobias Adrian, Bill Brainard, Vasco Cúrdia, Jamie McAndrews, Benoit Mojon, Tommaso Monacelli, Julio Rotemberg, and Argia Sbordone for helpful discussions, Luminita Stevens for research assistance, and the editors of this journal, David Autor, Chad Jones, and Timothy Taylor, for many useful comments on earlier drafts. I would also like to thank the National Science Foundation for research support under grant number SES-0820438.

References

Adrian, Tobias, Karin Kimbrough, and Dina Marchioni. 2010. “The Federal Reserve’s Commer-cial Paper Funding Facility.” Federal Reserve Bank of New York Staff Report 423.

Adrian, Tobias, Emanuel Moench, and Hyun Song Shin. 2010a. “Financial Intermediation, Asset Prices and Macroeconomic Dynamics.” Federal Reserve Bank of New York Staff Report 422.

Adrian, Tobias, Emanuel Moench, and Hyun Song Shin. 2010b. “Macro Risk Premium and

Intermediary Balance Sheet Quantities.” Federal Reserve Bank of New York Staff Report 428.

Adrian, Tobias, and Hyun Song Shin. 2009. “Prices and Quantities in the Monetary Policy Transmission Mechanism.” International Journal of Central Banking, 5(4): 131–42.

Adrian, Tobias, and Hyun Song Shin. Forth-coming (a). “The Changing Nature of Financial Intermediation and the Financial Crisis of 2007–09.” Annual Review of Economics.

Michael Woodford 43

Adrian, Tobias, and Hyun Song Shin. Forthcom-ing (b). “Financial Intermediaries and Monetary Economics.” In Handbook of Monetary Economics, vol. 3, ed. B. M. Friedman and M. Woodford. Amsterdam: Elsevier.

Ashcraft, Adam, Nicolae Gârleanu, and Lasse Heje Pedersen. Forthcoming. “Two Monetary Tools: Interest Rates and Haircuts.” In NBER Macroeconomics Annual 2010, ed. D. Acemoglu and M. Woodford. Chicago: University of Chicago Press.

Baker, Dean. 2010. “Blame It on the Bubble.” The Guardian, March 8.

Beaudry, Paul, and Amartya Lahiri. 2010. “Risk Allocation, Debt Fueled Expansion, and Financial Crisis.” June. http://faculty.arts.ubc.ca/alahiri/BeaudryLahiri.pdf.

Bennett, Paul, and Stavros Peristiani. 2002. “Are U.S. Reserve Requirements Still Binding?” Federal Reserve Bank of New York Economic Policy Review, 8(1): 53–68.

Bernanke, Ben S., and Alan S. Blinder. 1988. “Credit, Money, and Aggregate Demand.” American Economic Review, 78(2): 435–39.

Bernanke, Ben S., and Mark Gertler. 1995. “Inside the Black Box: The Credit Channel of Monetary Policy Transmission.” Journal of Economic Perspectives, 9(4): 27–48.

Bernanke, Ben S., Mark Gertler, and Simon Gilchrist. 1999. “The Financial Accelerator in a Quantitative Business Cycle Framework.” Chap. 21 in Handbook of Macroeconomics, vol. 1C, ed. J. B. Taylor and M. Woodford. Amsterdam: Elsevier.

Brunnermeier, Markus. 2009. “Deciphering the Liquidity and Credit Crunch 2007–2008.” Journal of Economic Perspectives, 23(1): 77–100.

Buiter, Willem H. 2010. “Housing Wealth Isn’t Wealth.” Economics: The Open-Access, Open Assess-ment E-Journal, vol. 4, no. 2010-22. http://www.economics-ejournal.org/economics/journalarticles/2010-22.

Christiano, Lawrence, Roberto Motto, and Massimo Rostagno. 2010. Financial Factors in Economic Fluctuations. European Central Bank Working Paper 1192. http://www.ecb.europa.eu/pub/pdf/scpwps/ecbwp1192.pdf.

Cúrdia, Vasco, and Michael Woodford. 2009. “Credit Frictions and Optimal Monetary Policy.” http://www.columbia.edu/~mw2230/credit_0709.

Cúrdia, Vasco, and Michael Woodford. 2010a. “Credit Spreads and Monetary Policy,” Journal of Money, Credit and Banking, 42(6, Supp.): 3–35.

