9512.net
甜梦文库
当前位置:首页 >> >>

The subprime credit crisis and contagion in financial markets


ARTICLE IN PRESS
Journal of Financial Economics 97 (2010) 436–450

Contents lists available at ScienceDirect

Journal of Financial Economics
journal homepage: www.elsevier.com/locate/jfec

The subprime credit crisis and contagion in ?nancial markets$
Francis A. Longstaff
UCLA Anderson School and NBER, USA

a r t i c l e i n f o
Article history: Received 12 December 2008 Received in revised form 22 September 2009 Accepted 29 November 2009 Available online 25 January 2010 JEL classi?cation: G01 G12 G14 Keywords: Contagion Asset-backed securities Subprime CDOs Liquidity Toxic assets

abstract
I conduct an empirical investigation into the pricing of subprime asset-backed collateralized debt obligations (CDOs) and their contagion effects on other markets. Using data for the ABX subprime indexes, I ?nd strong evidence of contagion in the ?nancial markets. The results support the hypothesis that ?nancial contagion was propagated primarily through liquidity and risk-premium channels, rather than through a correlated-information channel. Surprisingly, ABX index returns forecast stock returns and Treasury and corporate bond yield changes by as much as three weeks ahead during the subprime crisis. This challenges the popular view that the market prices of these ‘‘toxic assets’’ were unreliable; the results suggest that signi?cant price discovery did in fact occur in the subprime market during the crisis. & 2010 Elsevier B.V. All rights reserved.

1. Introduction During the past three years, ?nancial markets have suffered catastrophic losses. These were originally triggered by the threat of massive defaults by subprime borrowers in the mortgage markets. The resulting subprime crisis of 2007 led rapidly to massive declines in the market values of large portfolios of highly rated assetbacked securities (ABS) held by many ?nancial institutions. In addition, the subprime crisis brought about an almost complete halt to the ?edgling structured-credit
$ I am very grateful for helpful discussions with Joshua Anderson, Vineer Bhansali, Bruce Carlin, Richard Clarida, Rajna Gibson, Robert Gingrich, Anil Kayshup, Hanno Lustig, Alfred Murata, Steve Schulist, and Jiang Wang, and for the comments of seminar participants at Barclays Global Investors, the CFA Institute Conference, Claremont McKenna College, New York University, PIMCO, the University of Colorado, and UCLA. I am particularly grateful for the comments and suggestions of the editor Bill Schwert and two anonymous referees. All errors are my responsibility. E-mail address: francis.longstaff@anderson.ucla.edu

market, a serious credit crunch for both individuals and ?nancial institutions, and a major decline in the liquidity of debt securities in virtually every market. In 2008, the subprime crisis spilled over and became the catalyst for a much broader global ?nancial crisis. During the year, the markets reeled from the collapse or forced mergers/bailouts of Bear Stearns, AIG, Fannie Mae, Freddie Mac, Lehman Brothers, IndyMac Bank, Merrill Lynch, Wachovia, Washington Mutual, and many others. Concerns about the long-term ?nancial viability of the U.S. Treasury, which has provided an unprecedented amount of liquidity, capital, and ?nancial guarantees to the market, has resulted in credit default swaps on the U.S. Treasury trading at spreads as high as 100 basis points. Much of the intervention by the Treasury and the Federal Reserve in the ?nancial markets has been motivated by the objective of avoiding broader contagion and spillovers to other markets and sectors of the economy. Understanding the nature of contagion in ?nancial markets is of fundamental importance and there is an

0304-405X/$ - see front matter & 2010 Elsevier B.V. All rights reserved. doi:10.1016/j.j?neco.2010.01.002

ARTICLE IN PRESS
F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450 437

extensive literature addressing its causes and effects. Important recent papers on contagion include Allen and Gale (2000), Kyle and Xiong (2001), Kodres and Pritsker (2002), Kiyotaki and Moore (2002), Kaminsky, Reinhart, and Vegh (2003), Allen and Gale (2004), Brunnermeier and Pedersen (2005, 2009), and many others. From a research perspective, the crisis in the subprime assetbacked market provides a near-ideal ‘‘laboratory’’ for studying the role that contagion may play in ?nancial markets when an asset class becomes severely distressed.1 The contagion literature identi?es at least three possible mechanisms by which shocks in one market may spill over into other markets. First, Kiyotaki and Moore (2002), Kaminsky, Reinhart, and Vegh (2003), and others describe mechanisms in which negative shocks in one market represent the arrival of economic news that directly affects the collateral values or cash ?ows associated with securities in other markets. In this mechanism, contagion can be viewed as the transmission of information from more-liquid markets or markets with more rapid price discovery to other markets. Second, Allen and Gale (2000), Brunnermeier and Pedersen (2009), and others show how investors who suffer losses in one market may ?nd their ability to obtain funding impaired, potentially leading to a downward spiral in overall market liquidity and other asset prices via a ‘‘?ight to quality.’’ In this mechanism, contagion occurs through a liquidity shock across all markets. Third, Vayanos (2004), Acharya and Pedersen (2005), Longstaff (2008), and others imply that a severe negative shock in one market may be associated with an increase in the risk premium in other markets. In this mechanism, contagion occurs as negative returns in the distressed market affect subsequent returns in other markets via a time-varying risk premium. The objective of this paper is to shed some light on the mechanisms involved in ?nancial contagion by studying the subprime asset-backed collateralized debt obligation (CDO) market during the 2006–2008 period and exploring how negative shocks affected other markets as the subprime crisis of 2007 unfolded and then evolved into the global ?nancial crisis of 2008. The study is based on an extensive data set of prices for the ABX indexes of subprime mortgage-related asset-backed CDOs. Using a vector autoregression (VAR) framework, I examine the extent to which ABX returns are related to returns in other ?nancial markets as well as to market leverage and trading activity measures. Several key results emerge from this analysis. First, despite the lower liquidity of the asset-backed CDO market, I ?nd that ABX index returns developed signi?cant predictive ability (Granger causality) for subsequent stock market returns, Treasury yield changes, corporate

1 Important papers focusing on the valuation of distressed assets include Shleifer and Vishny (1992), Asquith, Gertner, and Scharfstein (1994), Opler and Titman (1994), Clark and Ofek (1994), John and Ofek (1995), Andrade and Kaplan (1998), Pulvino (1998), Kahl (2002), Longstaff (2004), Vayanos (2004), Acharya and Pedersen (2005), Brunnermeier and Pedersen (2005), Carlin, Lobo, and Viswanathan (2007), and Longstaff and Myers (2009).

bond spread changes, and changes in the VIX volatility index as the 2007 subprime crisis unfolded. In fact, ABX returns have signi?cant forecast power for stock returns, Treasury yield changes, corporate yield spread changes, and changes in the VIX as far as three weeks ahead. Treasury bond prices increase in response to negative shocks to asset-backed CDO values, consistent with a ?ight-to-quality pattern. This effect, however, is much stronger for short-term Treasury bonds than for longerterm Treasury bonds. In contrast, negative shocks to the ABX indexes map into signi?cant subsequent negative returns for the Standard and Poors (S&P) 500 index as well as for the subset of ?nancial ?rms in the S&P 500. Thus, I ?nd strong evidence of contagion in the ?nancial markets during the 2007 subprime crisis. Second, I ?nd that this forecast ability dissipates during 2008 as the subprime crisis gave way to the broader global ?nancial crisis. Thus, contagion appeared to spread from the ABX market at the beginning of the crisis when subprime losses were the primary concern. After concerns about a meltdown of the general ?nancial markets and the potential for a global depression became widespread in 2008, however, the ABX market no longer functioned as a vector of contagion (and no longer Granger-caused returns) in other markets. Intuitively, this is consistent with the usual view of contagion as a major shock or event in which there is a signi?cant but temporary increase in the linkages between different ?nancial markets. Taken together, these results provide a number of important insights about the nature of the mechanisms driving contagion across markets in the present crisis. For example, ?nding that shocks tended to be transmitted with a lag from the less-liquid ABX index market to the highly liquid stock and Treasury bond markets argues against a correlated-information view of ?nancial contagion. We would expect price effects to be contemporaneous in the highly liquid stock and Treasury bond markets if contagion was due to correlated information. Thus, the results (which, of course, are limited to the speci?c episode studied) appear to be more consistent with either the liquidity-induced contagion mechanisms presented by Allen and Gale (2000), Kodres and Pritsker (2002), and Brunnermeier and Pedersen (2005), or the risk-premium contagion mechanisms implied by Vayanos (2004), Acharya and Pedersen (2005), and Longstaff (2008). To explore this latter implication in more depth, I again use a VAR framework to explore the relation between ABX index returns and various measures of market activity, liquidity, and funding availability. I ?nd that shocks in the ABX market have signi?cant predictive power for trading activity in ?nancial stocks, trading disruptions in the ?xed-income markets, and the availability of short-term asset-backed ?nancing during the crisis. These results reinforce the view that market- and funding-liquidity effects were a major factor in the transmission of contagion during the subprime crisis. The remainder of this paper is organized as follows. Section 2 brie?y reviews the literature on contagion in ?nancial markets. Section 3 provides an introduction to

