Fisher College of Business Working Paper Series Charles A. Dice Center for Research in Financial Economics Liquidity Shocks and Hedge Fund Contagion Nicole M. Boyson, Northeastern University Christof W. Stahel, George Mason University and the FDIC René M. Stulz, Department of Finance, The Ohio State University, NBER, and ECGI Dice Center WP 2011-12 Fisher College of Business WP 2011-03-012 Revision Date: September 2011 Original Date: June 2011 This paper can be downloaded without charge from: http://www.ssrn.com/abstract=1868569 An index to the working paper in the Fisher College of Business Working Paper Series is located at: http://www.ssrn.com/link/Fisher-College-of-Business.html fisher.osu.edu Liquidity Shocks and Hedge Fund Contagion August 25, 2011 NICOLE M. BOYSON, CHRISTOF W. STAHEL, and RENÉ M. STULZ* ABSTRACT In Boyson, Stahel, and Stulz (2010), we investigate whether hedge funds experience worst return contagion – that is, correlations in extremely poor returns that are over and above those expected from economic fundamentals. We find strong evidence of contagion among hedge funds using eight separate style indices for the period from January 1990 to October 2008: the probability of a worst return in a particular index is increasing in the number of other indices that also have extremely poor returns. We then show that large adverse shocks to asset and funding liquidity strongly increase the likelihood of this contagion. In this paper, we further investigate contagion between hedge funds and main markets. We uncover strong evidence of contagion between hedge funds and small-cap, mid-cap and emerging market equity indices, high yield bonds, emerging market bonds, and the Australian Dollar. Finally, we show that this contagion between hedge funds and markets is also significantly linked to liquidity shocks, especially for small-cap domestic equities, Asian equities, high yield bonds, and the Australian Dollar. * Boyson is at Northeastern University, Stahel is at George Mason University and the FDIC, and Stulz is at The Ohio State University, NBER, and ECGI. Electronic copy available at: http://ssrn.com/abstract=1868569 The recent financial crisis has raised concerns among investors, policymakers, and world leaders about possible performance linkages across markets. For example, in an article regarding debt markets in Europe, Professor Nouriel Roubini of NYU stated that “[t]here’s financial contagion in Portugal, Spain, and to a smaller degree even countries like Italy, Belgium, and others in the Eurozone.”1 Generally, contagion means that a shock in one economy or one market spreads to other economies and markets in ways that cannot be explained by fundamentals. For instance, an event of contagion is one where an adverse shock in the stock market in one country leads to an adverse shock in another country’s stock market that cannot be explained by changes in fundamental determinants of stock prices. While identifying the existence of and transmission mechanism for contagion can be challenging, a better understanding of how and why contagion propagates, which markets are affected, and its severity is important for investors, policy makers, and scholars. Academics have conducted studies of contagion in international markets (see for example, Bae, Karolyi, and Stulz (2003) and Bekaert, Harvey, and Ng (2005)) as well as in financial institutions and hedge funds (see, for example, Adrian and Brunnermeier (2008), Cifuentes, Ferrucci, and Shin (2005), and Dudley and Nimalendran (2011)). Hedge funds are particularly relevant in terms of contagion, since managers often tout their funds as having low correlations with broad markets, which should provide downside protection for investors in times of a crisis. In this light, a number of hedge fund managers attempt to be “market neutral” or have a beta of zero, while focusing on absolute rather than relative performance. If hedge funds are affected by contagion during crises, funds that are market neutral outside of crises periods are unlikely to be so during crises periods, and the diversification provided by hedge funds outside of crises overstates the extent to which hedge funds provide diversification to portfolios when a crisis hits. In Boyson, Stahel, and Stulz (2010), we show that hedge funds experience contagion in worst returns and that this contagion is linked to asset and funding liquidity shocks. We define contagion as in Harvey, 1 See Culpan, Tim, December 1, 2010. “Roubini Sees European Financial Contagion Spreading Into Portugal, Spain,” at bloomberg.com. 1 Electronic copy available at: http://ssrn.com/abstract=1868569 Bekaert, and Ng (2005) as “correlation over and above what one would expect from economic fundamentals.” A “worst return” is a return belonging to the bottom 10th percentile of all returns for the entire time-series of returns under study (January 1990 to October 2008). We collect returns for this time frame for eight different single style hedge fund indices provided by Hedge Fund Research (HFR). The styles included are Convertible Arbitrage, Distressed Securities, Event Driven, Equity Hedge, Equity Market Neutral, Global Macro, Merger Arbitrage, and Relative Value Arbitrage. Using quantile and multinomial logit regressions, and controlling for a number of common factors in hedge fund performance, we show that the probability of a single style hedge fund experiencing a worst return is increasing in the number of worst returns experienced by the other seven hedge fund styles. The phenomenon we uncover is economically important: for example, the quantile regression approach shows that the probability that the Event Driven hedge