Cúrdia, Vasco, and Michael Woodford. 2010b.“The Central-Bank Balance Sheet as an Instrument of Monetary Policy.” NBER Working Paper 16208.

Dubecq, Simon, Benoit Mojon, and Xavier Ragot. 2009. “Fuzzy Capital Requirements, Risk Shifting, and the Risk-Taking Channel of Monetary Policy.” Banque de France Working Paper 254. http://www.banquedefrance.fr/gb/publications/telechar/ner/DT254.pdf.

Friedman, Milton. 1968. “The Role of Monetary Policy.” American Economic Review, 58(1): 1–17.

Friedman, Milton, and Anna J. Schwartz. 1963. A Monetary History of the United States, 1867–1960. Princeton: Princeton University Press.

Gagnon, Joseph, Matthew Raskin, Julie Remache, and Brian Sack. 2010. “Large-Scale Asset Purchases by the Federal Reserve: Did They Work?” Federal Reserve Bank of New York Staff Report 441.

Gârleanu, Nicolae, and Lasse Heje Pedersen. 2010. “Margin-Based Asset Pricing and Deviations from the Law of One Price.” http://pages.stern.nyu.edu/~lpederse/papers/MarginPricingLoOP.pdf.

Gaspar, Vitor, and Anil K. Kashyap. 2006. “Stability First: Refl ections Inspired by Otmar Issing’s Success as the ECB’s Chief Economist.” In Monetary Policy: A Journey from Theory to Practice, 86–118. Frankfurt: European Central Bank.

Geanakoplos, John. 1997. “Promises, Promises.” In The Economy as an Evolving Complex System II, ed. W. B. Arthur, S. N. Durlauf. and D. A. Lane, 285–320. Reading, MA: Addison Wesley.

Geanakoplos, John. 2003. “Liquidity, Default and Crashes: Endogenous Contracts in General Equilibrium.” Chap. 5 in Advances in Economics and Econometrics, Theory and Applications II, ed. M. Dewatripont, L. P. Hansen, and S. J. Turnovsky. Cambridge, Cambridge University Press.

Geanakoplos, John. 2010. “The Leverage Cycle.” In NBER Macroeconomics Annual 2009, ed. D. Acemoglu, K. Rogoff, and M. Woodford, 1–65. Chicago: University of Chicago Press.

Gerali, Andrea, Stefano Neri, Luca Sessa, and Federico M. Signoretti. 2010. “Credit and Banking in a DSGE Model of the Euro Area.” Journal of Money, Credit and Banking, 42(S1): 107–141.

Gertler, Mark, and Peter Karadi. 2010. “A Model of Unconventional Monetary Policy.” http://www.carnegie-rochester.rochester.edu/april10-pdfs/Gertler%20Karadi.pdf.

Gertler, Mark, and Nobuhiro Kiyotaki. Forth-coming. “Financial Intermediation and Credit Policy in Business Cycle Analysis.” In Handbook of Monetary Economics, vol. 3, ed. B. M. Friedman and M. Woodford. Amsterdam: Elsevier.

Giavazzi, Francesco. 2008. “Why Does the Spread between LIBOR and Expected Future Policy Rates Persist, and Should Central Banks Do Something About It?” VoxEU.org, posted June 2.

Gilchrist, Simon, Alberto Ortiz, and Egon


Zakrajsek. 2009. “Credit Risk and the Macroecon-omy: Evidence from an Estimated DSGE Model.” http://sws.bu.edu/aortizb/Credit%20Risk%20and%20the%20Macroeconomy%20Gilchrist%20Ortiz%20%20Zakrajsek.pdf.

Goodfriend, Marvin, and Bennett T. McCallum. 2007. “Banking and Interest Rates in Monetary Policy Analysis: A Quantitative Exploration.” Journal of Monetary Economics, 54(5): 1480–1507.

Gorton, Gary B., and Andrew Metrick. 2009. “Haircuts.” NBER Working Paper 15273.

Greenlaw, David, Jan Hatzius, Anil K. Kashyap, and Hyun Song Shin. 2008. Leveraged Losses: Lessons from the Mortgage Market Meltdown. U.S. Monetary Policy Forum Report No. 2. Chicago: University of Chicago Booth School of Business.