ARTICLE IN PRESS
438 F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

the asset-backed CDO market. Section 4 describes the ABX indexes and the other data used in the study. Section 5 presents the empirical test for contagion. Section 6 examines the implications of the subprime crisis for market liquidity. Section 7 summarizes the results and presents concluding remarks. 2. Contagion in ?nancial markets The literature on contagion in ?nancial markets is far too extensive to review fully here. Kindleberger (1978), Dornbusch, Park, and Claessens (2000), and Kaminsky, Reinhart, and Vegh (2003), however, provide excellent surveys. Generally, this literature has focused on contagion effects across countries. Contagion, however, is possible in virtually any set of ?nancial markets. In this section, I will simply summarize some of the key implications of the contagion literature for the behavior of security prices during periods of extreme market distress. Following Dornbusch, Park, and Claessens (2000), Kaminsky, Reinhart, and Vegh (2003), Bae, Karolyi, and Stulz (2003), and many others, I adopt a working de?nition of ?nancial contagion as an episode in which there is a signi?cant increase in cross-market linkages after a shock occurs in one market. The literature identi?es at least three major channels by which contagion effects can be propagated through different ?nancial markets. The ?rst channel can be termed the correlatedinformation channel. In this mechanism, a shock to one ?nancial market signals economic news that is directly or indirectly relevant for security prices in other markets. Note that this could be consistent with the revelation of information about economic factors affecting multiple markets. For example, Dornbusch, Park, and Claessens (2000) describe direct effects occurring through fundamentals such as trade links. Kiyotaki and Moore (2002) describe a balance-sheet channel in which losses in one market translate into declines in the equity of other ?rms holding the distressed assets. King and Wadhwani (1990) present a model in which contagion occurs as rational agents attempt to infer information from price changes in other markets. A common implication throughout the correlated-information literature is that contagion occurs rapidly via the price-discovery process. Thus, this channel should result in immediate price effects in the markets affected by the distress event, particularly when these markets are more liquid than the market in which the original distress event occurs. This implication of the correlated-information contagion mechanism can be directly tested using a VAR framework.2 The second channel can be designated the liquidity channel. In this mechanism, a shock to one ?nancial
2 This argument is clearly predicated on the assumption that markets are informationally ef?cient. If it takes an extended period to incorporate information revealed in one market into other markets, then it will clearly be more dif?cult to differentiate effects of the correlatedinformation channel from those implied by other channels. I am grateful to the referee for this insight.

market results in a decrease in the overall liquidity of all ?nancial markets. In turn, this may affect investor behavior and asset prices. For example, Allen and Gale (2000) present a model in which banks have cross holdings of deposits across regions. In this model, ?nancial shocks cause banks to liquidate these cross holdings, thereby denying liquidity to other regions. Kodres and Pritsker (2002) present a model in which contagion occurs as losses in one market force economic agents to either liquidate leveraged positions or to rebalance their portfolios in response. Brunnermeier and Pedersen (2009) argue that agents who experience losses in one market may ?nd their ability to obtain funding impaired, which would then result in declines in the liquidity of the other ?nancial assets in the markets. A key implication of this liquidity-related channel of contagion is that a distress event may be associated with subsequent declines in the availability of credit and increases in trading activity in other markets. Note that this spiraling mechanism might play out over an extended period. The third channel can be termed the risk-premium channel. In this mechanism, ?nancial shocks in one market may affect the willingness of market participants to bear risk in any market. Thus, prices in all markets may be affected as equilibrium risk premia adjust in response. For example, Vayanos (2004) and Acharya and Pedersen (2005) present models in which shocks such as those that might result from a distress event translate into major changes in the equilibrium risk premia of assets in the economy. An important implication of this time variation in risk premia is that return shocks to the distressed security may be predictive for the subsequent returns of other assets. This follows because when the risk premium for an asset increases during the current period, it also impacts the distribution of future asset returns. In turn, this feedback effect can induce predictability into the time series of realized asset returns. These contagion channels all have different implications for the behavior of security prices across markets when a distress event occurs. It is important to note, however, that there may also be similarities between the different channels.3 I will explore the empirical implications of the various channels later in the paper.

3. The subprime asset-backed CDO market In the current crisis, tranches or CDOs based on the cash ?ows of portfolios of subprime home-equity loans were originally the major source of credit losses for many ?nancial institutions. Accordingly, I focus primarily on these securities throughout this study. This section
3 For example, there is clearly a relation between credit risk and liquidity. In fact, a signi?cant factor during the subprime crisis of 2007 may have been credit-risk-induced illiquidity as investors were leary of taking positions in complex mortgage-related securities. On the other hand, an important factor in the global ?nancial crisis of late 2008 may have been illiquidity-induced credit risk as major ?nancial institutions faced default because they were unable to liquidate positions and collateralize their liabilities. I am grateful to the referee for pointing out this issue.

ARTICLE IN PRESS
F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450 439

Table 1 Countrywide Home Loans, Inc. subprime ABS CDO structure CWABS 2006-1. This table reports some of the contractual terms for the subprime ABS CDO structure issued by Countrywide Home Loans, Inc. through Lehman Brothers in February 2006. The issuing entity is designated as CWABS Asset-Backed Certi?cates Trust 2006-1. Of the total notional amount underlying the CDO, approximately $500 million is based on subprime ?xed-rate mortgages while $400 million is based on subprime ?oating-rate mortgages. The L in the initial pass-through rate represents one-month LIBOR. The seniority ranking n=m means that the tranche’s seniority is n-th out of m tranches. Tranche Notional amount Price to public 100.0000 99.9995 99.9998 99.9985 99.9987 99.9980 99.9981 99.9972 99.9965 99.4627 98.9985 98.5371 100.0000 100.0000 100.0000 100.0000 100.0000 100.0000 100.0000 100.0000 100.0000 100.0000 Underwriter fee 0.0521 0.1042 0.1563 0.2500 0.3333 0.4167 0.4167 0.5000 0.5833 0.8333 1.0000 1.2500 0.0522 0.1033 0.1033 0.4167 0.5000 0.8333 0.9167 0.9667 1.0000 1.0833 Initial pass-through rate L + 0.130% 5.281% 5.384% 5.714% 5.884% 5.526% 5.917% 6.016% 6.115% 6.200% 6.200% 6.200% L + 0.080% L +0.190% L + 0.300% L +0.390% L +0.410% L +0.440% L +0.560% L + 0.600% L +0.660% L +1.300% Maturity Initial Moody’s rating Aaa Aaa Aaa Aaa Aaa Aaa Aa1 Aa2 Aa3 A1 A2 A3 Aaa Aaa Aaa Aa1 Aa2 Aa3 A1 A2 A3 Baa1 Initial S&P rating AAA AAA AAA AAA AAA AAA AA+ AA+ AA AA ? A+ A AAA AAA AAA AA+ AA+ AA AA ? A+ A A Seniority ranking

AF-1 AF-2 AF-3 AF-4 AF-5 AF-6 MF-1 MF-2 MF-3 MF-4 MF-5 MF-6 AV-1 AV-2 AV-3 MV-1 MV-2 MV-3 MV-4 MV-5 MV-6 MV-7

147,232,000 22,857,000 90,995,000 21,633,000 38,617,000 44,200,000 13,260,000 12,155,000 7,293,000 6,409,000 6,188,000 5,525,000 139,560,000 115,712,000 25,042,000 14,320,000 13,067,000 7,518,000 6,802,000 6,802,000 5,907,000 5,549,000

Nov May Jul Sep Jul May May May Apr Apr Mar Feb Jul May Jun May May May Apr Apr Mar Mar

2025 2027 2033 2034 2036 2036 2036 2036 2036 2036 2036 2036 2028 2035 2036 2036 2036 2036 2036 2036 2036 2036

1/7 1/7 1/7 1/7 1/7 1/7 2/7 3/7 4/7 5/7 6/7 7/7 1/8 1/8 1/8 2/8 3/8 4/8 5/8 6/8 7/8 8/8

provides a brief introduction to the asset-backed CDO market. Large quantities of subprime asset-backed CDOs were issued during the past several years and were widely viewed as one of the most important ?nancial innovations of the past decade. According to the Securities Industry and Financial Markets Association, the total U.S. issuance of asset-backed securities during the 2005–2008 period was $2.154 trillion, and the total U.S. issuance of CDOs during the same period was $987 billion. Asset-backed tranches or CDOs share many features in common with CDOs for corporate bonds. As described in Longstaff and Rajan (2008) and Bhansali, Gingrich, and Longstaff (2008), a CDO is created by an issuer ?rst forming a portfolio of loans, either by lending money directly, or by buying debt securities in the marketplace.4 In the ABS market, these loans could consist of ?rst mortgages, second mortgages, loans on manufactured homes, credit card receivables, auto loans, student loans, and even account receivables.5 Once the portfolio is formed, the CDO issuer sells tranches based on the cash ?ows scheduled to be generated by the underlying loans. Typically, the tranches vary in terms of their subordination. For example, the equity or residual tranche receives a high coupon on its principal amount, but is ?rst in line to absorb any credit losses suffered by the underlying
4 Alternatively, a synthetic CDO could be constructed through the use of credit default swaps. 5 For an excellent review of the ABS market, see Rajan, McDermott, and Roy (2007).

portfolio. On the other hand, a supersenior tranche might only receive a coupon of LIBOR plus 20 basis points, but would not suffer any credit losses until after the total credit losses for the portfolio exceeded, say, 15%. In effect, an asset-backed CDO structure could be viewed as a synthetic lender where the assets consist of, say, subprime home-equity loans and where the capital structure consists of equity, subordinated debt, and senior debt (all often in the form of ?oating-rate notes).6 From a CDO issuer’s perspective, the advantage of issuing CDOs is that it allows the issuer to make loans, repackage them, and then sell them to third parties, thereby allowing the issuer to earn fees from originating and then servicing the loans without having to commit capital permanently. Of course, this originate-to-distribute mechanism creates a number of moral-hazard risks as the issuer is aware that he may bear very little of the credit losses on the loans he makes since they will be sold as repackaged CDOs. To provide an illustration of a typical subprime assetbacked CDO, Table 1 gives the details of a $900 million CDO sponsored by Countrywide Home Loans, Inc. and issued through Lehman Brothers in February 2006. The issuing entity is designed as CWABS Asset-Backed Certi?cates Trust 2006-1. Of the total notional amount underlying the CDO, about $500 million is based on subprime ?xed-rate mortgages, while $400 million is based on subprime ?oating-rate mortgages. On the ?xed-rate side, the CDO consists of 12 separate tranches. The ?rst six are equal in

6

See the discussion in Longstaff and Myers (2009).

