Vol. 5, No. 4 MAY 2010­­ EconomicLetter

Insights from the F e d e r a l R eserve Bank o f D a l l a s

The Term Facility’s Effectiveness in the Financial Crisis of 2007–09 by Tao Wu

In the second half of 2007, financial turmoil swept over the U.S. and The TAF and other other major economies. Triggered by a subprime mortgage meltdown, the crisis lending facilities quickly spread to other major financial markets and precipitated the worst eco- established during the nomic downturn since the Great Depression. Although financial and economic crisis were an conditions have improved substantially, the wrenching episode’s effects are still experiment that with us. proved effective in During the crisis, financial markets experienced tremendous strains, addressing severe and the cost of short-term funding rose sharply. The gap between the three- financial turbulence. month unsecured London interbank offered rate (Libor) and the overnight in-

dexed swap (OIS) rate is frequently used as a measure of tensions in interbank

money markets. As the crisis began in early August, this spread jumped from

less than 0.1 percentage point to almost 1 percentage point within a month. At

the crisis’s peak in September 2008—just after the collapse of Lehman Brothers and the rescue of American Interna- lending market’s breakdown rendered tional Group—the gap soared to an those tools inadequate for addressing unprecedented 3.7 percentage points. the unusual financial market pressures The spikes in the Libor–OIS gap during this crisis. testify to the severity of a crisis that Open-market operations are the posed serious challenges to central Fed’s most powerful and frequently banks around the world. In response, used policy tool, providing overnight several of them created new lending credits at the rate. Every To find out whether the facilities to quickly provide liquidity day, the Fed trades on the open mar- to the banking sector and improve ket with a select group of primary liquidity facilities have market functioning. The list includes dealers, directly buying or selling the , Bank of Treasury or government agency securi- been working as England, Bank of Canada and Swiss ties or repurchase agreements against National Bank. On Dec. 12, 2007, the such securities. intended, it’s important established its ver- In normal times, primary deal- sion—the term auction facility (TAF). ers distribute the liquidity increases to to understand the Researchers have yet to reach other financial institutions through the a consensus on the effectiveness of interbank money market, increasing nature of the heightened such facilities. This Economic Letter, the flow of credit to the overall econo- based on a recent study, provides an my. In turbulent times, however, finan- strains on interbank econometric evaluation of whether the cial institutions are reluctant to lend to TAF helped relieve strains in the U.S. each other, and the channel can clog. money markets. money market.1 The findings reveal The gives the that the TAF has reduced liquidity risk Fed an alternative means of adding premiums paid by banks; however, it liquidity. Depository institutions in has been less effective in cutting coun- sound condition can obtain fully col- terparty risk premiums. lateralized overnight loans at an inter- est rate that’s usually higher than the The Financial Crisis federal funds rate. The subprime mortgage market’s From 2003 to the summer of 2007, growing problems began to draw the discount rate had been 100 basis attention in early 2007. However, it points above the federal funds target. took several months for the financial After the initial jump in money-market crisis to spread to money markets. On interest rates, the Fed narrowed the Aug. 9, 2007, French investment bank discount rate premium to 50 basis BNP Paribas halted redemptions from points on Aug. 17, 2007, and to 25 three of its subsidiary mutual funds, basis points on March 17, 2008. The and in response, overnight interest terms of discount window loans were rates shot up in Europe and the U.S. extended to up to 30 days in August The interbank money market 2007 and later to 90 days. The Fed is the main gateway for commercial also made the loans renewable at the banks to quickly obtain funding to request of the borrowers. make loans. Deteriorating condi- These measures were taken to tions greatly impaired the stability of encourage banks to use the discount this critical short-term funding and window, but the effects were modest posed severe challenges to central due to the so-called stigma problem. banks’ ability to provide ample liquid- During a financial crisis, banks may be ity through regular reluctant to borrow from the discount channels. window because of concerns that Under normal circumstances, the markets would interpret it as a sign of Fed injects liquidity into the economy financial weakness. The stigma might by two means—conducting open-mar- damage their reputations, lower their ket operations and lending at the dis- market values and reduce their ability count window. However, the interbank to borrow in the market.

