Bank Liquidity Creation, Monetary Policy, and Financial Crises

Transcription

1 Bank Liquidity Creation, Monetary Policy, and Financial Crises Allen N. Berger University of South Carolina, Wharton Financial Institutions Center, and CentER Tilburg University Christa H.S. Bouwman MIT Sloan School of Management (visiting), Wharton Financial Institutions Center, and Case Western Reserve University July 2009 We study the relationship between aggregate bank liquidity creation and monetary policy in the U.S. from 1984:Q1 to 2008:Q4 by first examining the effectiveness of monetary policy during normal times and financial crises, and by then analyzing the behavior of both monetary policy and liquidity creation around financial crises. Our main findings are as follows. First, liquidity creation has increased substantially over the sample period, from $1.398 trillion to $5.304 trillion (in real 2008:Q4 dollars). The share of large banks (assets over $3 billion) in aggregate liquidity creation has also gone up over this time period, as has the fraction of liquidity created off the balance sheet. Second, the effect of monetary policy on bank liquidity creation is significant only for small banks, and additional interesting size-dependent differences appear after splitting liquidity creation into on- and offbalance sheet components. There appears to be no significant difference between the effects of monetary policy on liquidity creation during normal times and financial crises. If the Federal Reserve had used the Taylor rule, liquidity creation would have been similar during normal times, but higher during financial crises, and would have resulted in a 3.3% increase in liquidity creation over the sample period. Third, detrended liquidity creation is relatively high prior to financial crises, tends to drop significantly during financial crises and stays low afterwards. The data suggest that the level of detrended liquidity creation is a better indicator of a crisis occurring than detrended GDP, the federal funds rate, or the return on the stock market. The predicted probability of a crisis striking was only 0.02% when detrended liquidity creation was at its lowest point over the sample period, and 79.5% when detrended liquidity creation was at its highest point. Fourth, we find that the subprime lending crisis has made monetary policy more effective for small banks, less effective for medium banks, and equally effective for large banks, and left undisturbed the result that monetary policy has no different effect during normal times and crises. In addition, the high level of detrended liquidity creation before the crisis was far more pronounced than that before other financial crises, and it was driven by a high level of off-balance sheet illiquid guarantees (primarily loan commitments). When the crisis hit, illiquid assets (primarily business loans) increased for several quarters before declining. During the crisis, the drop in illiquid guarantees was far more pronounced than the drop in illiquid assets, and contributed significantly to the decline in detrended liquidity creation. Contact details: Moore School of Business, University of South Carolina, 1705 College Street, Columbia, SC Tel: Fax: aberger@moore.sc.edu. On leave from Case Western Reserve University. Contact details: MIT Sloan School of Management, 77 Massachusetts Avenue, Cambridge MA Tel: Fax: cbouwman@mit.edu. Keywords: Financial Crises, Liquidity Creation, and Banking. JEL Classification: G28, and G21. This is a significantly expanded version of the first part of an earlier paper, Financial Crises and Bank Liquidity Creation. The authors thank Bob DeYoung, John Driscoll, Bill English, Paolo Fulghieri, Thomas Kick, Loretta Mester, Steven Ongena, Bruno Parigi, Peter Ritchken, David Romer, Asani Sarkar, Klaus Schaeck, Greg Udell, Todd Vermilyea, Egon Zakrajsek, and participants at presentations at the CREI / JFI / CEPR Conference on Financial Crises at Pompeu Fabra, the Philadelphia Federal Reserve, the San Francisco Federal Reserve, the Cleveland Federal Reserve, the International Monetary Fund, the Summer Research Conference in Finance at the ISB in Hyderabad, the Unicredit Conference on Banking and Finance, the University of Kansas Southwind Finance Conference, Erasmus University, and Tilburg University for useful comments. We are indebted to Christopher Crowe for providing us with monetary policy shock data.

2 Bank Liquidity Creation, Monetary Policy, and Financial Crises 1. Introduction Over the past quarter century, the U.S. has experienced a number of financial crises. At the heart of these financial crises are often issues surrounding liquidity provision by the banking sector (e.g., Acharya, Shin, and Yorulmazer 2009). For example, in the current subprime lending crisis, liquidity seemed to have dried up for an extended time period. The practical importance of liquidity during financial crises is buttressed by financial intermediation theory, which indicates that the creation of liquidity is an important reason why banks exist. 1 Early contributions argue that banks create liquidity by financing relatively illiquid assets such as business loans with relatively liquid liabilities such as transactions deposits (e.g., Bryant 1980, Diamond and Dybvig 1983). More recent contributions suggest that banks also create liquidity off the balance sheet through loan commitments and similar claims to liquid funds (e.g., Holmstrom and Tirole 1998, Kashyap, Rajan, and Stein 2002). 2 The creation of liquidity makes banks fragile and susceptible to runs (e.g., Diamond and Dybvig 1983, Chari and Jagannathan 1988), and such runs can lead to financial crises via contagion effects. Such crises can affect the real economy if they rupture the creation of liquidity (e.g., Dell Ariccia, Detragiache, and Rajan 2008). 3 Because of the importance of liquidity provision by banks to the real economy, the central bank s monetary policy seeks to counter the effects of negative liquidity shocks. According to the bank lending channel literature, monetary policy may affect bank lending and deposits (for survey papers on this, see Bernanke and Gertler 1995, Kashyap and Stein 1997). Moreover, it may also affect off-balance sheet activities like loan commitments (e.g., Woodford 1996, Morgan 1998). It is plausible that the conduct of monetary policy varies across normal and financial crisis periods and that monetary policy affects and possibly interacts with aggregate liquidity creation by banks. Yet, we know little about the relationship 1 According to the theory, another central role of banks in the economy is to transform credit risk (e.g., Diamond 1984, Ramakrishnan and Thakor 1984, Boyd and Prescott 1986). Recently, Coval and Thakor (2005) theorize that banks may also arise in response to the behavior of irrational agents in financial markets. See Bhattacharya and Thakor (1993) and Freixas and Rochet (2008) for a summary of financial intermediary existence theories. 2 James (1981) and Boot, Thakor, and Udell (1991) endogenize the loan commitment contract due to informational frictions. The loan commitment contract is subsequently used in Holmstrom and Tirole (1998) and Kashyap, Rajan, and Stein (2002) to show how banks can provide liquidity to borrowers. 3 Acharya and Pedersen (2005) show that liquidity risk also affects the expected returns on stocks. 1