Kacperczyk, Marcin, and Philipp Schnabl. 2010. “When Safe Proved Risky: Commercial Paper during the Financial Crisis of 2007–2009.” Journal of Economic Perspectives, 24(1): 29–50.

Kashyap, Anil, and Jeremy Stein. 1994. “Mone-tary Policy and Bank Lending.” In Monetary Policy, ed. N. G. Mankiw, 221–56. Chicago: University of Chicago Press.

Kim, Don H., and Jonathan H. Wright. 2005. “An Arbitrage-Free Three-Factor Term Structure Model and the Recent Behavior of Long-Term Yields and Distant-Horizon Forward Rates.” Finance and Economics Discussion Series No. 2005-33, Federal Reserve Board, August.

Kiyotaki, Nobuhiro, and John Moore. 1997. “Credit Cycles.” Journal of Political Economy, 105(2): 211–48.

Romer, David. 2000. “Keynesian Macro-

economics without the LM Curve.” Journal of Economic Perspectives, 14(2): 149–69.

Sarkar, Asani. 2009. “Liquidity Risk, Credit Risk, and the Federal Reserve’s Responses to the Crisis.” Financial Markets and Portfolio Management, 23(4): 335–48.

Smant, David. 2002. “Bank Credit in the Transmission of Monetary Policy: A Critical Review of the Issues and Evidence.” MRPA paper 19816, Munich Personal RePEc Archive.

Taylor, John B. 1993. “Discretion versus Policy Rules in Practice.” Carnegie-Rochester Conference Series on Public Policy, vol. 39, pp 195–214.

Taylor, John B. 2008. “Monetary Policy and the State of the Economy.” Testimony before the Committee on Financial Services, U.S. House of Representatives, February 26.

Taylor, John B. 2009. Getting Off Track: How Government Actions and Interventions Caused, Prolonged, and Worsened the Financial Crisis. Stanford: Hoover Press.

Taylor, John B., and John C. Williams. 2009. “A Black Swan in the Money Market,” American Economic Journal: Macroeconomics, 1(1): 58–83.

Woodford, Michael. 2003. Interest and Prices: Foundations of a Theory of Monetary Policy. Princeton: Princeton University Press.

Woodford, Michael. 2007. “The Case for Forecast Targeting as a Monetary Policy Strategy.” Journal of Economic Perspectives, 21(4): 3–24.

Zigrand, Jean-Pierre, Hyun Song Shin, and Jon Danielsson. 2010. “Risk Appetite and Endog-enous Risk.” Financial Markets Group Discussion Paper 647.

This article has been cited by:

1. Giuliana Passamani, Roberto Tamborini. 2013. Monetary policy through the ‘credit-cost channel’:Italy and Germany pre- and post-EMU. Applied Economics 45:29, 4095-4113. [CrossRef]

2. Stijn Claessens, M. Ayhan Kose, Marco E. Terrones. 2012. How do business and financial cyclesinteract?. Journal of International Economics 87:1, 178-190. [CrossRef]

3. Piotr Radziszewski, Jerzy Pi#at, Karol J. Kowalski, Jan Król. 2012. Use Of Aggregate FromGlacier Deposits In High-Traffic Asphalt Pavements: A Polish Experience. The Baltic Journal ofRoad and Bridge Engineering 7:1, 5-12. [CrossRef]

4. PAOLO ANGELINI, ANDREA NOBILI, CRISTINA PICILLO. 2011. The Interbank Marketafter August 2007: What Has Changed, and Why?. Journal of Money, Credit and Banking 43:5,923-958. [CrossRef]

5. M. Ayhan Kose. 2011. Journal of International Economics . [CrossRef]

http://dx.doi.org/10.1080/00036846.2012.748176

http://dx.doi.org/10.1016/j.jinteco.2011.11.008

http://dx.doi.org/10.3846/bme.2012.01

http://dx.doi.org/10.1111/j.1538-4616.2011.00402.x

http://dx.doi.org/10.1016/j.jinteco.2011.03.001

Date post:	28-Apr-2015
Category:	Documents
Upload:	oscar-neira-cuadra
View:	44 times
Download:	0 times

Michael Woodford 2010 Financial Intermediation and Macroeconomic Analysis

Documents