ARTICLE IN PRESS
440 F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

seniority but differ in terms of their coupon rates and collateral. The other six tranches are subordinated sequentially, with the MF-6 tranche absorbing the ?rst $5.525 million in losses, the MF-5 tranche absorbing the next $6.188 million in losses, etc. A similar structure applies on the ?oating-rate side of the portfolio with the MV-7 tranche absorbing the ?rst $5.549 million of losses, the MV-6 tranche absorbing the next $5.907 million of losses, etc. The average FICO credit score for the ?xed-rate and ?oating-rate loans is 611 and 618, respectively, placing these loans squarely in the subprime category. Interestingly, while some of the underlying mortgages bear low ‘‘teaser’’ rates, many carry very high mortgage rates; the mortgage rates for the loans in the underlying portfolio vary from 4.95% to 12.00%. Given the different positions of the tranches in the capital structure ‘‘pecking order,’’ it is not surprising that the tranches can have different credit ratings. Table 1 shows that the initial credit ratings for the tranches offered range from Aaa/AAA to Baa1/A. Since each of these CDO tranches can be viewed as either a ?xed-rate bond or a ?oating-rate note, the prices of these securities are generally quoted per $100 notional. To illustrate, the MF-1 tranche in the CWABS 2006-1 example has a Bloomberg quoted price of 65.00 on December 4, 2007. Thus, an investor who acquired this tranche at the issue price of 99.99814 on February 8, 2006 would have a markto-market loss of nearly 35%. Given that this tranche initially had a credit rating of AA1/AA+, the subsequent large decline in the value of the tranche argues that the initial credit ratings may have been overly optimistic. From the perspective of the asset-backed CDO markets, there are several key events or threads that underlie the current distressed state of the market. First, the recent wave of subprime defaults and declines in housing values has created severe uncertainty about what the ultimate magnitude of credit losses will be. Second, given the inherent complexity of the underlying loan portfolios on which assetbacked CDOs are based (as evidenced from the Countrywide example in Table 1), participants in the ?nancial markets apparently placed too much reliance on the credit ratings provided by the ratings agencies in making investment and pricing decisions. For example, see the discussion in Benmelech and Dlugosz (2009). When the rating agencies began to backtrack from their previous optimistic ratings in mid-2007 and the liquidity in secondary CDO markets dried up, many investors were left with what Clarida (2007) describes as almost-Knightian uncertainty as to what their asset-backed CDO positions were actually worth. 4. The ABX indexes To measure the returns on subprime CDOs, I use market quotations for the widely known ABX indexes maintained by Markit Group Ltd. These indexes consist of daily closing values obtained from market dealers for subprime home-equity-related CDOs of various credit ratings.7 In particular, the ABX indexes consist of ?ve
Market makers for the ABX indexes during most of the sample period included Bank of America, BNP Paribas, Deutsche Bank, Lehman
7

separate indexes, where each of these indexes is based on the market quotations of a speci?c basket of distinct subprime CDO tranches. The AAA index is based on a portfolio of 20 subprime home-equity CDOs with initial credit ratings of AAA. The AA index is based on a portfolio of 20 subprime homeequity CDOs with initial credit ratings of AA. Similarly, the other three indexes are based on portfolios of subprime home-equity CDOs with credit ratings of A, BBB, and BBB ? , respectively. Each index is a simple average of the prices for the 20 CDOs or tranches in the basket, where prices are quoted relative to a $100 notional position. The 20 subprime deals that appear in each basket are chosen from among the qualifying deals of the largest subprime home-equity ABS shelf programs during the six-month period preceding the formation of the indexes. The algorithm for choosing the 20 subprime CDOs to be included in each index limits the same loan originator to four deals and the same master servicer to six deals. The minimum deal size is $500 million. Each CDO (tranche) must have a weighted-average life between four to six years as of the issuance date (except the AAA tranche which must be greater than ?ve years). The tranches must be rated by Moody’s and Standard and Poors; the lesser of the ratings applies. At least 90% of a deal’s assets must be ?rst-lien mortgages, and the weighted-average FICO credit score for loans underlying the tranche must be less than 620. Deals must pay on the 25th of each month and referenced tranches must bear interest at a ?oating-rate benchmark of one-month LIBOR. The ?ve ABX indexes are reconstituted every six months. The ?rst series of ABX indexes were formed in January 2006 and designated the ABX.HE 1 AAA, AA, A, BBB, and BBB ?indexes. The second series of ABX indexes were formed in July 2006 and designated the ABX.HE 2 AAA, AA, A, BBB, and BBB ?indexes. Similarly formed were the ABX.HE 3 and ABX.HE 4 indexes in January 2007 and July 2007, respectively. Once the subprime crisis began in the latter part of 2007, however, subprime CDO issuance declined precipitously and new ABX indexes were no longer formed. Thus, the ABX.HE 4 index remains the on-therun or most-recently created ABX index from mid-2007 to the end of the sample period. Market quotations for the ABX indexes can be dif?cult to obtain. Fortunately, I was given access to a proprietary data set by a major ?xed-income asset management ?rm that includes daily closing values for all of the ABX.HE 1, 2, 3, and 4 indexes for the three-year period from the inception of the ABX index in January 19, 2006 to December 31, 2008. Table 2 provides a brief chronology of some of the major crisis events during the 2006–2008 period. This timeline suggests that the ongoing crisis could be viewed as having two distinct phases. The ?rst was the subprime crisis of 2007 in which investors and ?nancial institutions

(footnote continued) Brothers, Morgan Stanley, Barclays Capital, Citigroup, Goldman Sachs, RBS Greenwich Capital, UBS, Bear Stearns, Credit Suisse, JP Morgan, Merrill Lynch, and Wachovia.

ARTICLE IN PRESS
F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450 441

Table 2 Timeline of the subprime and ?nancial market crises. Source: Reuters, Federal Reserve of St. Louis. Late 2006 Feb-7-2007 Apr-2-2007 Jun-20-2007 Jul-10-2007 Jul-17-2007 Jul-20-2007 Aug-9-2007 Sep-13-2007 Oct-1-2007 Oct-30-2007 Nov-4-2007 Dec-12-2007 Jan-1-2008 Feb-13-2008 Mar-112008 Mar-162008 Mar-242008 Jun-5-2008 Jul-11-2008 Sep-7-2008 Sep-15-2008 Sep-16-2008 Sep-25-2008 Sep-29-2008 Oct-3-2008 Nov-252008 Dec-19-2008 The U.S. housing market slows after two years of increases in of?cial interest rates. Delinquencies rise; a wave of bankruptcies. Europe’s biggest bank, HSBC Holdings, blamed soured U.S. subprime loans for its ?rst-ever pro?t warning. Subprime lender New Century Financial Corp. ?les for bankruptcy. Two Bear Stearns funds sell $4 billion of assets to cover redemptions and expected margin calls arising from subprime losses. Standard & Poor’s said it may cut ratings on some $12 billion of subprime debt. Bear Stearns says two hedge funds with subprime exposure have very little value; credit spreads soar. Home foreclosures soar 93% from the previous year. BNP Paribas suspends redemptions in $2.2 billion of asset-backed funds; says it cannot determine security values. UK mortgage lender Northern Rock seeks ?nancial support from the Bank of England; report sparks a run by worried depositors. Swiss bank UBS said it would write down $3.4 billion in its ?xed-income portfolio; ?rst quarterly loss in nine years. Merrill Lynch ousts Chairman and Chief Executive Stan O’Neal after reporting biggest quarterly loss in company’s history. Citigroup announces a further $8-11 billion of subprime-related writedowns and losses. Charles Prince resigns as CEO. Central banks coordinate the launch of the temporary Term Auction Facility (TAF) to address pressures in short-term funding markets. Bank of America purchases Countrywide Financial in an all-stock transaction. President Bush signs the Economic Stimulus Act of 2008 into law. Federal Reserve announces creation of Term Securities Lending Facility (TSLF). Federal Reserve announces creation of Primary Dealer Credit Facility (PDCF). JP Morgan acquires Bear Stearns in rescue partially ?nanced by Federal Reserve Bank of New York. Standard & Poor’s announces downgrade of monoline insurers AMBAC and MBIA. Of?ce of Thrift Supervision closes IndyMac Bank, F.S.B. Federal Housing Finance Agency places Fannie Mae and Freddie Mac in government conservatorship. Bank of America announces purchase of Merrill Lynch; Lehman Brothers ?les Chapter 11 bankruptcy. Federal Reserve authorizes lending up to $85 billion to AIG. Of?ce of Thrift Supervision closes Washington Mutual Bank. Federal Deposit Insurance Corporation (FDIC) announces that Citigroup will purchase the banking operations of Wachovia Corp. Congress passes Emergency Economic Stabilization Act establishing $700 billion The Troubled Asset Relief Program (TARP). Federal Reserve Board announces creation of Term Asset-Backed Securities Lending Facility (TALF). U.S. Treasury authorizes loans for General Motors and Chrysler from the TARP.