fund index experiences a worst return is 55% when the equally-weighted index of all other hedge fund styles also experiences a worst return, compared to an unconditional probability of 10% if there were no dependence among hedge fund style returns. We also show that this dependence among returns increases dramatically when returns are low (below the 50th percentile), but there is no corresponding increase when returns are high (see Figure 1 in Boyson, Stahel, Stulz (2010)). In other tests, we find that when seven of the hedge fund style indices experience worst returns, the probability that the eighth style index will also experience a worst return is greater than 70%, compared to an unconditional probability of about 6%. After documenting the existence of contagion among hedge funds, we investigate possible explanations for this contagion. For theoretical motivation, we turn to a recent paper by Brunnermeier and Pedersen (2009). In their model, an adverse shock to speculators’ funding liquidity (the availability of funding) forces them to reduce their leverage and provide less liquidity to the markets, which decreases asset liquidity (the ease with which assets trade). When the impact of the funding liquidity shock on asset liquidity is strong enough, asset values become depressed, which leads to further tightening of funding liquidity for speculators and causes a self-reinforcing liquidity spiral in which funding and asset liquidity 2 Electronic copy available at: http://ssrn.com/abstract=1868569 and asset values continue to deteriorate across various asset classes. While the predictions of their model may be most significant during a severe financial crisis like the credit crunch that began in 2007, we study periods of coincident poor performance in hedge funds since 1990 and show that large adverse shocks to asset and hedge fund funding liquidity make hedge fund contagion more likely. Specifically, we show that shocks to credit spreads, the TED spread, stock market liquidity, bank and prime broker stock performance, and hedge fund flows are positively correlated with the existence of hedge fund contagion. In this paper, we build on our prior analysis to examine contagion between hedge funds and main financial markets. Such an investigation is important in helping understand which hedge fund strategies are most affected by poor performance in the main markets. We conduct this examination using 10 hedge fund indices instead of 8 and consider three broad-based main market indices for equities, bonds, and currencies, and 15 sub-indices: 7 for equities, 3 for bonds, and 5 for currencies. Importantly, we link the existence of contagion between hedge funds and main markets to liquidity shocks. For the main market-hedge fund contagion analysis, we find strong evidence of contagion between hedge funds and a number of main markets including the broad-based equity index and the sub-indices for small-cap equities, mid-cap equities, Asian stocks, emerging market equities, high yield bonds, emerging market bonds, and the Australian Dollar. In addition, we find that for six markets, there is strong evidence linking the main market-hedge fund contagion to liquidity shocks. Using the expanded set of hedge fund indices, we also find evidence of contagion among hedge funds, a result that is strongly consistent with our previous work. Also consistent with our prior work, we find evidence of a significant linkage between liquidity shocks and contagion among hedge funds. Taking the current results together with our prior results, we have strong and comprehensive evidence for the existence of contagion not just within the hedge fund universe, but between hedge funds and some components of the main financial markets as well. Further, liquidity shocks appear to exacerbate both main market-hedge fund contagion as well as contagion among hedge funds. 3 The remainder of the paper is organized as follows. Section I describes the data. Section II documents the existence of contagion among hedge funds. Section III documents the existence of contagion between hedge funds and main markets. Section IV links liquidity shocks to both types of contagion, and Section V concludes. I. Data In this section, we first introduce our hedge fund data and then our data on the returns of the main financial markets. A. Hedge Fund Data The hedge fund style returns we use are the monthly style index returns provided by Hedge Fund Research (HFR). The returns are equally weighted and net of fees. The indices include both domestic and offshore funds. These data extend from January 1990 to October 2008 for a total of 226 monthly observations. We use ten single-strategy indices (Event Driven Distressed/Restructuring, Event Driven Merger Arbitrage, Equity Hedge Market Neutral, Equity Hedge Quantitative Directional, Equity Hedge Short Bias, Emerging Markets, Macro Systematic Diversified, Relative Value Fixed Income Convertible Arbitrage, Relative Value Fixed Income Corporate, and Relative Value Multi-Strategy). These indices include over 2,000 funds and exclude funds of funds. Since the publication of Boyson, Stahel, and Stulz (2010), HFR has reclassified and expanded their eight previous indices. They explain the reclassification as follows on their website: “The classifications reflect the evolution of strategic trends in the hedge fund industry, cognizant of the reality that over market cycles the classification system is likely to continue to evolve, as new opportunities attract investor capital. The objective of the system is to define pure strategy and sub-strategy buckets which can be used to characterize pure strategy return at each level of analysis, 