3 between monetary policy and bank liquidity creation, and particularly about how this relationship changes during and around financial crises. This paper seeks to fill this void in the literature. Since monetary policy potentially influences both sides of the bank s balance sheet as well as off-balance sheet activities, we use a comprehensive measure of bank liquidity creation that includes all on- and off-balance-sheet activities. Specifically, we address four broad questions. First, what has been the magnitude of liquidity creation (and some key components) by US banks of different size classes and how has this changed over time? Second, what has been the effect of monetary policy on aggregate bank liquidity creation, and does this effect differ in financial crises from normal times? Third, how have bank liquidity creation and monetary policy behaved during financial crises and in the periods immediately preceding and following those crises? Fourth, from the perspective of bank liquidity creation, how is the current subprime lending crisis different from other financial crises? To address these questions, our analyses use data on virtually all banks in the U.S. from 1984:Q1 2008:Q4. The sample period includes five financial crises: the 1987 stock market crash, the credit crunch of the early 1990s, the Russian debt crisis plus the Long-Term Capital Management meltdown in 1998, the bursting of the dot.com bubble plus the September 11 terrorist attack of the early 2000s, and the subprime lending crisis of 2007?. For our first question, we calculate the amount of liquidity created by the banking sector using Berger and Bouwman s (forthcoming) preferred liquidity creation measure. 4 This measure takes into account the fact that banks create liquidity both on and off the balance sheet and is constructed using a three-step procedure. In the first step, all bank assets, liabilities, equity, and off-balance sheet activities are classified as liquid, semi-liquid, or illiquid. This is done based on the ease, cost, and time for customers to obtain liquid funds from the bank, and the ease, cost, and time for banks to dispose of their obligations in order to meet these liquidity demands. In the second step, weights are assigned to these activities. The weights are consistent with the theory in that maximum liquidity is created when illiquid assets (e.g., business loans) are transformed into liquid liabilities (e.g., transactions deposits) and maximum liquidity is 4 While the preferred measure treats the securitizability of assets as fixed over time, we also construct an alternative liquidity creation measure (as in Berger and Bouwman forthcoming) that recognizes the fact that the ability to securitize assets has changed over time (see Section 4.2). 2

4 destroyed when liquid assets (e.g., treasuries) are transformed into illiquid liabilities (e.g., subordinated debt) or equity. In the third step, we combine the activities as classified in the first step and as weighted in the second step to construct a liquidity creation measure. We apply this measure to quarterly data on virtually all U.S. commercial and credit card banks from 1984:Q1 to 2008:Q4. The dollar amount of liquidity creation by these individual banks is added up to obtain aggregate liquidity creation by the banking sector. We find that liquidity creation has increased substantially over the sample period, from $1.398 trillion to $5.304 trillion (in real 2008:Q4 dollars), and that this growth has outpaced GDP growth in percentage terms over the same time period. The share of large banks in aggregate liquidity creation has also gone up over time, as has the fraction of liquidity created off the balance sheet. Turning to the second question, we examine the effect of monetary policy on aggregate bank liquidity creation during normal times and crisis periods by regressing the change in liquidity creation on the change in monetary policy, alternatively proxied by the change in the federal funds rate and Romer and Romer s (2004) monetary policy shocks, using several specifications. 5 We also address how the effect of monetary policy would have changed if the Federal Reserve had used the Taylor rule in setting the federal funds rate. Our main findings on this issue are that monetary policy seems to significantly affect liquidity creation only for small banks. Liquidity creation by large and medium banks (which create roughly 90% of aggregate bank liquidity) is mostly unaffected by monetary policy. There appears to be no significant difference between the effects of monetary policy on liquidity creation during normal times and financial crises. To the extent that monetary policy affects liquidity creation, the channel through which this effect occurs varies depending on bank size. Monetary policy seems to mostly affect on-balance sheet liquidity creation by small and medium banks, and off-balance sheet liquidity creation by large banks. However, the large-bank effect is positive, suggesting that increased demand for off-balance sheet guarantees outweighs any drop in supply caused by tightening monetary policy. Finally, we find that if the Federal Reserve had used the Taylor rule in conducting monetary policy, liquidity creation would have been similar during normal times, but higher during financial crises, and would have resulted in a 3.3% increase in liquidity 5 While Romer and Romer s (2004) monetary policy shock data ends in 1996:Q4, we use monetary policy shock data provided by Christopher Crowe (used in Barakchian and Crowe 2009) that covers our entire sample period. 3