EconomicLetter 2 Federal Reserve Bank of Dallas Federal Reserve Bank of Dallas EconomicLetter As strains in money markets persisted and worsened in early Chart 1 December 2007, the Fed established a new lending alternative—the term Libor–OIS Spread Widens in 2008, Raising Doubts auction facility. Through this facil- About TAF ity, the Fed auctioned preannounced amounts of credit, twice a month, Percentage points, annualized to eligible depository institutions in 4.5 Failure of G7 action, sound financial condition for a term of Lehman Lehman CDS 4 2 Brothers; AIG settlement one month instead of overnight. The (Sept. 15–16) (Oct. 10) 3.5 TAF accepted the same kinds of collat- TAF eral as the discount window. established 3 (Dec. 12) The TAF was initially set at $20 Bear Stearns 2.5 billion for each auction. It was gradual- bailout (Mar. 17) ly increased to $150 billion in January 2 2009 before it was scaled back. The 1.5 final auction was held March 8, 2010. After the TAF’s establishment, 1

credit conditions in the interbank .5 market improved significantly. The three-month Libor spread over the OIS 0 2006 2007 2008 2009 rate dropped sharply from more than 1 percentage point in early December SOURCE: Haver Analytics. 2007 to less than 0.3 percentage point in late January 2008 (Chart 1). However, the spread widened again to sheets is likely to become more costly of the price per unit of risk and the about 0.8 percentage point in spring and harder to obtain. Fund managers perceived amount of risk—should also 2008. may also demand that extra liquidity decline. As macroeconomic and financial be readily available to cover potential The final channel entailed offering market conditions worsened substan- redemptions.4 other readily available funding sources tially in the second half of 2008, the Through lending facilities to pro- to discourage banks from excessively spread jumped. The upswing and the vide credit to financial institutions in hoarding liquidity purely out of indi- later surge in Libor spreads raised need, the Fed and other central banks vidual precautionary concerns. doubts about the new liquidity facil- sought to relieve financial strains These channels provided reason ity’s effectiveness.3 through several channels. to believe the TAF and other liquidity To find out whether a liquid- The first and most direct channel facilities might alleviate financial strains ity facility is working as intended, it’s involved providing additional funding in the interbank money market. important to understand the nature of to banks in immediate need of liquid- the heightened strains on interbank ity, lowering short-term borrowing Quantifying TAF’s Effects money markets. During financial stress, costs.5 Money market strains come from banks become increasingly reluctant to The second channel focused on both the larger demand for liquidity lend to each other for two reasons. reducing the pressure on banks to during a financial crisis and heightened First, counterparty risk—the pos- liquidate assets, helping counteract counterparty risk. I first examine the sibility that the institution on the other upward pressure on banks’ funding TAF’s effect by addressing these two side of the transaction may default— costs from deterioration in money mar- concerns separately, and then quantify increases with the uncertainty about ket conditions. All else equal, this may the TAF’s overall effect. banks’ financial conditions. contribute to a decline in counterparty Reducing liquidity premiums. Second, banks tend to build up risk. I focus on examining the TAF’s effect precautionary liquidity to guard against The third channel centered on in relieving banks’ liquidity concerns mounting uncertainty about the mar- strengthening confidence so that inves- by controlling for the variation in sys- ket value of their assets—for instance, tors would demand less compensation tematic counterparty risk premiums. various structured credit products. In for a given unit of risk—i.e., the price Because measures of these premiums times of financial stress, funding to of risk may decline in the TAF’s pres- aren’t readily available, I construct one keep these assets on banks’ balance ence. The risk premium—the product based on the observed credit default

EconomicLetter Federal Reserve Bank of Dallas Federal Reserve Bank of Dallas 3 EconomicLetter dollar Libor survey and use it as a Chart 2 proxy for the major banks’ systematic default risk premiums. Credit Default Swap Premiums Gyrate During Crisis Chart 2 displays this systematic counterparty risk factor along with the Percentage points, annualized individual CDS rates of several major 5 banks included in the Libor survey.

4.5 The constructed indicator captures the variations of individual CDS rates quite 4 Citigroup Bank of America well, showing waves of volatility dur- 3.5 Libor CDS factor ing the crisis. 3 Counterparty Risk Index I then regress the Libor spreads J.P. Morgan 2.5 HSBC on the constructed systematic counter- party default risk factor and on a dum- 2 my variable that accounts for the TAF’s 1.5 creation on Dec. 12, 2007. Estimation 1 results from the two regressions indi-