4 to be used for purposes of quantitative index construction.”2 HFR did not restate the performance of any past index, but rather, merged the eight existing indices into the new indices, according to their new classification system. As with the prior eight indices, the new indices include all the funds in the HFR database, with no required minimum track record or asset size. Additionally, these indices are not directly investible; that is, they include some funds that are closed to new investors. To address backfilling and survivorship bias, when a fund is added to an index, the index is not recomputed with past returns of the new fund. Similarly, when a fund is dropped from an index, past returns of the index are left unchanged. While the new classification system includes more than 10 new indices, we chose the 10 indices we use in this paper based on the following criteria: first, all eight of the indices in the Boyson, Stahel, and Stulz (2010) paper must be included in the reclassified indices we use, and second, the index data must extend back to January 1990 so as to be comparable to the Boyson, Stahel, and Stulz (2010) paper. Panel A of Table I reports summary statistics for the hedge fund indices. The Equity Hedge Quantitative Directional (EHQD) index has the highest rate of return at about 1.15% per month, closely followed by the Emerging Market Index (EMKT) with a return of 1.14% per month. The Equity Hedge Short Bias Index (EHSB) has both the lowest monthly rate of return at about 40 basis points per month as well as the highest standard deviation of 5.68% per month. Given the bull market during most of the 1990s and much of the 2000s, the poor performance of the short bias index is not surprising. Correlations among all the indices are generally positive and statistically significant, consistent with the results of Boyson, Stahel, and Stulz (2010). The one exception is the Equity Hedge Short Bias index which, as expected, exhibits a negative and statistically significant correlation with the majority of the other indices. 2 See https://www.hedgefundresearch.com/index.php?fuse=indices_class&1295886859 for more information regarding the new classification system and for detailed descriptions of the new indices. 5 B. Main Markets Index Data We also gather data for a number of market indices. We use three main market factors: a broad-based equity market index (the Russell 3000), a broad-based bond index (the Lehman Brothers bond index), and a broad-based currency index (the change in the trade-weighted U.S. dollar exchange rate index published by the Board of Governors of the U.S. Federal Reserve System). In addition, we separately investigate sub-indices of the main markets. These include seven equity market sub-indices: the Russell 2000 for small-cap domestic stocks, the Wilshire U.S. mid-cap index for mid-cap domestic stocks, the S&P 500 index for large-cap domestic stocks, the Datastream total return Asia index, the Datastream total return Europe index, the MSCI Emerging Markets index, and the MSCI Emerging Markets Latin America Index. For fixed-income, there are 3 sub-indices: the Datastream investment grade corporate bond index, the Bank of America/Merrill Lynch high yield bond index, and the JP Morgan emerging markets bond index. Finally, for currencies, we use exchange rate data from the Board of Governors of the Federal Reserve System for the Australian dollar (AUS), the Swiss franc (SFR), the Euro (EUR), the British Pound (PD), and the Japanese Yen (YEN). All exchange rates are defined as the dollar price of the foreign currencies.3 Panel B of Table I reports summary statistics for the main markets indices and sub-indices. Here, the best performers are investment grade bonds (IGR) and Latin American equities (LAT) with monthly return of about 1.36% for each. Both these indices also have rather high standard deviations. The worst performer is the US Dollar Index (DLR), with a return of just three basis points per month. Correlations between the indices seem reasonable, with fairly high correlations among the equity indices, and low correlations between investment grade bonds (IGR) and most other indices. The US Dollar Index is, as expected, significantly correlated with the individual exchange rates, which in turn are also correlated among themselves. The Australian Dollar is positively correlated with the Russell 3000 and all the stock market subindices. The Asian and European stock markets are each positively correlated with four out of 3 We convert the published rates from the Board of Governors for the Swiss franc and the Japanese yen. Also, we use the ECU as the European currency before inception of the Euro. 6 five currencies. The Swiss franc, Euro and British Pound are all correlated with the investment grade bond index (IGR), and the Australian Dollar exhibits a positive correlation with both high yield and emerging bond market indices. Interestingly, the Japanese Yen is negatively correlated with these indices. II. Tests for Contagion Among Hedge Funds We first examine whether there is contagion among hedge fund returns. As a first look, Figure 1 plots the number of hedge fund indices (of a possible 10) that experience worst returns in a given month for the period January 1990 to October 2008. The periods with the most simultaneous extreme returns include the period around the Long-Term Capital Management crisis as well as the recent financial crisis. In the following, we use a lower 10% cutoff of the overall distribution of returns to identify worst returns among hedge fund styles (i.e., returns that are in the bottom decile of all returns for an index’s entire time series). Since