holding subprime credit-related assets experienced major losses. The second is the global ?nancial crisis of 2008 which was marked by massive deleveraging as well as by failures of major ?nancial institutions with general credit exposure as the economy slid rapidly into recession. Fig. 1 plots the time series of ABX index values for each of the three years 2006, 2007, and 2008. As illustrated, the ABX indexes were generally close to par during much of 2006, although the ABX BBB and BBB ? began to decline toward the end of 2006. During the ?rst part of 2007, the ABX BBB and BBB? indexes continued their decline. Around the middle of 2007, however, the other ABX indexes began to decrease. By the end of 2007, the ABX AAA index was below 80 and the other indexes were all below 50. During 2008, all of the ABX indexes continued to decline steadily and ended the year below 10, with the exception of the ABX AAA index which dipped below 30 but recovered somewhat to 40 at the end of 2008. Table 3 provides summary statistics for the ABX index returns. Fig. 2 plots the time series of ABX index returns for each of the three years in the sample period. Note that these returns are based on weekly changes in the ABX index levels (weekly coupon accruals are not included in the ABX index returns). Table 3 shows that the returns on the ABX indexes became increasingly negative during the sample period. During 2006, the two highest-rated indexes actually experienced positive returns. During 2007 and 2008, the AAA index experienced negative returns, but these were not nearly as severe as for the other indexes. Not surprisingly, the volatility of ABX index

returns was signi?cantly higher in 2007 and 2008 than in 2006. Interestingly, the volatility of ABX index returns is not monotonically related to credit rating; the ABX A index was the most volatile index during 2007 while the ABX AA index was the most volatile index during 2008. Table 3 also shows that there were major changes in the relation between the different ABX indexes during the sample period. During 2006, the average correlation of returns across all indexes was 0.500. During 2007, this measure increased to 0.744. During 2008, the average correlation of returns across all indexes declined to 0.587, approximating its value during the 2006 pre-crisis period.

5. Testing for contagion In studying the nature of contagion in ?nancial markets, it is helpful to have two key elements. First, I must be able to identify an event window for the distress event. Second, I must be able to identify a vector of contagion which can then be used to test for changes in linkages across markets associated with the distress event. The subprime crisis of 2007 provides a nearly textbook example of a potential contagion event in which both of these elements are present. In particular, the subprime crisis began during early 2007 as market participants gradually began to fear that the cash ?ows from their holdings of asset-backed CDOs might ultimately be far less than they had anticipated given the high credit ratings that these securities initially carried.

ARTICLE IN PRESS
442 F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

102 Price 99 96 Jan Mar Jun Sep Dec

90 Price 60 30 Jan 75 45 15 Jan Mar Jun Sep Jan Mar Jun Sep Dec

Fig. 1. The upper, middle, and lower panels plot the ABX subprime indexes weekly for 2006, 2007, and 2008, respectively. The solid gray line represents the AAA index; the dotted line, the AA index; the dashed-dotted line, the A index; the dashed line, the BBB index; the x’s, the BBB ? index.

Table 3 Summary statistics for ABX home-equity CDO tranche weekly returns. This table reports summary statistics for the weekly percentage price changes for the indicated ABX indexes for each year. Each of the ?ve ABX indexes represents an average of the prices of 20 subprime residential mortgage-backed CDOs with the same rating. Speci?cally, the AAA index is an average of 20 subprime CDOs with the rating of AAA; the AA index is an average of 20 subprime CDOs with the rating of AA; etc. The ABX indexes are maintained by Markit Group Ltd. The ABX indexes are reconstituted every six months, and the most-recently constructed indexes are denoted the on-the-run indexes. The sample consists of weekly data for the on-the-run ABX indexes from January 25, 2006 to December 31, 2008. Year Rating Mean Std. dev Min. Max AAA 2006 AAA AA A BBB BBB ? AAA AA A BBB BBB ? AAA AA A BBB BBB ? 0.002 0.008 ? 0.012 ? 0.067 ? 0.087 ? 0.551 ? 1.447 ? 2.229 ? 2.779 ? 2.840 ? 1.016 ? 3.499 ? 3.544 ? 3.407 ? 3.203 0.022 0.042 0.100 0.393 0.462 3.465 6.867 8.077 6.666 6.824 6.443 8.527 6.996 5.373 5.792 -0.045 ? 0.119 ? 0.301 ? 1.979 ? 2.081 ? 12.230 ? 29.754 ? 28.787 ? 21.429 ? 26.618 ? 16.573 ? 29.697 ? 23.980 ? 20.000 ? 16.238 0.090 0.130 0.140 0.465 0.535 9.737 21.416 18.774 13.595 12.940 14.839 14.838 9.845 6.594 12.030 1.00 0.44 0.36 0.23 0.33 1.00 0.85 0.80 0.58 0.51 1.00 0.75 0.44 0.44 0.39 AA Correlation A BBB BBB ?

Price

1.00 0.50 0.32 0.44

1.00 0.77 0.76

1.00 0.85

1.00

2007

1.00 0.89 0.69 0.60

1.00 0.80 0.77

1.00 0.95

1.00

2008

1.00 0.69 0.62 0.59

1.00 0.53 0.52

1.00 0.90

1.00

ARTICLE IN PRESS
F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450 443

0.01 0 ?0.01

Return

Feb 0.2 Return 0 ?0.2 Jan 0.2 Return 0 ?0.2 Jan

Mar

Jun

Sep

Dec

Mar

Jun

Sep

Dec

Mar

Jun

Sep

Jan

Fig. 2. The upper, middle, and lower panels plot the weekly ABX subprime index returns for 2006, 2007, and 2008, respectively. The solid gray line represents the AAA index; the dotted line, the AA index; the dashed-dotted line, the A index; the dashed line, the BBB index; the x’s, the BBB ? index.

Thus, asset-backed CDOs can clearly be viewed as the prime vector of contagion. By early 2008, however, the subprime crisis began to evolve into the global ?nancial crisis as these fears were realized with the failures of Bear Stearns, IndyMac Bank, Washington Mutual, Lehman Brothers, AIG, Fannie Mae, Freddie Mac, Merrill Lynch, and many others. Thus, it is natural to divide the sample period into three distinct periods: the 2006 pre-crisis period, the 2007 subprime-crisis period, and the 2008 global-crisis period.8 To explore the empirical implications of the contagion literature for the subprime crisis, the approach will be to test whether there is an increase in the cross-market linkages between the asset-backed CDO market and other major ?nancial markets during the subprime crisis. This approach is motivated by the standard de?nition in the literature of contagion as a change in the linkages between markets following a distress event. Speci?cally, I apply a vector autoregression (VAR) framework that allows us to estimate the relation between asset-backed CDO returns and returns in other ?nancial markets separately during the three subperiods of the sample period. This allows us to examine directly whether crossmarket linkages during the 2007 subprime crisis differed from those during the other two periods.

5.1. The VAR variables As measures of the returns in the distressed assetbacked CDO market, I use the returns on the ABX indexes (formed from the on-the-run series, e.g., rolling the series from ABX-HE 1 to ABX-HE 2 when the latter index is constructed, etc.). Speci?cally, I use the weekly (Wednesday to Wednesday) returns for the corresponding on-the-run ABX index. Altogether, I have ?ve such on-therun series of returns, each representing a different credit rating, which I designate ABXAAA, ABXAA, ABXA, ABXBBB, and ABXBBB ? . In testing for ?nancial contagion in other ?nancial markets, I will focus on a number of major ?xed-income, equity, and volatility markets. To capture changes in the Treasury bond market, I use weekly changes (over the same period as for the ABX returns) in the constant maturity one- and 10-year Treasury yields (obtained from the Federal Reserve Board’s Web site). Yields are measured in percentage terms. Thus, a one-basis point yield change from, say, 4.50 to 4.51 equals 0.01. To capture changes in corporate bond spreads, I use the Moody’s Aaa and Baa corporate yield indexes and compute the spread by subtracting the 10-year Treasury yield from these index values. The weekly Moody’s data are obtained from the Federal Reserve Board. To capture changes in the stock market, I use two different measures. Speci?cally, I collect weekly return

8

I am grateful to the referee for suggesting this approach.

ARTICLE IN PRESS
444 F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

data for both the S&P 500 index and the S&P 500 subindex of ?nancial ?rms (dividends omitted from both return series). This subindex consists of roughly 80 to 90 commercial and investment banks, insurance companies, and home lenders during the sample period. The data for the S&P 500 indexes are obtained from the Bloomberg system. As the measure of volatility, I use weekly changes in the VIX volatility index. The data for the VIX are also obtained from the Bloomberg system.

5.2. The VAR results Turning now to the question of whether the subprime crisis resulted in increased cross-market linkages between the asset-backed CDO market and other major markets, I estimate the following VAR system: Yt ? a ?
4 X k?1

bk Yt-k ? gk ABX t-k ? et ;

?1?

separately for each of the seven different dependent variables Yt described in the previous section. Speci?cally, as the dependent variable Yt, I use the changes in the oneand 10-year Treasury yields, changes in the Moody’s Aaa and Baa credit spreads, the returns on the S&P 500 index, the returns on the subindex of S&P 500 ?nancial ?rms, and changes in the value of the VIX index. The four-week lag structure is suggested by the data and is consistent with the Akaike Information Criterion (AIC). Note that for every speci?cation of the dependent variable Y, I estimate the VAR ?ve different times, each time using a different ABX index. In addition, I estimate the VAR separately for each of the three years in the sample period: 2006, 2007, and 2008. Table 4 summarizes the VAR estimation results. For each of the three periods in the sample, I report the Newey–West t-statistics for the gk coef?cients in Eq. (1) and the R2s from the VARs. Table 4 also reports the p-values for the F-test that the gk coef?cients are jointly zero.9 This F-test can also be viewed as a test of the hypothesis that ABX returns Granger-cause subsequent changes or returns in the other ?nancial markets examined. These tests also allow us to determine whether there is a signi?cant difference in the relation between ABX index returns and the other ?nancial markets during the 2007 subprime period. Turning ?rst to the results from the Treasury bond VARs, Table 4 shows that there is a very clear pattern of contagion during the subprime crisis. In particular, few of the individual t-statistics for the lagged ABX index returns are signi?cant for the 2006 VARs. Similarly, none of the F-statistics are signi?cant for the 2006 VARs. These 2006 results are intuitive since the asset-backed CDO market is much less liquid than the Treasury market. Thus, in ordinary circumstances, I would anticipate that there would be very little information in the ABX indexes that might be useful in forecasting Treasury yield changes.
9 For a discussion of this test of joint signi?cance for the VAR coef?cients, see Chapter 11 of Hamilton (1994).