.5 cate that the TAF had effectively low- ered major banks’ liquidity concerns by 0 2006 2007 2008 2009 decreasing the three-month Libor–OIS spread about 26 basis points (Table NOTES: The Libor CDS factor summarizes the five-year rates for the 16 banks in the Libor U.S. dollar survey. The Counterparty Risk Index is based on the average credit spread of five-year credit default swap contracts traded by 15 major financial firms. 1). This result is consistent with James SOURCES: Bloomberg; author’s calculations. McAndrews, Asani Sarkar and Zhenyu Wang’s estimates of TAF effects using different model specification.6 Counterparty default risk pre- Table 1 miums. Next, I evaluate the TAF’s effect on reducing systematic coun- TAF Lowers Liquidity Premiums terparty default risk premiums. These premiums may depend on a variety Three-month Libor–OIS spread of fundamental macroeconomic and financial variables—in particular, Regression 1 Regression 2 aggregate risks in the macroeconomy Counterparty default risk 1.0804** .0677** and financial markets. To capture this (8.9614) (3.1736) variable, I incorporate three measures TAF dummy –.2605** –.0448** of aggregate risk (Chart 3): (–2.8216) (–3.5727) • Merrill Lynch’s Merrill Option Lag of Libor–OIS spread .9697** Volatility Estimate (MOVE) Index to (56.2749) track the implied volatility in the lon- 2 Adj. R .6037 .9881 ger-term U.S. Treasury market • The Chicago Board Options NOTES: t-statistics are displayed in parentheses (based on Newey–West standard errors). * (**) denotes statistical signifi- cance at the 5 percent (1 percent) level. Exchange Volatility Index (VIX) mea- SOURCE: Author’s estimates. sure of implied volatility from options on the S&P 500 index to gauge the uncertainty in the stock market swap (CDS) rates of major financial ance premium, and a higher CDS rate • The implied volatility from firms. means the market perceives a higher three-month Eurodollar futures options A CDS is a contract insuring risk of default for the company. to measure uncertainty regarding the against the default risk of a specific The CDS market quotes rates only near-term path of monetary policy company. The CDS buyer makes peri- for individual companies. To obtain a Given the central role of subprime odic payments to its seller, receiving measure of overall or systematic coun- mortgages in the most recent financial full compensation for losses if the terparty risk, I extract the first principal crisis, it’s necessary to control for their company defaults on its debt. The component of the individual five-year risks to more accurately determine the CDS rate can be viewed as an insur- CDS rates for all 16 banks in the U.S. TAF’s effect. However, there are few

EconomicLetter 4 Federal Reserve Bank of Dallas Federal Reserve Bank of Dallas EconomicLetter Chart 3 Evaluating Aggregate Risks in the Microeconomy and Financial Markets

Implied Volatility on Longer-Term Treasury Securities (MOVE Index) Implied Volatility on S&P 500 (VIX Index)

Percentage points, annualized Percentage points, annualized 3 90

80 2.5 70

2 60

50 1.5 40

1 30

20 .5 10

0 0 2006 2007 2008 2009 2006 2007 2008 2009

Implied Volatility on Three-Month-Ahead Eurodollar Futures Options Average Delinquency Rate of Residential Mortgage Loans

Percentage points, annualized Percentage points

2 8

1.8 7 1.6 6 1.4 5 1.2

1 4

.8 3 .6 2 .4 1 .2

0 0 2006 2007 2008 2009 2006 2007 2008

Systematic CDS Factor Among Mortgage-Related Firms

Percentage points, annualized 18

16

14

12

10

8

6

4

2

0 2006 2007 2008 2009

SOURCES: Bloomberg; author’s calculations.