5 creation over the sample period. To address the third question, we analyze the behavior of bank liquidity creation and monetary policy around financial crises. Since liquidity creation has grown considerably over time and also contains cyclical components, we do not focus on the actual dollar amount of liquidity creation for these analyses but follow an approach commonly used in the macroeconomics literature (e.g. Barro, 1997). Specifically, we first deseasonalize the data using the U.S. Census Bureau s X11 procedure, and then detrend it using the Hodrick Prescott (HP) filter. Henceforth, we call deseasonalized, detrended liquidity creation detrended liquidity creation for short. Our main results on this issue are that detrended liquidity creation is relatively high prior to financial crises, but tends to drop significantly during these crises and stays low afterwards. The federal funds rate is also often high prior to financial crises, but tends to fall as monetary policy generally loosens during the crises. Thus, detrended liquidity creation generally drops during financial crises despite loosened monetary policy. The high level of detrended liquidity creation that tends to precede financial crises suggests a possible dark side of bank liquidity creation. While financial fragility may be needed to induce banks to create liquidity (e.g., Diamond and Rajan 2000, 2001), our analysis raises the intriguing possibility that the causality may also be reversed in the sense that too much liquidity creation may lead to financial fragility. Results from logit regressions suggest that the level of detrended liquidity creation is significantly related to the probability of a crisis occurring one to four quarters hence. The predicted probability of a crisis striking is 0.02% when detrended liquidity creation is at its lowest point over the sample period, and it rises to 79.5% when detrended liquidity creation is at its highest point. In fact, our results show that detrended liquidity creation is a better indicator of a financial crisis than detrended GDP, the federal funds rate (monetary policy), or the return on the stock market. Finally, on the fourth question, we find that the subprime lending crisis has made monetary policy more effective for small banks, less effective for medium banks, and equally effective for large banks, and left undisturbed the result that monetary policy has no different effect during normal times and crises. Further differences, however, exist when we split liquidity creation into on- and off-balance sheet components. The behavior of detrended liquidity creation around the subprime lending crisis also differs in some important respects from previous financial crises. The high level of detrended liquidity creation 4

6 before the subprime lending crisis was driven by a high level of off-balance sheet illiquid guarantees (primarily loan commitments). When the crisis hit, illiquid assets (primarily business loans) increased for several quarters before declining. During the crisis, the drop in illiquid guarantees was far more pronounced than the drop in illiquid assets, and contributed significantly to the decline in detrended liquidity creation. The remainder of this paper is organized as follows. Section 2 discusses some related literature. Section 3 describes the five financial crises. Section 4 explains the monetary policy and liquidity creation measures, discusses our sample, and presents summary statistics. Section 5 examines the relationship between monetary policy and bank liquidity creation during normal times and financial crises. Section 6 describes the behavior of aggregate bank liquidity creation and monetary policy before, during, and after financial crises, and draws some general conclusions. Section 7 concludes. 2. Related literature This paper is primarily related to two literatures. The first is the literature on financial crises. 6 One strand of this literature focuses on financial crises and fragility. Some papers analyze contagion, showing that a small liquidity shock in one area may have a contagious effect throughout the economy (e.g., Allen and Gale 1998, 2000). Other papers focus on the determinants of financial crises and their policy implications (e.g., Bordo, Eichengreen, Klingebiel, and Martinez-Peria 2001, Demirguc-Kunt, Detragiache, and Gupta 2006, Lorenzoni 2008, Claessens, Klingebiel, and Laeven forthcoming). A second strand examines the effect of financial crises on the real sector (e.g., Friedman and Schwarz 1963, Bernanke 1983, Bernanke and Gertler 1989, Dell Ariccia, Detragiache, and Rajan 2008, Shin forthcoming). These papers find that financial crises increase the cost of financing and reduce credit, which adversely affects investment and may lead to reduced growth and recessions. That is, financial crises have independent real effects (see Dell Ariccia, Detragiache, and Rajan 2008). In contrast to these papers, we examine how the amount of liquidity created by the banking sector behaved around financial crises in the U.S. The second literature to which this paper is related focuses on the effect of monetary policy on bank behavior. The credit channel literature hypothesizes two distinct mechanisms through which 6 Allen and Gale (2007) provide a detailed overview of the causes and consequences of financial crises. 5

7 monetary policy may affect real economic activity, of which the bank lending channel is most relevant for our paper. 7 According to the bank lending channel, a tightening of monetary policy may cause a decline in bank deposits, which lessens the amount of loanable funds available. The reason is that the decline in deposits may not be offset by increases in other sources of funds such as federal funds or large CDs, or deposits may have to be replaced by funds with higher marginal costs (e.g., Bernanke and Blinder 1992, Stein 1998). In this way, a monetary contraction may result in a reduced supply of bank credit. Additionally, an increase in market interest rates caused by a tightening of monetary policy reduces the present value of fixed-rate loans in a bank s portfolio, which causes a decline in the bank s net worth, thereby resulting in a reduction in bank credit supply, particularly for banks that are constrained by relatively low capital ratios. 8 Although the bank lending channel literature generally does not discuss offbalance sheet activities, it may be argued that a monetary policy tightening would also reduce the supply of loan commitments, letters of credit, and similar off-balance sheet guarantees, given that these are present commitments to lend in the future. Thus, it seems appropriate to use a bank liquidity creation measure that incorporates both sides of the balance sheet as well as off-balance sheet activities, like the Berger and Bouwman (forthcoming) measure used here. 3. Five financial crises This section describes five financial crises that occurred between 1984:Q1 and 2008:Q4. The crises include the 1987 stock market crash, the credit crunch of the early 1990s, the Russian debt crisis plus Long- Term Capital Management (LTCM) bailout of 1998, the bursting of the dot.com bubble and the September 11 terrorist attacks of the early 2000s, and the current subprime lending crisis. Financial crisis #1: Stock market crash (1987:Q4) On Monday, October 19, 1987, the stock market crashed, with the S&P500 index falling about 20%. During the years before the crash, the level of the stock market had increased dramatically, causing some 7 The second mechanism is the balance sheet channel, which argues that monetary tightening affects real activity because the associated higher interest rates impair collateral values or otherwise reduce the net worth of certain borrowers, thereby making them less creditworthy and diminishing their ability to obtain funds. For more details on both mechanisms, see the survey paper by Bernanke and Gertler (1995). 8 See also the survey paper by Kashyap and Stein (1997). 6