In striking contrast, all of the F-statistics for the 2007 Treasury yield VARs are signi?cant, indicating that the ABX returns have predictive ability for (or Granger-cause) Treasury yield changes. In addition, many of the individual t-statistics are highly signi?cant in these VARs. The AAA and AA indexes have signi?cant forecast power for both one- and 10-year Treasury yields about one to three weeks ahead, while the other ABX indexes have signi?cant forecast power three to four weeks ahead. Table 4 also shows that all of the signi?cant coef?cients for the ABX returns in the 2007 VARs are positive in sign, indicating that a negative shock to the ABX index translates into a decline in Treasury yields, which, in turn, implies an increase in the value of Treasury bonds. Thus, these results are consistent with a ?ight-to-quality in the Treasury bond market in response to shocks in the subprime market. Interestingly, the magnitude of the coef?cients for the 10-year Treasury bonds is roughly the same as that for the one-year Treasury bonds. Recall, however, that the duration and, therefore, the price effect on the value of a 10-year bond is many times that for the one-year bonds. Thus, these results imply large increases in the value of 10-year Treasury bonds stemming from declines in the value of asset-backed CDOs during 2007. The R2 s for the VARs are also very high and compare favorably to those for the forward rate forecasting models presented in Cochrane and Piazzesi (2005). Finally, Table 4 shows that the forecast ability of ABX returns for Treasury yields largely disappears in the 2008 VARs. The exception is that the F-statistic for the AA index is signi?cant at the 10% level for the 10-year Treasury yield. In addition, a number of the individual t-statistics are signi?cant during 2008. In summary, the onset of the subprime crisis resulted in a signi?cant change in the relation between ABX returns and Treasury bond yields. Prior to the crisis, the ABX indexes have little or no forecast power for the highly liquid Treasury bond market. During the crisis, however, ABX returns developed signi?cant forecast power for Treasury yields. Once the subprime crisis evolved into another form, the ABX indexes were no longer a vector of contagion and the relation between ABX index returns and Treasury bond prices reverted to its pre-crisis nature. Thus, these results provide strong support for the view that the 2007 subprime crisis was accompanied by ?nancial contagion as shocks in the asset-backed CDO market were transmitted to the Treasury bond market. Focusing next on the corporate bond market, Table 4 shows a very similar pattern. There is little evidence of forecast ability during 2006. The sole exception is that the F-statistic for the AA index is signi?cant for the Moody’s Aaa spread. In contrast, all of the ABX indexes have signi?cant forecast ability for both the Moody’s Aaa and Baa spreads during 2007. The signi?cant coef?cients in these VARs are all negative in sign, implying that declines in ABX values map into wider subsequent corporate spreads. During 2008, the forecast power of the ABX indexes for corporate spreads again dissipates and none of the F-statistics are signi?cant for either the Moody’s Aaa or Baa spreads.

Table 4 VAR Estimation results. This table reports the Newey–West t-statistics for the indicated coef?cients from the estimation of the VAR speci?cation shown below, where each VAR is estimated separately for the indicated year. Also reported is the p-value for the F-test of the hypothesis that g1 ? g2 ? g3 ? g4 ? 0. In this speci?cation, Y denotes the ?nancial market variable that appears as the dependent variable while ABX denotes the ABX index return whose lagged values (along with lagged values of Y) appear as explanatory variables. Each of the ?ve ABX indexes represents an average of the prices of 20 subprime residential mortgage-backed CDOs with the same rating. Speci?cally, the AAA index is an average of 20 subprime CDOs with the rating of AAA; the AA index is an average of 20 subprime CDOs with the rating of AA; etc. The ABX indexes are maintained by Markit Group Ltd. The ABX indexes are reconstituted every six months, and the most-recently constructed indexes are denoted the on-the-run indexes. One- and 10-year Treasury denote weekly changes in the respective constant maturity Treasury yields. Moody’s Aaa and Baa corporate spread denote weekly changes in the spread of these yield indexes over the 10-year Treasury rate. S&P 500 Financials denotes the weekly return (excluding dividends) of the ?nancial stocks in the S&P 500 index. S&P 500 denotes the weekly return (excluding dividends) on the S&P 500 index. VIX denotes weekly changes in the VIX volatility index. The superscript ** denotes signi?cance at the 5% level; the superscript * denotes signi?cance at the 10% level. The sample period is January 25, 2006 to December 31, 2008. Yt ? a ?
4 X k?1

bk Yt?k ? gk ABX t?k ? et

F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

Y

ABX

2006

2007

2008

g1
One-year Treasury AAA AA A BBB BBB ? ? 1.04 0.18 ? 1.22 ? 1.47 ? 0.74

g2
? 0.30 1.48 1.02 0.21 0.85

g3
0.29 0.01 ? 1.21 0.33 ? 0.56

g4
0.29 0.78 ? 1.04 ? 0.55 ? 0.31

R

2

p 0.91 0.98 0.41 0.91 0.93

g1
0.04 2.59** ? 0.12 0.49 0.83

g2
2.52** 1.36 1.03 ? 0.18 ? 0.11

g3
1.90* 3.44** 7.05** 3.75** 3.56**

g4
? 0.28 ? 0.61 2.06** 2.45** 2.66**

R

2

p 0.00** 0.00** 0.00** 0.00** 0.00**

g1
? 1.35 ? 1.46 ? 1.09 ? 2.71* ? 2.56**

g2
0.15 1.04 1.21 0.40 ? 0.10

g3
1.29 0.21 ? 1.29 ? 0.19 ? 0.06

g4
2.31** ? 0.04 0.90 ? 0.14 ? 1.00

R2 0.12 0.09 0.10 0.11 0.12

p 0.43 0.64 0.58 0.51 0.44

0.08 0.06 0.15 0.08 0.07

0.40 0.48 0.54 0.43 0.39

ARTICLE IN PRESS

10-year Treasury

AAA AA A BBB BBB ?

0.31 2.89** ? 0.45 ? 0.74 ? 0.48

? 0.25 2.64** 0.52 ? 0.41 0.91

1.33 1.24 ? 1.37 ? 0.81 ? 1.44

0.65 0.52 ? 0.92 0.36 0.13

0.07 0.17 0.11 0.06 0.08

0.89 0.27 0.65 0.94 0.87

? 0.46 0.97 0.66 1.65 2.68**

7.02** 1.90* 0.99 0.17 ? 0.32

2.00** 1.71* 3.50** 2.09** 2.34**

? 0.22 0.32 1.97* 1.92* 2.65**

0.33 0.28 0.23 0.33 0.39

0.01** 0.03** 0.01** 0.01** 0.00**

? 0.51 0.80 0.46 ? 0.43 ? 0.58

2.30** 2.69** 1.53 1.06 1.33

0.62 ? 0.72 ? 1.24 ? 0.41 ? 0.82

0.57 ? 0.33 0.74 0.33 ? 0.05

0.23 0.26 0.19 0.13 0.16

0.19 0.09* 0.36 0.86 0.66

Moody’s Aaa corporate spread

AAA AA A BBB BBB ?

1.59 2.17** 1.55 1.51 ? 1.45

0.16 0.44 ? 1.30 ? 0.98 ? 1.02

0.92 1.58 1.27 0.43 1.14

? 0.98 ? 1.57 ? 1.06 ? 0.16 ? 0.01

0.16 0.28 0.18 0.10 0.11

0.37 0.04** 0.28 0.76 0.70

1.01 ? 0.54 0.46 0.14 ? 0.25

? 3.07** ? 1.31 ? 1.60 ? 0.94 ? 1.01

? 2.62** ? 7.28** ? 9.04** ? 2.22** ? 2.22**

1.81* 2.07** 0.51 ? 0.58 ? 1.08

0.34 0.41 0.45 0.42 0.33

0.02** 0.00** 0.00** 0.00** 0.03**

? 0.63 ? 0.27 ? 0.42 0.38 0.41

? 0.88 0.44 0.56 1.38 2.00**

? 0.31 1.09 1.35 1.39 1.39

0.81 0.59 0.10 0.22 1.13

0.22 0.20 0.21 0.22 0.24

0.78 0.89 0.86 0.72 0.57

Moody’s Baa corporate spread

AAA AA A BBB BBB ?

0.49 ? 1.08 1.11 0.87 0.37

? 1.46 ? 0.39 ? 2.18** ? 1.87* ? 1.86*

? 2.09** ? 0.16 0.25 ? 0.02 0.06

? 0.68 ? 1.46 ? 0.57 0.13 0.01

0.17 0.24 0.20 0.13 0.13

0.40 0.12 0.24 0.68 0.63

1.58 ? 0.13 0.72 ? 0.31 ? 0.54

? 5.89** ? 2.58** ? 2.48** ? 0.96 ? 1.12

? 5.12** ? 4.07** ? 4.67** ? 2.28** ? 2.43**

1.54 1.22 ? 0.70 ? 1.28 ? 2.56**

0.53 0.52 0.54 0.45 0.40

0.00** 0.00** 0.00** 0.00** 0.01**

? 0.02 0.07 0.00 1.09 0.91

? 0.62 0.17 0.19 1.50 1.71*

1.40 1.42 1.63 1.08 1.08

1.07 1.05 0.34 0.17 1.11

0.23 0.22 0.21 0.24 0.23

0.53 0.63 0.72 0.47 0.47

S&P 500 Financials

AAA AA A BBB BBB ?