EconomicLetter Federal Reserve Bank of Dallas Federal Reserve Bank of Dallas 5 EconomicLetter readily available measures of these spreads, I construct a second mortgage mortgage risks. For this reason, I use default risk factor with a procedure two alternative measures. similar to the one used for the vari- The first is the seasonally adjusted able that tracks systematic counterparty average delinquency rate of residen- default risk. tial mortgage loans owned by the 100 This mortgage default risk factor largest U.S. commercial banks. This is defined as the first principal com- measures the portion of loans past due ponent of individual CDS rates for a The TAF was designed for 30 days or more on these banks’ group of the largest subprime mort- balance sheets. A substantial rise in the gage lenders, mortgage bond insurers to alleviate liquidity delinquency rate would endanger the and residential construction companies. banks’ health and increase the prob- These firms represent sectors most premiums rather than ability of default. heavily exposed to subprime mortgage The average delinquency rate had market turmoil but have no access to address the insolvency been below 2 percent until first quarter the TAF. Therefore, their CDS rates are 2007, when it began to rise sharply, an ideal measure of financial markets’ risk that is reflected more than tripling to 6.9 percent in perception of the underlying mortgage fourth quarter 2008. Since the onset default risk. in credit default of the financial crisis, a substantial rise Table 2 reports the TAF’s impact in the delinquency rate has preceded on CDS spreads of banks active in swap premiums. almost every major spike in Libor–OIS the Libor market. Results from two spreads and counterparty default risk regressions suggest that uncertainties premiums, confirming that the wors- reflected in the Treasury bond market ening mortgage situation was a major and mortgage default risk premiums driver in the crisis. are closely related to financial strains However, the delinquency rate is in the Libor market. Uncertainties available only quarterly with a six- to about the stock market and near-term seven-week lag. To model daily Libor monetary policy actions have far less effect on the counterparty default risk premiums. At the same time, the TAF’s Table 2 effect on counterparty default risk TAF Offers Limited Relief from Counterparty Default Risk premiums is negligible. This indicates that the facility has been unable to significantly reduce counterparty Systematic counterparty default risk factor default risk premiums among major Regression 1 Regression 2 commercial banks. The TAF, however, was designed MOVE .5600** .4342** to alleviate liquidity premiums rather (4.9254) (3.4536) than address the insolvency risk that VIX –.0032 –.0051 (–.9792) (–1.3582) is reflected in CDS premiums. As McAndrews, Sarkar and Wang Eurodollar volatility .1254 .5523* (.4651) (2.3049) point out, “The TAF is not expected Delinquency rate .1638** to exert large or immediate effects in (3.4940) reducing credit risks of banks. Credit Mortgage default risk factor .0458** risks are largely determined by banks’ (3.4059) earnings and asset value. In the cur- TAF dummy .0977 .1637 rent situation, it is likely that changes (0.9875) (1.5500) in asset values are the driving force Adj. R2 .8744 .8936 for the credit risk of banks. “Much of the change in banks’ NOTES: t-statistics are displayed in parentheses (based on Newey–West standard errors). * (**) denotes statistical signifi- cance at the 5 percent (1 percent) level. asset values is determined by the SOURCE: Author’s estimates. valuation of mortgages and related financial products. Since the valua-

EconomicLetter 6 Federal Reserve Bank of Dallas Federal Reserve Bank of Dallas EconomicLetter tion of mortgages is determined by ket can’t be attributed to heightened the homeowners’ long-term ability to uncertainty about the stock market or pay for their debt, there is no reason about the near-term course of mon- to expect the TAF to affect the value etary policy. of banks’ mortgage and other assets.”7 Finally, estimation results again TAF’s overall effect. Quanti- reveal that the TAF has a substan- fying the TAF’s broad impact on tial and statistically significant effect reducing money market strains com- in narrowing the Libor–OIS spread. bines the lowering of both liquidity Controlled for various macroeco- Evidence indicates and counterparty risk premiums. For nomic and financial volatility and this purpose, I regress the three- risk measures, the results show that that the term auction month Libor–OIS spread on the the presence of the TAF on average variables measuring macroeconomic reduced the three-month Libor–OIS facility was effective and financial-market volatilities, the spread by 50 or 55 basis points, mortgage default risks and the TAF depending on the proxy of mortgage in reducing liquidity dummy. risks used in the regression. The estimation’s results show The high adjusted R² values—85 risk premiums paid that the implied volatility of longer- percent even without including CDS term Treasury securities (MOVE) has rates or the lagged Libor spread in by banks. a substantially positive and statisti- the regression—indicate that these cally significant effect on the Libor volatility and risk measures along spreads (Table 3). A 1 percentage- with the TAF dummy account for point increase in the MOVE index most of the variation in Libor spreads. tends to increase the three-month Libor–OIS spread by almost 1 per- TAF in Retrospect centage point. Facing potentially dire conse- Heightened mortgage risks have quences from the financial crisis, also contributed substantially to several central banks established jumps in Libor spreads in the past two years. For instance, a 1 percent- age-point rise in the delinquency rate tends to increase the three-month Table 3 Libor spread by 24 basis points, and Overall TAF Effects Significant a 1 percentage-point rise in the CDS rates of mortgage firms increases the Libor spread by 6 basis points. Three-month Libor–OIS spread From mid-2007 to fourth quarter Regression 1 Regression 2 2008, the delinquency rate rose about MOVE .9833** .9904** 4.5 percentage points, and mortgage (5.3512) (5.8595) CDS rates rose from less than 1 per- VIX .0083 .0109 cent to a peak of 16 percent, both (1.3181) (1.6506) suggesting that heightened mortgage Eurodollar volatility –.4092 –.2618 risk alone has increased the Libor (–1.1172) (–1.0616) spread by about 1 percentage point. Delinquency rate .2371** Such an effect is significant at the 1 (5.2418) percent level. Mortgage default risk factor .0559** The implied volatilities on the (3.9497) S&P 500 (VIX) and three-month TAF dummy –.5492** –.4879** Eurodollar futures options have much (–3.7547) (–3.4788) smaller influences on the Libor–OIS Adj. R2 .8501 .8477