8 concern that the market had become overvalued. 9 A few days before the crash, two events occurred that may have helped precipitate the crash: 1) legislation was enacted to eliminate certain tax benefits associated with financing mergers; and 2) information was released that the trade deficit was above expectations. Both events seemed to have added to the selling pressure and a record trading volume on Oct. 19, in part caused by program trading, overwhelmed many systems. Financial crisis #2: Credit crunch (1990:Q1 1992:Q4) During the first three years of the 1990s, bank commercial and industrial lending declined in real terms, particularly for small banks and for small loans (see Berger, Kashyap, and Scalise 1995, Table 8, for details). The ascribed causes of the credit crunch include a fall in bank capital from the loan loss experiences of the late 1980s (e.g., Peek and Rosengren 1995), the increases in bank leverage requirements and implementation of Basel I risk-based capital standards during this time period (e.g., Berger and Udell 1994, Hancock, Laing, and Wilcox 1995, Thakor 1996), an increase in supervisory toughness evidenced in worse examination ratings for a given bank condition (e.g., Berger, Kyle, and Scalise 2001), and reduced loan demand because of macroeconomic and regional recessions (e.g., Bernanke and Lown 1991). The existing research provides some support for each of these hypotheses. Financial crisis #3: Russian debt crisis / LTCM bailout (1998:Q3 1998:Q4) Since its inception in March 1994, hedge fund Long-Term Capital Management ( LTCM ) followed an arbitrage strategy that was avowedly market neutral, designed to make money regardless of whether prices were rising or falling. When Russia defaulted on its sovereign debt on August 17, 1998, investors fled from other government paper to the safe haven of U.S. treasuries. This flight to liquidity caused an unexpected widening of spreads on supposedly low-risk portfolios. By the end of August 1998, LTCM s capital had dropped to $2.3 billion, less than 50% of its December 1997 value, with assets standing at $126 billion. In the first three weeks of September, LTCM s capital dropped further to $600 million without shrinking the portfolio. Banks began to doubt its ability to meet margin calls. To prevent a potential systemic meltdown triggered by the collapse of the world s largest hedge fund, the Federal Reserve Bank 9 E.g., Raging bull, stock market s surge is puzzling investors: When will it end? on page 1 of the Wall Street Journal, Jan. 19,

9 of New York organized a $3.5 billion bail-out by LTCM s major creditors on September 23, In 1998:Q4, several large banks had to take substantial write-offs as a result of losses on their investments. Financial crisis #4: Bursting of the dot.com bubble and Sept. 11 terrorist attack (2000:Q2 2002:Q3) The dot.com bubble was a speculative stock price bubble that was built up during the mid- to late-1990s. During this period, many internet-based companies, commonly referred to as dot.coms, were founded. Rapidly increasing stock prices and widely available venture capital created an environment in which many of these companies seemed to focus largely on increasing market share. At the height of the boom, many dot.com s were able to go public and raise substantial amounts of money even if they had never earned any profits, and in some cases had not even earned any revenues. On March 10, 2000, the Nasdaq composite index peaked at more than double its value just a year before. After the bursting of the bubble, many dot.com s ran out of capital and were acquired or filed for bankruptcy (examples of the latter include WorldCom and Pets.com). The U.S. economy started to slow down and business investments began falling. The September 11, 2001 terrorist attacks may have exacerbated the stock market downturn by adversely affecting investor sentiment. By 2002:Q3, the Nasdaq index had fallen by 78%, wiping out $5 trillion in market value of mostly technology firms. Financial crisis #5: Subprime lending crisis (2007:Q3?) The subprime lending crisis has been characterized by turmoil in financial markets as banks have experienced difficulty in selling loans in the syndicated loan market and in securitizing loans. According to press reports, banks also seem to be reluctant to provide credit: they appear to have cut back their lending to firms and individuals, and have also been reticent to lend to each other. Some banks have experienced substantial losses in capital. Massive losses at Countrywide resulted in a takeover by Bank of America. Bear Stearns suffered a fatal loss in confidence and was sold at a fire-sale price to J.P. Morgan Chase with the Federal Reserve guaranteeing $29 billion in potential losses. Washington Mutual, the sixth-largest bank, became the biggest bank failure in the U.S. financial history. J.P. Morgan Chase purchased the banking business while the rest of the organization filed for bankruptcy. IndyMac Bank was seized by the FDIC after it suffered substantive losses and depositors had started to run on the bank. The FDIC sold all 8