? 1.36 ? 3.56** ? 2.11** ? 0.87 ? 1.54

? 0.45 ? 0.36 ? 1.31 ? 1.25 ? 1.27

? 0.96 ? 0.49 ? 1.09 ? 1.97* ? 0.64

0.86 0.75 0.65 0.73 0.54

0.19 0.22 0.17 0.20 0.18

0.56 0.35 0.17 0.48 0.61

1.09 1.91* 2.43** 1.77* 3.24**

1.55 ? 2.13** ? 0.34 0.92 0.76

0.37 2.87** 2.39** 1.68* 2.08**

? 0.48 ? 0.10 0.46 1.29 1.73*

0.36 0.47 0.51 0.50 0.52

0.41 0.02** 0.01** 0.01** 0.00**

? 0.90 ? 0.02 0.57 ? 0.25 ? 0.39

? 0.14 0.16 ? 0.88 ? 1.16 ? 0.90

0.45 ? 1.58 ? 3.01** 0.41 0.30

? 0.66 ? 1.29 ? 0.93 ? 0.54 ? 1.00

0.14 0.16 0.24 0.16 0.15

0.88 0.73 0.19 0.85 0.82 445

ARTICLE IN PRESS
446 F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

AAA AA A BBB BBB ?

Table 4 (Continued)

ABX

S&P 500

VIX

AAA AA A BBB BBB ?

Turning to the results for the S&P 500 indexes, Table 4 shows that there is little or no evidence that ABX returns were able to forecast S&P 500 index returns during the pre-crisis period. In contrast, ABX index returns became highly predictive of stock index returns during the 2007 subprime crisis. Speci?cally, the F-statistics for the lagged AA, A, BBB, and BBB ?index returns are signi?cant at the 5% level for the S&P 500 ?nancials, and the F-statistics for all ?ve ABX indexes are signi?cant at either the 5% or 10% level for the S&P 500 index. Most of the many signi?cant coef?cients are positive in sign, indicating that a negative shock in the ABX index results in a subsequent negative return for the S&P 500 stock index. Again, these results are consistent with the view that the subprime crisis resulted in contagion being spread from the asset-backed subprime market to other much larger and more liquid markets like the stock market. The VAR results also show that the stock market returns are highly predictable on the basis of ex ante data during the 2007 subprime crisis. In fact, the S&P 500 ?nancial subindex displays a stunning amount of predictability, with R2 s ranging from 36% to 52%. These values far exceed most of the stock market predictability results previously shown in the literature.10 The R2s for the S&P 500 index returns are also very high, with values ranging from 29% to 41%. Table 4 also shows that the much of the predictability of ABX index returns for the stock market dissipates in 2008. The F-statistics for the lagged ABX index returns are only signi?cant at the 10% level for the AAA and AA indexes in the S&P 500 VARs. Only one of the t-statistics for the lagged ABX index returns is signi?cant in the S&P 500 ?nancials VARs. On the other hand, all ?ve of the VARs for the S&P 500 index returns have at least one t-statistic that is signi?cant at the 10% level. Thus, the relation between ABX index returns and S&P 500 index returns does not completely revert to its pre-crisis pattern once the subprime crisis evolves. Finally, Table 4 shows that the results for the VIX VARs are very similar to the others. In particular, there is little evidence of any lead–lag relation between ABX index returns and changes in the VIX during 2006. During 2007, however, all ?ve of the ABX indexes have signi?cant predictive ability at the 10% level for subsequent changes in the VIX. The negative sign for all of the signi?cant t-statistics indicates that the VIX increases as negative shocks to the ABS indexes occur. This is very intuitive since the VIX is often designated as a ‘‘fear’’ index; negative ?nancial news often is linked to increases in the volatility of markets as measured by the VIX. During 2008, the ability of the ABX indexes to forecast changes in the VIX dissipates signi?cantly, with only the F-statistic for the AA index being signi?cant at the 5% level. Note, however, that a number of the individual t-statistics for the lagged ABX index returns remain signi?cant in 2008.

0.25 0.25 0.21 0.21 0.21

? 1.47 ? 0.87 ? 0.36 ? 0.68 ? 0.60

2008

g4

0.46 ? 2.01** ? 2.95** ? 0.99 ? 1.26

g3

0.15 0.75 0.37 ? 0.05 ? 0.23

g2

? 2.19** ? 1.89* ? 1.41 ? 1.74* ? 1.81*

g1

0.09* 0.09* 0.01** 0.00** 0.00**

? 0.40 0.23 1.25 2.68** 3.19**

2007

g4

0.20 2.92** 3.11** 1.96* 2.18**

g3

2.84** 0.91 0.78 1.34 0.76

g2

g1

0.29 0.18 0.26 0.55 0.80

1.16 1.04 0.79 1.20 0.78

0.27 0.30 0.28 0.23 0.20

2006

g4

? 0.91 ? 0.82 ? 1.17 ? 1.41 ? 0.22

g3

? 0.04 ? 0.21 ? 0.93 ? 1.14 ? 1.08

g2

? 1.53 ? 4.11** ? 1.33 ? 0.62 ? 0.98

g1

1.14 2.10** 1.41 1.62 1.71*

? 0.25 ? 1.09 0.13 ? 0.35 ? 0.14

? 0.86 0.23 ? 0.10 ? 0.62 ? 1.29

? 0.77 ? 0.46 0.69 0.17 1.09

0.18 0.16 0.20 0.16 0.20

R2

0.70 0.84 0.54 0.86 0.48

p

? 4.15** ? 2.27** ? 3.97** ? 1.82* ? 2.67**

2.08** 1.28 1.35 0.58 1.92*

? 2.85** ? 0.90 ? 1.42 ? 2.59** ? 1.63

1.55 ? 0.93 ? 3.76** ? 2.82** ? 2.89**

0.46 0.31 ? 0.61 ? 2.09** ? 2.82**

0.34 0.34 0.44 0.52 0.49

0.29 0.30 0.36 0.44 0.41

R2

0.05** 0.05** 0.00** 0.00** 0.00**

p

2.07** 2.70** 2.39** 1.50 1.28

0.20 ? 1.38 ? 0.64 0.58 0.79

0.90 2.53** 1.84* 0.55 0.53

0.22 0.03 0.65 0.72 0.79

0.18 0.29 0.16 0.19 0.15

R2

0.22 0.02** 0.32 0.21 0.39

p

0.06* 0.07* 0.17 0.15 0.16

10 As examples of the recent market predictability literature, see Lettau and Ludvigson (2001) and Cochrane (2008).

Y

ARTICLE IN PRESS
F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450 447

5.3. Discussion Taken together, the evidence that ABX index returns developed signi?cant predictive ability for returns or changes in other major ?nancial markets such as Treasury bonds, corporate bonds, S&P 500 stock indexes, and the VIX during the 2007 subprime crisis provides strong support for the hypothesis that there were spillover contagion effects during this crisis. Cross-market linkages became much stronger and signi?cant during the subprime crisis, consistent with the standard de?nition of ?nancial contagion. Equivalently, ABX index returns are able to Granger-cause returns in other markets during the subprime crisis (but not before or after the subprime crisis, indicating a change in cross-market linkages).11 These results also shed light on the earlier discussion about the nature of the contagion mechanism in ?nancial markets. Recall that the literature on contagion identi?es at least three possible channels by which contagion in ?nancial markets might be propagated: the correlatedinformation channel, the liquidity channel, and the riskpremium channel. The strong evidence that ABX index returns were able to forecast changes or returns several weeks ahead in much larger and more liquid markets during the 2007 subprime crisis argues against the correlated-information channel as the contagion mechanism. Intuitively, the reason for this is simply that we would expect any relevant information discovered in the ABX markets to be very rapidly incorporated into the actively traded stock, stock index option, and Treasury bond markets. Thus, we would expect that there would be a nearly contemporaneous relation between shocks in the ABX index market and these other ?nancial markets if contagion was spread via the correlated-information channel. Note that illiquidity in the pricing of ABX tranches cannot explain the ability of ABX index returns to forecast the returns in other markets. By essentially ruling out the correlated-information channel, I am left with the possibility that ?nancial contagion may have been propagated primarily through either the liquidity channel or the risk-premium channel during the subprime crisis (or both). To address this issue more de?nitively, however, I need to explore in more depth whether a link between the ABX market and trading and liquidity/funding patterns in other markets emerged during the subprime crisis. This analysis is the focus of the next section.

6. Was there liquidity contagion? To explore the effects of the subprime crisis on market/ funding liquidity, I again use the VAR framework introduced in the previous section. Rather than using asset returns or yields as the dependent variables in the VARs, however, I use a number of measures that attempt to capture weekly changes in the trading patterns or liquidity pro?les of key ?nancial markets.