spread, and in most specifications, NOTES: t-statistics are displayed in parentheses (based on Newey–West standard errors). * (**) denotes statistical signifi- their coefficient estimates are statisti- cance at the 5 percent (1 percent) level. cally insignificant. This suggests that SOURCE: Author’s estimates. increased strains in the money mar-

EconomicLetter Federal Reserve Bank of Dallas Federal Reserve Bank of Dallas 7 EconomicLetter EconomicLetter is published by the Federal Reserve Bank of Dallas. The views expressed are those of the authors and should not be attributed to the Federal Reserve Bank of Dallas or the Federal liquidity facilities designed to reduce Notes Reserve System. financial stresses on the interbank 1 The study used is “The U.S. Money Market and Articles may be reprinted on the condition that money market. The Fed’s version was the Term Auction Facility in the Financial Crisis of the source is credited and a copy is provided to the Research Department of the Federal Reserve Bank of the term auction facility. 2007–2009,” by Tao Wu, The Review of Econom- The TAF appears to have had Dallas. ics and Statistics, forthcoming. Economic Letter is available free of charge only a limited effect in reduc- 2 Occasionally, the Federal Reserve also conducts by writing the Public Affairs Department, Federal ing counterparty risk premiums. in anticipation of special needs for Reserve Bank of Dallas, P.O. Box 655906, Dallas, TX However, evidence indicates that liquidity. For instance, several forward auctions 75265-5906; by fax at 214-922-5268; or by telephone at 214-922-5254. This publication is available on the the facility was effective in reduc- were conducted to ease year-end pressures on Dallas Fed website, www.dallasfed.org. ing liquidity risk premiums paid by the demand for liquidity in late 2008. The matu- banks. rity of the TAF loans in such cases can vary and is Estimates indicate that the pres- sometimes as long as 85 days. ence of the TAF lowered the three- 3 See, for example, “A Black Swan in the Money month Libor–OIS spread by 50 or 55 Market,” by John B. Taylor and John C. Williams, basis points during the crisis of 2007– American Economic Journal: Macroeconomics, 09, mainly by addressing concerns vol. 1, issue 1, 2009, pp. 58–83. about the banking sector’s liquidity. 4 See note 1. These are concerns over funding In 2010, the crisis has abated liquidity; that is, the risk that an institution may be and financial markets are functioning unable to raise cash to maintain its balance-sheet normally. The TAF and other lend- position. These are in contrast to concerns over ing facilities established during the trading liquidity, which refers to banks’ difficulties crisis were an experiment that proved executing transactions at the prevailing market effective in addressing severe finan- price due to a temporary lack of appetite for the cial turbulence, and similar facilities transactions by other traders on the market. can be a useful part of the Federal 5 See note 1. Richard W. Fisher Reserve’s tool kit in the event of 6 See “The Effect of the Term Auction Facility on President and Chief Executive Officer future crises. the London Interbank Offered Rate,” by James McAndrews, Asani Sarkar and Zhenyu Wang, Helen E. Holcomb Wu is a senior economist and advisor in the Federal Reserve Bank of New York, Staff Report First Vice President and Chief Operating Officer Research Department of the Federal Reserve Bank no. 335, July 2008. Harvey Rosenblum of Dallas. 7 See note 6. Executive Vice President and Director of Research

Robert D. Hankins Executive Vice President, Banking Supervision

Director of Research Publications Mine Yücel

Executive Editor Jim Dolmas

Associate Editor Kathy Thacker

Graphic Designer Ellah Piña

Federal Reserve Bank of Dallas 2200 N. Pearl St. Dallas, TX 75201