10 deposits and most of the assets to OneWest Bank, FSB. The Federal Reserve also intervened in some unprecedented ways in the market. It extended its safety-net privileges to investment banks and one insurance company (AIG) and began holding mortgage-backed securities and lending directly to investment banks. The Treasury set aside $250 billion out of its $700-billion bailout package (TARP program) to enhance capital ratios of selected banks. 4. Monetary policy measures, bank liquidity creation, and the sample This section first describes the two monetary policy measures, the change in the federal funds rate and Romer and Romer s monetary policy shocks, and provides summary statistics. It then explains Berger and Bouwman s (forthcoming) preferred and an alternative liquidity creation measure. Finally, it describes the sample and provides summary statistics Two monetary policy measures To examine the relationship between monetary policy and liquidity creation, we select two monetary policy measures. The first measure is the federal funds rate. Since the Federal Reserve explicitly targeted the federal funds rate over our entire sample period (see Romer and Romer 2004), the change in the federal funds rate measures the change in monetary policy. 10 A drawback of this measure, however, is that it may contain anticipatory movements. That is, movements in the federal funds rate may respond to information about future developments in the economy, making it harder to isolate the effect of monetary policy on bank output. The second measure takes into account such endogeneity. Our second (change in) monetary policy measure is a monetary policy shock measure developed by Romer and Romer (2004). As a first step to construct this measure, the intended federal funds rate changes around meetings of the Federal Open Market Committee (FOMC), the institution responsible for setting monetary policy in the U.S., are obtained by examining narrative accounts of each FOMC meeting. Next, anticipatory movements are removed by regressing the intended federal funds rate on the Federal Reserve s internal forecasts of inflation and real activity. The residuals from this regression are the monetary policy shocks, i.e., the changes in the intended federal funds rate that are not made in response to forecasts of 10 We use the actual federal funds rate (as in Romer and Romer 2004) rather than the target federal funds rate since bank behavior will be affected most by the actual rate. 9

11 future economic conditions. While Romer and Romer s (2004) monetary policy shock data ends in 1996:Q4, Barakchian and Crowe (2009) extend this data through 2008:Q2, and Crowe provides a further extension to 2008:Q4. 11 Figure 1 shows the change in the federal funds rate (Panel A) and the Romer and Romer monetary policy shocks (Panel B) over time. The broad movements in the two series tend to be similar, but differences do exist. For example, if the Federal Reserve lowers interest rates by less than it normally would based on its internal forecasts of low inflation and sluggish growth, monetary loosening based on the change in the federal funds rate coincides with monetary tightening based on the monetary policy shock data Bank liquidity creation: preferred and alternative measure To construct a measure of liquidity creation, we follow Berger and Bouwman s (forthcoming) three-step procedure (see Table 1). Below, we briefly discuss these three steps. In Step 1, we classify all bank activities (assets, liabilities, equity, and off-balance sheet activities) as liquid, semi-liquid, or illiquid. For assets, this is based on the ease, cost, and time for banks to dispose of their obligations in order to meet these liquidity demands. For liabilities and equity, this is based on the ease, cost, and time for customers to obtain liquid funds from the bank. We follow a similar approach for off-balance sheet activities, classifying them based on functionally similar on-balance sheet activities. For all activities other than loans, this classification process uses information on both product category and maturity. Due to data restrictions, we classify loans entirely by category. In Step 2, we assign weights to all the bank activities classified in Step 1. The weights are consistent with liquidity creation theory, which argues that banks create liquidity on the balance sheet when they transform illiquid assets into liquid liabilities. We therefore apply positive weights to illiquid assets and liquid liabilities. Following similar logic, we apply negative weights to liquid assets and illiquid liabilities and equity, since banks destroy liquidity when they use illiquid liabilities to finance liquid assets. We use weights of ½ and -½, because only half of the total amount of liquidity created is attributable to the source or use of funds alone. For example, when $1 of liquid liabilities is used to finance $1 in illiquid 11 We are grateful to Christopher Crowe for making this data available to us. 10

12 assets, liquidity creation equals ½ * $1 + ½ * $1 = $1. In this case, maximum liquidity is created. However, when $1 of liquid liabilities is used to finance $1 in liquid assets, liquidity creation equals ½ * $1 + -½ * $1 = $0. In this case, no liquidity is created as the bank holds items of approximately the same liquidity as those it gives to the nonbank public. Maximum liquidity is destroyed when $1 of illiquid liabilities or equity is used to finance $1 of liquid assets. In this case, liquidity creation equals -½ * $1 + -½ * $1 = -$1. An intermediate weight of 0 is applied to semi-liquid assets and liabilities. Weights for off-balance sheet activities are assigned using the same principles. In Step 3, we combine the activities as classified in Step 1 and as weighted in Step 2 to construct Berger and Bouwman s (forthcoming) preferred liquidity creation measure. This measure classifies loans by category, while all activities other than loans are classified using information on product category and maturity, and includes off-balance sheet activities. 12 To obtain the dollar amount of liquidity creation at a particular bank, we multiply the weights of ½, -½, or 0, respectively, times the dollar amounts of the corresponding bank activities and add the weighted dollar amounts. Since the ability to securitize assets has changed greatly over time, we also construct an alternative liquidity creation measure as in Berger and Bouwman (forthcoming). This measure is identical to the preferred measure, except for the way we classify loans. For each loan category, we use U.S. Flow of Funds data on the total amount of loans outstanding and the total amount of loans securitized to calculate the fraction of loans that has been securitized in the market at each point in time. Following Loutskina (2006), we then assume that each bank can securitize that fraction of its own loans. To give an example, in 1993:Q4, $3.1 trillion in residential real estate loans were outstanding in the market, and 48.4% of these loans were securitized. If a bank has $10 million in residential real estate loans in that quarter, we assume that 48.4% thereof can be securitized. Hence, we classify $4.84 million of these loans as semi-liquid and the remainder as illiquid. Note that this alternative measure faces a significant drawback. While the theories suggest that the ability to securitize matters for liquidity creation, this measure uses the actual amount of securitization. Thus, while the vast majority of these residential real estate loans may be securitizable, this alternative measure treats only about half of them as such. 12 Berger and Bouwman (forthcoming) construct four liquidity creation measures by alternatively classifying loans by category or maturity, and by alternatively including or excluding off-balance sheet activities. However, they argue that the measure we use here is the preferred measure since for liquidity creation, banks ability to securitize or sell loans is more important than loan maturity, and banks do create liquidity both on and off the balance sheet. 11