6.1. The liquidity variables First, to explore whether the subprime crisis is associated with changes in trading patterns in the equity markets, I compute the ratio of the aggregate weekly trading volume for the ?rms in the S&P 500 subindex of ?nancial ?rms to the aggregate weekly trading volume for all ?rms in the S&P 500 index. In computing this ratio for week t, I use the volume data for the week immediately before and including the Wednesday of week t. The rationale for considering this variable is to examine if ABX index returns forecast or Granger-cause changes in the amount of trading of ?nancials relative to that for a broader set of stocks. Finding that the stocks in the S&P 500 subindex of ?nancial ?rms are traded more intensively than the remaining S&P 500 ?rms during the crisis could be consistent with a ?ight-to-quality or a major rebalancing of portfolios in the ?nancial markets.12 Second, as one measure of the trading/liquidity patterns in the ?xed-income markets, I collect data on the aggregate amount of fails reported by the New York Federal Reserve. Fails represent repurchase (repo) transactions in which one party fails to deliver the ?xed income securities that are the collateral for the repo contract. Since it is costly to fail on a repo contract, market participants attempt to avoid failures whenever possible. Thus, a sudden increase in the amount of fails in the market signals that some type of market disruption may have occurred in bond markets. This implies that the amount of fails each week (measured in $ millions) can provide a measure of liquidity/trading shocks in the ?xedincome markets. Third, as suggested by Brunnermeier and Pedersen (2005), liquidity shocks may take the form of declines in the amount of funding available to leveraged players in the ?nancial markets. In general, measuring changes in the aggregate size of the credit sector over a short period such as a week is challenging. In the current subprime crisis, however, considerable attention has focused on the commercial paper market. Over the past decade, this market has grown to a notional size on the order of $2 trillion and has become a major source of short-term funding for ?nancial institutions, investors, and corporations. Given that the median maturity of commercial paper is on the order of 30 days, changes in the size of this
12 On the other hand, a change in this trading pattern could also be consistent with an equilibrium in which agents hedge against consumption risks by trading certain sectors of the market more actively than others. I am grateful to the referee for this observation.

11 I also examined the reverse VAR speci?cation to determine whether returns in these other markets Granger-caused ABX index returns during the subprime crisis. Intuitively, ?nding some evidence of Granger-causality in this direction would not be surprising given that most of the other markets examined are much more liquid and actively traded than are the ABX indexes. In actuality, however, the number of signi?cant F-statistics for this reverse speci?cation was far fewer than for those reported in Table 4. Also, when the F-statistic for the reverse speci?cation was signi?cant, it was nearly always less signi?cant than for the corresponding speci?cation in Table 4.

448 Table 5 VAR Estimation results for liquidity and ?nancing variables. This table reports the Newey–West t-statistics for the indicated coef?cients from the estimation of the VAR speci?cation shown below, where each VAR is estimated separately for the indicated year. Also reported is the p-value for the F-statistic of the hypothesis that g1 ? g2 ? g3 ? g4 ? 0. In this speci?cation, Y denotes the liquidity or ?nancing variable that appears as the dependent variable while ABX denotes the ABX index return whose lagged values (along with lagged values of Y) appear as explanatory variables. Each of the ?ve ABX indexes represents an average of the prices of 20 subprime residential mortgagebacked CDOs with the same rating. Speci?cally, the AAA index is an average of 20 subprime CDOs with the rating of AAA; the AA index is an average of 20 subprime CDOs with the rating of AA; etc. The ABX indexes are maintained by Markit Group Ltd. The ABX indexes are reconstituted every six months, and the most-recently constructed indexes are denoted the on-the-run indexes. Ratio of trading denotes the ratio of trading volume for the S&P 500 ?nancials to the total trading volume for the S&P 500 index for the week. Fails denotes the total value (measured in $ millions) of settlement failures by primary dealers in the Treasury, agency, mortgage, and corporate bond markets for the week. Change in ABS CP is the weekly change (measured in $ billions) in the aggregate amount of asset-backed commercial paper outstanding. The superscript ** denotes signi?cance at the 5% level; the superscript * denotes signi?cance at the 10% level. The sample period is January 25, 2006 to December 31, 2008. Yt ? a ?
4 X k?1

F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

bk Yt?k ? gk ABX t?k ? et

ARTICLE IN PRESS

Y

ABX

2006

2007

2008

g1
Ratio of trading volume AAA AA A BBB BBB ? ? 0.48 ? 0.96 ? 1.79* ? 1.54 ? 1.22

g2
? 0.12 ? 2.70** ? 0.81 ? 2.19** ? 1.13

g3
? 0.58 ? 1.40 ? 2.01** ? 3.10** ? 2.17**

g4
? 0.29 ? 1.51 0.35 0.22 0.27

R2 0.27 0.33 0.36 0.42 0.38

p 0.95 0.43 0.25 0.06* 0.16

g1
? 4.01 ? 3.70** ? 5.00** ? 3.39** ? 3.28**
**

g2
? 2.02 ? 0.67 ? 1.02 ? 1.84* ? 3.25**
**

g3
? 0.41 ? 1.90* ? 2.73** ? 2.72** ? 2.84**

g4
? 1.96 ? 0.74 ? 1.04 ? 1.31 ? 1.34
*

R2 0.90 0.91 0.92 0.91 0.91

p 0.00 0.00** 0.00** 0.00** 0.00**
**

g1
0.25 ? 1.00 ? 2.90** ? 0.47 ? 0.63

g2
? 0.54 ? 1.46 ? 0.97 ? 0.24 0.53

g3
0.35 0.77 2.46** 1.15 0.78

g4
? 2.55 ? 2.87** ? 0.55 ? 0.78 0.19
**

R2 0.56 0.55 0.53 0.47 0.47

p 0.06* 0.08* 0.22 0.86 0.95

Fails

AAA AA A BBB BBB ?

1.18 ? 0.11 0.02 ? 0.35 ? 0.76

3.87** 0.62 ? 0.34 0.42 0.27

2.07** 2.65** 1.47 1.47 1.00

? 0.60 ? 2.28** ? 0.39 ? 0.75 ? 0.87

0.42 0.41 0.30 0.29 0.32

0.07* 0.09* 0.75 0.85 0.62

0.57 ? 0.70 ? 0.04 ? 0.88 ? 0.32

0.02 3.02** 2.70** 1.53 1.01

? 2.82** ? 2.32** ? 1.80* ? 1.56 ? 2.01**

? 1.44 ? 2.16** ? 1.06 ? 1.33 ? 0.64

0.39 0.40 0.36 0.37 0.32

0.09* 0.08* 0.23 0.17 0.56

? 1.05 ? 0.29 ? 0.53 ? 0.90 ? 1.52

0.82 2.47 1.00 1.71* 2.10**

0.47 ? 1.80* ? 1.75* 0.06 0.02

1.43 0.27 0.30 ? 0.38 ? 0.27

0.78 0.79 0.78 0.79 0.79

0.60 0.27 0.63 0.35 0.21

Change in ABS CP

AAA AA A BBB BBB ?

? 0.65 ? 0.10 1.96* 1.10 1.47

? 0.08 ? 0.00 ? 3.13** ? 1.11 ? 1.23

1.45 0.65 1.31 2.38** 2.38**

? 0.05 ? 0.21 ? 1.46 ? 1.41 ? 2.10**

0.35 0.33 0.48 0.41 0.43

0.84 0.99 0.05** 0.30 0.16

3.45** 2.26** 2.90** 1.34 1.64

? 2.68** ? 1.03 ? 0.59 ? 0.05 ? 0.14

? 0.82 ? 2.75** ? 3.30** ? 1.81* ? 2.11**

0.71 2.00** ? 0.01 ? 0.67 ? 0.81

0.46 0.50 0.48 0.42 0.41

0.07* 0.02** 0.04** 0.25 0.26

? 1.99* ? 1.64 ? 0.52 ? 1.07 ? 1.59

2.62** 1.49 0.90 1.32 1.54

0.74 ? 0.35 ? 1.22 ? 1.24 ? 1.59

1.71* 1.93* 2.02** 1.63 1.49

0.28 0.18 0.13 0.15 0.16

0.02** 0.14 0.36 0.28 0.22

ARTICLE IN PRESS
F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450 449

market (measured in $ billions) may provide a useful proxy for discretionary changes in the amount of shortterm credit provided in the ?nancial markets. I obtain weekly (Wednesday) data on the size of the asset-backed commercial paper market from the Federal Reserve Board’s Web site.

6.2. The liquidity VAR results Table 5 reports the VAR estimation results for the liquidity and ?nancing variables. Focusing ?rst on the ratio of trading volume, Table 5 shows that there is some limited predictability by ABX index returns for the ratio during the 2006 pre-crisis period. The F-statistic is signi?cant at the 10% level for the BBB index, while a number of the t-statistics are individually signi?cant during 2006. With the onset of the subprime crisis in 2007, however, the evidence of predictability becomes much stronger. In particular, the F-statistics for all ?ve ABX indexes are highly statistically signi?cant. In addition, many of the individual t-statistics are signi?cant. All of the signi?cant coef?cients for the lagged ABX index returns are negative, implying that a negative shock to asset-backed CDO values is associated with an increase in the trading activity of ?nancial ?rms relative to other ?rms in the S&P 500. These results suggest that investors did not simply trade the market as the subprime distress event unfolded, but concentrated their trading in the ?nancial sector. During 2008, most of the predictive power of the ABX index dissipates, with only the ABX AAA and AA indexes having F-statistics that are signi?cant at the 10% level. Thus, the cross-market linkage between ABX index returns and the ratio of trading activity spiked during 2007, but then essentially returns to its pre-crisis pattern during 2008. The results for the amount of fails in the ?xed-income markets provide some evidence, albeit mixed, that the relation between ABX index returns and ?xed-income market liquidity became more pronounced during the 2007 subprime crisis. In particular, four of the individual t-statistics for the lagged ABX index returns are signi?cant during 2006. During 2007, however, seven of the individual t-statistics are signi?cant. On the other hand, the F-statistics for both the AAA and AA indexes are signi?cant during both 2006 and 2007. In contrast, the relation between ABX index returns and fails becomes much weaker during 2008. Table 5 shows that while there is a weak relation between ABX index returns and changes in ABS commercial paper during 2006, there is a much stronger relation during 2007. Speci?cally, the F-statistics for the AAA, AA, and A indexes are signi?cant at the 5–10% level during 2007, while only the F-statistic for the A index is signi?cant during 2006. During 2008, the relation between ABX index returns and changes in ABS commercial paper returns to a level similar to those for 2006. These results are consistent with the model presented by Brunnermeier and Pedersen (2005) in which funding shocks in one market may translate into broad liquidity

and valuation shocks in other markets, thereby generating pervasive contagion effects in ?nancial markets. In summary, these results do provide evidence that the 2007 subprime crisis resulted in signi?cant changes in the patterns of trading activity, liquidity, and funding in the ?nancial markets. Thus, these results are consistent with both the Brunnermeier and Pedersen (2005) funding-illiquidity contagion mechanism as well as with the portfolio rebalancing implications of Allen and Gale (2000), Kodres and Pritsker (2002), and others, and support the view that contagion during the subprime crisis was spread through a liquidity channel which, in turn, was associated with major portfolio rebalancing by market participants. These results are also consistent with Aragon and Strahan (2009) who study the impact of the Lehman bankruptcy on hedge funds.