13 We provide descriptive statistics on the preferred and the alternative liquidity creation measures in Section 4.4. Since we obtain qualitatively similar regression results based on the alternative measure, all reported regression results are based on the preferred measure for brevity Sample description We include virtually all commercial and credit card banks in the U.S. in our study. 13 For each bank, we obtain quarterly Call Report data from 1984:Q1 to 2008:Q4. We keep a bank in the sample if it: 1) has commercial real estate or commercial and industrial loans outstanding; 2) has deposits; 3) has gross total assets or GTA exceeding $25 million; 4) has an equity capital to GTA ratio of at least 1%. 14 For each bank, we calculate the dollar amount of liquidity creation in each quarter (933,209 bankquarter observations from 18,294 distinct banks) using the process described in Section 4.2. We aggregate these amounts to obtain the dollar amount of liquidity creation by the banking sector, and put these (and all other financial values) into real 2008:Q4 dollars using the implicit GDP price deflator. We thus end up with a final sample that contains 100 inflation-adjusted, quarterly liquidity creation amounts. Given documented differences between banks of different sizes in terms of portfolio composition (e.g., Kashyap, Rajan, and Stein 2002, Berger, Miller, Petersen, Rajan, and Stein 2005), we also split the sample by bank size into large, medium, and small banks, and perform our analyses separately for three sets of banks. Large banks have gross total assets (GTA) exceeding $3 billion, medium banks have GTA exceeding $1 billion and up to $3 billion, and small banks have GTA up to $1 billion Bank liquidity creation summary statistics Figure 2 Panel A shows the dollar amount of liquidity created by the banking sector, calculated using the preferred liquidity creation measure over our sample period. It also shows the breakout into on- and offbalance sheet liquidity creation. As shown, liquidity creation increased substantially over time: it almost 13 Berger and Bouwman (forthcoming) include only commercial banks. We also include credit card banks to avoid an artificial $0.19 trillion drop in bank liquidity creation in the fourth quarter of 2006 when Citibank N.A. moved its credit-card lines to Citibank South Dakota N.A., a credit card bank. 14 GTA equals total assets plus the allowance for loan and lease losses and the allocated transfer risk reserve (a reserve for certain foreign loans). Total assets on Call Reports deduct these two reserves, which are held to cover potential credit losses. We add these reserves back to measure the full value of the loans financed and the liquidity created by the bank on the asset side. 12

14 quadrupled from $1.398 trillion in 1984:Q1 to $5.304 trillion in 2008:Q4 (in real 2008:Q4 dollars). Using the alternative liquidity creation measure (see Section 4.2), liquidity creation grew from $1.715 trillion to $5.797 trillion over this period. Since the mid-1990s, off-balance sheet liquidity creation has exceeded and grown faster than on-balance sheet liquidity creation. Figure 2 Panel B shows that most of the liquidity in the banking sector is created by large banks and that their share of liquidity creation has increased from 76% in 1984 to 86% in Using the alternative measure, it increased from 70% to 87%. Over this same time frame, the shares of medium and small banks dropped from 8% to 5% and from 16% to 9%, respectively. Using the alternative measure, their shares dropped from 9% to 4% and from 21% to 8%, respectively. Figure 2 Panel C shows per capita liquidity creation over time. The picture reveals that per capita liquidity creation more than tripled from $5.9K in 1984:Q1 to $17.4K in 2008:Q4. Using the alternative measure, it almost tripled from $7.3K to $19.0K over this time period. Interestingly, the picture looks very similar to the one shown in Panel A, because the annual U.S. population growth rate is low relative to the growth in liquidity creation. Figure 2 Panel D shows the dollar amount of liquidity creation divided by GDP. The picture reveals that bank liquidity creation has increased from 19.9% of GDP in 1984:Q1 to 37.3% of GDP in 2008:Q4. Using the alternative liquidity creation measure, it increased from 24.4% to 40.8% of GDP over this time period. This picture also resembles the one shown in Panel A, but the increase over time is less pronounced. While liquidity creation more than quadrupled over the sample period, GDP doubled. 5. The effect of monetary policy on bank liquidity creation during normal times and financial crises The goals of monetary policy are to promote maximum sustainable output and employment and to promote stable prices. The FOMC sets the federal funds rate at a level it believes will foster conditions consistent with achieving these goals. Given the potential impact of monetary policy on both sides of the bank s balance sheet and on off-balance sheet guarantees via the bank lending channel, monetary policy can be expected to affect liquidity creation by banks. This section focuses on the second goal of the paper and examines the effect of monetary policy on liquidity creation during normal times and financial crises. We also analyze how the impact of monetary 13

15 policy would have been affected if the Federal Reserve had used the Taylor rule in setting monetary policy. We use regression analysis (in the spirit of Romer and Romer 2004) for this purpose, but acknowledge that our results, while indicative of causation, may merely reflect association Monetary policy and bank liquidity creation We start with the simplest possible regression. We regress the percentage change in aggregate liquidity creation (%ΔLC) on the (lagged) change in monetary policy and quarterly dummies (D SEASON) to control for seasonal effects. The change in monetary policy is alternatively defined as the change in the federal funds rate (ΔFEDFUNDS) and Romer and Romer s monetary policy shocks (POLICYSHOCKS). The specification based on ΔFEDFUNDS is: % (1) Inference is based on robust standard errors. Since tighter monetary policy (i.e. an increase in the federal funds rate or a positive monetary policy shock) is expected to be associated with a decrease in liquidity creation, we expect to find a negative coefficient on the change in monetary policy. Table 2 shows the results based on the change in the federal funds rate (Panel A) and the Romer and Romer monetary policy shocks (Panel B). 15 As can be seen in column (i), the coefficient on the change in monetary policy is positive for the all-bank sample, suggesting that tighter monetary policy is associated with an increase in liquidity creation (significant using the change in the federal funds rate). This result is puzzling and seems to be driven by large banks. For medium banks, the coefficient on monetary policy is not significant. For small banks, the coefficient is negative (significant based on the change in the federal funds rate), suggesting that for these banks, monetary policy works as expected, a finding consistent with Kashyap and Stein (2000). One reason why we find a positive effect of monetary policy on liquidity creation by large banks may be that large banks engage in far more off-balance sheet liquidity creation relative to on-balance sheet liquidity creation than small banks. While tighter monetary policy reduces the supply of on- and offbalance-sheet liquidity creation by for example restricting the supply of loans and loan commitments, the 15 Since Panel A regresses the percentage change in liquidity creation on the lagged change in the federal funds rate, these regressions have 98 (rather than 100) observations. Since Panel B regresses the percentage change in liquidity creation on lagged monetary policy shocks (available for every quarter), these regressions have 99 observations. 14

16 increase in the demand for loan commitments by borrowers may outweigh the supply effect. This is consistent with Thakor (2005), who shows that the value of the guarantee provided by loan commitments increases during periods of rationing. In Section 5.2 we provide evidence consistent with this. To examine whether the impact of monetary policy on liquidity creation varies across normal times and financial crises, we add a crisis dummy (D CRIS), and the (lagged) change in monetary policy interacted with the crisis dummy (alternatively defined as ΔFEDFUNDS * D CRIS or POLICYSHOCKS * D CRIS) to our specification. The crisis dummy equals one if there was a crisis in that quarter and is zero otherwise. The specification based on the change in the federal funds rate is: % (2) In these regressions, the coefficient on the change in monetary policy picks up the effect of monetary policy during normal times, while the coefficient on the interaction term shows whether monetary policy has a differential effect during financial crises. For example, if monetary policy is more effective during financial crises than during normal times, the coefficient on the interaction term will be negative and significant. Also, if liquidity creation is negatively affected during financial crises as one might expect, the coefficient on the crisis dummy will be negative. Table 2 column (ii) contains the results. As expected, the coefficient on the crisis dummy is negative and significant (in most cases), suggesting that liquidity creation indeed declines during financial crises. The coefficient on the change in monetary policy for the all-bank sample is now negative, but not significant. As before, this coefficient is significant only for small banks (based on the change in the federal funds rate). This suggests that monetary policy does not affect liquidity creation by large and medium banks, but provides some evidence that it does affect liquidity creation by small banks during normal times. The coefficients on the interaction terms are not significant, suggesting that the effect of monetary policy on liquidity creation is not different across crisis and non-crisis periods. It is intriguing that monetary policy does not seem to affect large and medium banks. We offer two possible reasons for this result. First, large and medium banks have better access to capital markets, which means that their cost and availability of funds are likely to be less affected by changes in monetary policy than those of small banks. Second, while the bank lending channel argues that tighter monetary policy 15

17 leads to a drop in lending, deposits, and off-balance sheet guarantees, the associated drop in liquidity creation may be offset by increased customer demand for off-balance sheet guarantees. It is possible that our results suffer from an omitted variables bias. One could argue, for example, that we need to control for demand for banking services in our regressions. 16 If this demand increased in the previous period, liquidity creation is expected to increase this period. We now address this issue by including the percentage change in GDP (%ΔGDP) in the regressions. Henceforth, we call this the complete specification. The specification based on the change in the federal funds rate is: % % (3) If changes in monetary policy affect liquidity creation, the coefficients on the change in monetary policy should be significant after controlling for changes in demand. Table 2 column (iii) contains the results. The coefficient on the change in GDP is positive and significant in the all-bank and large-bank regressions, suggesting that higher demand indeed fuels liquidity creation for these banks. However, controlling for demand does not materially affect the monetary policy results. To gauge the economic significance of the small-bank results, we focus on the coefficient on the change in the federal funds rate of (column (iii) in Panel A). This result suggests that if the federal funds rate increases by one percentage point, liquidity creation by small banks decreases by 1.3%. Evaluated at the average dollar amount of liquidity created by small banks of $333 billion, this translates into a drop in liquidity creation of $4.33 billion Monetary policy and on- versus off-balance sheet liquidity creation According to the bank lending channel, monetary tightening induces a drop in deposits and lending, and may have a negative effect on the supply of off-balance sheet guarantees such as loan commitments and letters of credit as well. While this suggests a decrease in on-balance sheet liquidity creation, off-balance sheet liquidity creation does not necessarily decline. Faced with rationing by banks in the spot credit 16 Regression specifications that also include the lagged percentage change in liquidity creation as a regressor or include the annual (rather than quarterly) change in monetary policy yield qualitatively similar results as the ones that are reported. Details are available on request. 16