7. Conclusion The 2007 subprime crisis provides an ideal opportunity for studying the effects of contagion in ?nancial markets. I use data for the ABX indexes of subprime asset-backed CDOs to examine whether contagion occurred across markets as the crisis developed. Motivated by the frequently adopted de?nition of contagion in the literature as a signi?cant temporary increase in cross-market linkages after a major distress event, I use a VAR framework to test for changes in the relation between the ABX market and other ?nancial markets after the onset of the crisis. The results provide strong evidence of an increase in cross-market linkages. Prior to the subprime crisis, ABX returns contain little useful information for forecasting returns in other major markets. After the crisis began, however, the ABX indexes became highly predictive for Treasury bond yields, corporate yield spreads, stock market returns, and changes in the VIX volatility index. In many cases, the less-liquid ABX indexes are able to forecast Treasury yields, corporate yield spreads, stock market returns, and changes in the VIX up to three weeks ahead with surprisingly high R2s. These results provide strong support that ?nancial contagion spread across markets as the subprime crisis developed. Since I focus only on the subprime crisis, it is important to acknowledge that my results are limited to this speci?c episode in the markets. A key aspect of the study is that the results allow us to contrast among the different models of contagion that appear in the extensive literature on the subject. For example, the length of the forecast horizon, in many cases as long as three weeks, argues against the view that contagion is spread via the correlated-information channel. The reason for this is simply that I would expect that price-discovery in the highly liquid stock, Treasury bond, corporate bond, and VIX markets would occur much more rapidly if the source of contagion was correlated information. Furthermore, the evidence that ABX index shocks during the subprime crisis became predictive for equity and ?xed-income market trading patterns as well as for the amount of securitized ?nancing is consistent with

ARTICLE IN PRESS
450 F.A. Longstaff / Journal of Financial Economics 97 (2010) 436–450

contagion having been spread via a liquidity/?nancing channel as argued by Allen and Gale (2000), Brunnermeier and Pedersen (2005), and others. References
Acharya, V., Pedersen, L., 2005. Asset pricing with liquidity risk. Journal of Financial Economics 77, 375–410. Allen, F., Gale, D., 2000. Financial contagion. Journal of Political Economy 108, 1–33. Allen, F., Gale, D., 2004. Financial intermediaries and markets. Econometrica 72, 1023–1061. Andrade, G., Kaplan, S., 1998. How costly is ?nancial (not economic) distress? evidence from highly leveraged transactions that became distressed. Journal of Finance 53, 1443–1493. Aragon, G., Strahan, P., 2009. Hedge funds as liquidity providers: evidence from the Lehman bankruptcy. Unpublished working paper, University of Arizona. Asquith, P., Gertner, R., Scharfstein, D., 1994. Anatomy of ?nancial distress: an examination of junk-bond issuers. Quarterly Journal of Economics 109, 625–658. Bae, K., Karolyi, A., Stulz, R., 2003. A new approach to measuring ?nancial market contagion. Review of Financial Studies 16, 717–764. Benmelech, E., Dlugosz, J., 2009. The credit rating crisis. In: Acemoglu, D., Rogoff, K., Woodford, M., NBER Macroeconomics Annual 2009, vol. 24. University of Chicago Press, Chicago, IL (Chapter 3). Bhansali, V., Gingrich, R., Longstaff, F., 2008. Systemic credit risk: what is the market telling us? Financial Analysts Journal 64, 16–24 Brunnermeier, M., Pedersen, L., 2005. Predatory trading. Journal of Finance 60, 1825–1863. Brunnermeier, M., Pedersen, L., 2009. Market liquidity and funding liquidity. Review of Financial Studies 22, 2201–2238. Carlin, B., Lobo, M., Vishwanathan, S., 2007. Episodic liquidity crises: cooperative and predatory trading. Journal of Finance 62, 2235–2274. Clarida, R., 2007. A hard day’s knight: the global ?nancial market confronts uncertainty, not Just risk (and the difference is important). Global Perspectives /www.pimco.comS. Clark, K., Ofek, E., 1994. Mergers as a means of restructuring distressed ?rms: an empirical investigation. Journal of Financial and Quantitative Analysis 29, 541–565.

Cochrane, J., 2008. The dog that did not bark: a defense of return predictability. Review of Financial Studies 21, 1533–1575. Cochrane, J., Piazzesi, M., 2005. Bond risk premia. American Economic Review 95, 138–160. Dornbusch, R., Park, Y., Claessens, S., 2000. Contagion: understanding how it spreads. The World Bank Research Observer 15, 177–197. Hamilton, J., 1994. Time Series Analysis. Princeton University Press, Princeton, NJ. John, K., Ofek, E., 1995. Asset sales and increase in focus. Journal of Financial Economics 37, 105–126. Kahl, M., 2002. Economic distress, ?nancial distress, and dynamic liquidation. Journal of Finance 57, 135–168. Kaminsky, G., Reinhardt, C., Vegh, C., 2003. The unholy trinity of ?nancial contagion. Journal of Economic Perspectives 17, 51–74. Kindleberger, C., 1978. Manias, Panics, and Crashes. Basic Books, New York, NY. King, M., Wadhwani, S., 1990. Transmission of volatility between stock markets. Review of Financial Markets 3, 5–33. Kiyotaki, N., Moore, J., 2002. Evil is the root of all money. American Economic Review: Papers and Proceedings 92, 62–66. Kodres, L., Pritsker, M., 2002. A rational expectations model of ?nancial contagion. Journal of Finance 57, 769–800. Kyle, A., Xiong, W., 2001. Contagion as a wealth effect. Journal of Finance 56, 1401–1440. Lettau, M., Ludvigson, S., 2001. Consumption, wealth, and expected returns. Journal of Finance 55, 815–849. Longstaff, F., 2004. The Flight-to-Liquidity premium in U.S. Treasury bond prices. Journal of Business 77, 511–526. Longstaff, F., 2008. Flight-from-leverage in distressed asset markets. Unpublished working paper, UCLA. Longstaff, F., Myers, B., 2009. How does the market value toxic assets? Unpublished working paper, UCLA. Longstaff, F., Rajan, A., 2008. An empirical analysis of the pricing of collateralized debt obligations. Journal of Finance 63, 529–563. Opler, T., Titman, S., 1994. Financial distress and corporate performance. Journal of Finance 49, 1015–1040. Pulvino, T., 1998. Do asset ?re sales exist? An empirical investigation of commercial aircraft transactions. Journal of Finance 53, 939–978. Rajan, A., McDermott, G., Roy, R., 2007. The Structured Credit Handbook. John Wiley & Sons, Hoboken, NJ. Shleifer, A., Vishny, R., 1992. Liquidation values and debt capacity: a market equilibrium approach. Journal of Finance 47, 1343–1366. Vayanos, D., 2004. Flight to quality, ?ight to liquidity, and the pricing of risk. Unpublished working paper, London School of Economics.


赞助商链接

更多相关文章:
week 1---financial crisis
practice o o o o o o o 1.4 Subprime lending...contagion (蔓延) 3 Effects on the global economy...financial crisis (often called the Credit Crunch ...
Financial markets assessment
Some have called for restrictive measures to avoid any potential contagion ... between the debt crisis evidenced in Greece and the US subprime crisis? ...
When there is blood in the water, it is only natura...
contagion spreads from subprime mortgages to other ...crisis and the credit-card market certainly exist...Perlin, Capital One’ s chief financial officer....
money and bank test bank chap09
Chapter 9 Financial Crises and the Subprime Melt... the source of contagion is the A) free-rider...B) credit bust. C) deleveraging. D) market ...
Contagion Fears Hit Markets
The subprime credit cris... 15页 20财富值 Contagion in CDS, bankin.....Its policy of restoring financial-market credibility and international ...
论文常用英语句型
financial crises has again re-emerged with the ...the subprime crises that began in Auguest 2007 ...contagion in emerging markets may be regional ...
关于金融脆弱性与美国金融危机理论与现实的思考
financial crisis shows its "contagion" integrity ...Key words: subprime crisis realestate market ...(一)次贷危机的本质是信用危机和道德危机 根据上述...
The international financial crisis on China's econo...
subprime mortgage crisis triggered by the ...(B) The mechanism of financial contagion Financial...crisis and triggered the chain of credit in ...
The origins of the financial crisis
subprime” borrowers 次级借款人 with poor credit histories who struggled to...financial crisis by other means, as markets have agonised【suffer agony 】 ...
THE SUBPRIME FINANCIAL CRISIS OF 2007
THE SUBPRIME FINANCIAL CRISIS OF 2007_英语考试_...market 次级住房抵押贷款市场上对金融创新的管理不当... Structured credit products 结构化信用产品: ...
更多相关标签:

All rights reserved Powered by 甜梦文库 9512.net

copyright ©right 2010-2021。
甜梦文库内容来自网络,如有侵犯请联系客服。zhit325@126.com|网站地图