18 market, customers may try to insure themselves against future rationing. The associated increase in demand for off-balance sheet guarantees may offset any decline in supply, and could lead to an increase off-balance sheet liquidity creation (Thakor 2005). Since large banks create far more liquidity off the balance sheet than small banks (see Berger and Bouwman forthcoming), the impact on off-balance sheet liquidity creation is expected to be strongest for these banks. We now investigate how monetary policy affects on- and off-balance sheet liquidity creation by re-running regressions (1) (3) from the previous section, with the percentage change in on- and off-balance sheet liquidity creation as dependent variables. Table 3 shows the regression results for on-balance sheet liquidity creation (Panel I) and offbalance sheet liquidity creation (Panel II). Subpanels A and B use the change in the federal funds rate and monetary policy shocks to measure the change in monetary policy, respectively. Results are shown in a similar way as in Table 2. Our discussion focuses on the complete specification in column (iii), which regresses the percentage change in (on- or off-balance sheet) liquidity creation on the change in monetary policy, the crisis dummy, the change in monetary policy interacted with the crisis dummy, the percentage change in GDP, and seasonal dummies. The results shown before (Table 2) suggested that monetary policy is effective only for small banks. The results in Table 3 Panel I provide evidence that monetary tightening reduces on-balance sheet liquidity creation of small banks (based on both monetary policy measures) and that of medium banks (significant based on the change in the federal funds rate). The small-bank effect on on-balance sheet liquidity creation seems stronger than the effect on overall liquidity creation in Table 2 based on both the federal funds rate (coefficient of versus ) and the Romer and Romer monetary policy shocks (coefficient of versus ). While monetary policy does not seem to affect total liquidity creation by large banks (see Table 2), the results in Table 3 Panel II suggest that it does have a significant effect on off-balance sheet liquidity creation for these banks (based on the monetary policy shock measure). There is no such effect for medium and small banks. The positive coefficient for large banks supports the hypothesis that firms try to obtain more off-balance sheet loan commitments from banks during periods of monetary tightening, and that large banks are most affected by this. The coefficient of suggests that a one percentage point bigger monetary policy shock is associated with a 6.1% increase in off-balance sheet liquidity creation, which 17

19 evaluated at the sample average of $1.773 trillion translates into a gain in off-balance sheet liquidity creation of $108.2 billion The effect of monetary policy on liquidity creation excluding the current subprime lending crisis The previous two subsections examine how monetary policy affects liquidity creation over the sample period. Since the current subprime lending crisis has been called the most serious crisis since the Great Depression, one might wonder to what extent the results presented above are driven by this crisis. To examine this, we rerun the regressions from Tables 2 and 3 using data from 1984:Q1 2004:Q4, a period that excludes the current crisis and the tremendous growth of liquidity creation that preceded it. 17 We only discuss the results based on the complete specification in column (iii) here (not shown for brevity). As before, we first focus on the effect of monetary policy on total liquidity creation. Using this shorter sample period, we find that monetary policy was somewhat less effective for small banks than reported in Table 2 above: the coefficients on liquidity creation are smaller (less negative) while still significant (based on the federal funds rate). Monetary policy was more effective for medium banks: the coefficients on liquidity creation are bigger (more negative) and are now significant. Monetary policy was equally ineffective for large banks: the coefficients and t-statistics are similar to those reported in Table 2 above. As above, we find that monetary policy affects liquidity creation during normal times and crises in a similar way for all size classes (the interaction term was not significant). These results suggest that the subprime lending crisis made monetary policy more effective for small banks, less effective for medium banks, and equally ineffective for large banks. Next, we split liquidity creation into on- and off-balance sheet liquidity creation. We find that using this shorter sample period, monetary policy affects on-balance sheet liquidity creation by small and medium banks more than reported in Table 3 Panel A above: the coefficients on on-balance sheet liquidity creation are more negative than before, while the t-statistics are similar. Monetary policy again does not significantly affect on-balance sheet liquidity creation by large banks. As above, we find no differential effect of monetary policy between normal periods and crises. These findings suggest that the subprime lending crisis has made monetary policy less effective for small and medium banks and had no effect on 17 We thank Loretta Mester for suggesting this additional analysis. 18

20 large banks in terms of on-balance sheet liquidity creation. Turning to off-balance sheet liquidity creation, we find that based on the shorter sample period, monetary policy again does not significantly affect off-balance sheet liquidity creation by small and medium banks, although the small-bank coefficients are now positive (not significant) based on the federal funds rate. Monetary policy has a larger positive effect for large banks during normal times than reported in Table 3 Panel B above (based on the Romer and Romer monetary policy shocks): the coefficients on offbalance sheet liquidity creation are bigger (more positive) as are the t-statistics. However, in contrast to our earlier findings, monetary policy has no effect on large banks during crisis periods based on the Romer and Romer monetary policy shocks (the sum of the coefficients on monetary policy and the interaction term is approximately zero). These results suggest that in terms of off-balance sheet liquidity creation, the subprime lending crisis has (insignificantly) enhanced the effectiveness of monetary policy for small banks; has not affected medium banks; has reduced the positive effect of monetary policy for large banks during normal times; and has increased the positive effect of monetary policy for these banks during crises Monetary policy and bank liquidity creation: the Taylor rule Taylor-type rules are widely used to explain how monetary policy has been set in the past and how it should be set in the future (e.g., Judd and Rudebusch 1998, Carlstrom and Fuerst 2003). This raises an interesting question: would aggregate liquidity creation have been significantly different if the Federal Reserve had used a Taylor rule? As a first step, in each quarter we estimate the Taylor rule federal funds rate, i.e., the federal funds rate that would have obtained if the Federal Reserve had used a Taylor rule. Taylor (1993) argues that the federal funds rate should be set as follows: where Infl is the four-quarter rate of inflation; NatFedFunds is the natural real federal funds rate (a rate that is consistent with full employment), which Taylor assumed to be 2%; TargetInfl is the target inflation rate, which also equals 2% by assumption; and OutputGap is the percent deviation of potential GDP from 19