there are 226 observations, we have 23 worst returns for each style. The analyses use a linear probability model with regressions performed separately for each index, in which the dependent variable is set to 1 if the hedge fund index of interest has a worst return, and zero otherwise.4 The independent variables include a number of control variables that have been shown to affect hedge fund performance, including the returns on the three broad-based main markets (stocks, bonds, and currencies), the monthly change in the 10-year constant maturity Treasury yield, based on Fama and French (1993) and Fung and Hsieh (2004), and the return on the Treasury Bill. We also use ABS (asset-based strategy) factors and a size spread factor (Wilshire Small Cap 1750 monthly return minus Wilshire Large Cap 750 monthly return), both from Fung and Hsieh (2004).5 The ABS factors are look-back straddles on bonds, currencies, commodities, short-term interest rates, and equities. We use changes in the VIX index to capture stock market volatility. Finally, we include a variable that represents the equally-weighted index of all hedge 4 Using more sophisticated regression approaches like binomial and multinomial logit models and explicitly controlling for autocorrelation and non-normality in returns generate qualitatively similar results. See Boyson, Stahel, and Stulz (2010). 5 We thank David Hsieh for making this data available. 7 fund returns excluding the returns on the hedge fund index in the dependent variable. This factor is included to control for the overall correlations among hedge fund indices. The independent variable of interest is called COUNT9 which is the number of other hedge fund indices (excluding the index represented by the dependent variable) that have an extreme worst return in a month. A positive and significant coefficient on COUNT9 provides evidence of contagion between the hedge fund index of interest and all other hedge fund indices. Table II presents results. The most striking result is that all ten indices exhibit evidence of hedge fund contagion as the coefficients on COUNT9 are positive and statistically significant at the 5% or better level. This result is strongly consistent with the results in Boyson, Stahel, and Stulz (2010), and provides additional evidence of hedge fund contagion, given the larger number of hedge fund indices examined in this paper. The estimates are also economically significant: a one standard deviation increase in COUNT9 increases the probability of observing an extreme negative return by about 12%. III. Tests for Contagion Between Hedge Funds and Main Markets We next examine whether there is contagion between hedge fund returns and returns on the main markets. As in the prior section, we use a lower 10% cutoff of the overall distribution of returns to identify worst returns among hedge funds and main market indices. In addition to the control variables described above, the independent variables also include indicator variables for each of the market indices, set to 1 if the index has a worst return, and zero otherwise. For each of the 10 hedge fund indices, we perform four different regression specifications for a total of 40 separate analyses. The first specification includes indicator variables for worst returns in the three broad-based main markets. The second includes indicator variables for each of the equity market sub-indices, excluding the Latin American index since it is subsumed by the Emerging Markets index, as well as indicators for the broad-based bond and currency indices. The third includes indicator variables for each of the bond sub-indices, as well as indicators for 8 the broad-based equity and currency indices, and the final set of regressions includes indicators for each of the currency sub-indices, as well as indicators for the broad-based equity and bond indices. Rather than tabulating 40 separate regressions, we perform four summary regressions that aggregate the above analyses. We report the summarized results in Table III, while describing the detailed results in the exposition. Each summary regression is a pooled time-series, cross-sectional regression across all ten hedge fund indices over the entire time period. We include fixed effects to provide a separate intercept for each of the hedge fund indices that controls for within-style variation for each hedge fund index. A positive and statistically significant coefficient on a main market indicator variable implies that there is contagion, on average, between hedge funds and the main market of interest. The first regression specification in Table III includes all the control variables, indicator variables for the three broad-based main markets, and the equally-weighted return on all hedge fund indices to control for normal correlations among indices. For brevity, the control variables and individual intercepts on each of the hedge fund indices are not reported. In Specification 1, there is strong evidence of contagion between the stock market index and hedge funds. An examination of the 10 detailed regressions for each hedge fund index (results not tabulated) indicates that 7 of the 10 separate indices (Event Driven Merger Arbitrage, Quantitative Directional, Short Bias, Emerging Markets, Macro Systematic Diversified Index, Relative Value Fixed Income Corporate, and Relative Value Multi-Strategy) have positive and significant coefficients on the Russell 3000 indicator variable. Turning to Specification 2, this analysis includes the six equity sub-indices (which exclude the Latin American index as it is subsumed by the Emerging Markets index) as independent variables. Four of these indices: Small-Cap Domestic, Mid-Cap Domestic, Asian Stock, and Emerging Markets show evidence of contagion with hedge funds. Examining the 10 detailed regressions for this specification, there is evidence of contagion between 4 of the 10 hedge fund indices (Event Driven Merger Arbitrage, Relative Value Fixed Income Corporate, Relative Value Convertible Arbitrage, and Relative Value MultiStrategy) and the small cap stock market index. Four of the 10 indices (Event Driven Distressed 9 Securities, Event Driven Merger Arbitrage, Quantitative Directional, and Macro Systematic Diversified) show evidence of mid-cap stock market contagion with hedge funds. By contrast, 2 of the 10 indices (Event Driven Merger Arbitrage and Quantitative Directional) show evidence of contagion between large cap stocks and hedge funds, while 2 of the 10 indices (Event Driven Distressed Securities and Relative Value Multi-Strategy) show a negative relationship between poor performance in the index and poor performance in large cap stocks, suggesting that these two indices may actually provide downside protection against extreme moves in large cap stocks for their investors. Two of the 10 indices (Relative Value Fixed Income Corporate and Relative Value Convertible Arbitrage) have contagion with the Asian stock index, and finally, only the Emerging Markets index shows evidence of emerging markets contagion with hedge funds. Taken together, we present fairly strong evidence of contagion between hedge funds and small and mid-cap stock markets, weaker evidence of contagion with Asian stock markets, and mixed evidence for contagion with large cap stocks: some indices show contagion but even more indices indicate weak evidence of downside protection. These results are consistent with academic work showing that hedge funds have significant exposure to small stocks (see, for example, Agarwal and Naik (2004), and Billio, Getmanksy, and Pellizzon (2010)). Specification 3 provides strong evidence of contagion between hedge funds and high yield bonds and weaker evidence of contagion between hedge funds and emerging market bonds. Examining the detailed regressions, 5 of the 10 indices (Event Driven Distressed Securities, Event Driven Merger Arbitrage, Relative Value Fixed Income Convertible Arbitrage, Relative Value Fixed Income Corporate, and Relative Value Multi-Strategy) provide evidence of contagion with high yield bonds. These results are intuitive in that 2 of the 5 hedge fund indices have a fixed-income focus, and a third has a distressed focus. Finally, only one index, Emerging Markets, shows contagion with the Emerging Markets bond index. Specification 4 provides evidence of contagion between hedge funds and the Australian Dollar. Specifically, the Emerging Market and the Equity Hedge Quantitative Directional indices exhibit significant positive exposure. One possible explanation for this contagion is that it may reflect the 10 unwinding of carry trades by hedge funds during poor economic times. There is also weak evidence of downside protection being provided by holding Japanese Yen in the aggregate regressions, but none of the 10 separate index regressions have a statistically significant result for the Japanese Yen. Given that these regressions are linear probability models and the explanatory variables are indicators, the estimated coefficients can directly be interpreted as increases in the average probability of observing an extreme negative return. Focusing on the largest significant coefficient in Specification 2, the estimate suggests that if the Dow Jones Mid-Cap Equity Index has an extreme return, the probability of experiencing an extreme return in the 10 hedge fund indices increases on average by 14.4%. Similarly for specifications 3 and 4, the results suggest that the probabilities increase on average by 18.8% and 4.0% with a large negative return in the BOA/ML Corporate High Yield Index and the Australian Dollar, respectively. In summary, the results from this section provide significant evidence of contagion between main markets and hedge funds, particular with respect to the broad-based equity index, the broad-based currency index, small and mid-cap stock markets, high yield bond markets, and the Australian Dollar. IV. Channels of Hedge Fund Contagion In this section, we explore the channels through which contagion takes place. In the first part of the section, we consider the channels of contagion among hedge funds. In the second part, we investigate the channels of contagion between hedge funds and the main markets. A. Contagion across Hedge Funds We are interested in identifying factors that explain the contagion we observe. Brunnermeier and Pedersen (2009) provide a theoretical framework from which to conduct our investigation. They argue that speculators, notably hedge funds, help smooth price fluctuations in markets that are caused by order imbalances across buyers and sellers by providing liquidity. In order to provide liquidity, speculators 11 must finance their trades through collateralized borrowing from financiers, including commercial and investment banks. As a result, speculators can face funding liquidity constraints, either through higher margins or a decline in the value of the assets they hold, or both. For example, in the case of a liquidity shock, a financier may raise its margin requirements, forcing the speculator to delever in a time of crisis, reducing prices and market liquidity even further (a margin liquidity spiral).6 Concurrently, any large positions that the speculator holds will lose value, which can lead to further margin calls at the now higher margin rate, causing further deleveraging in a weak market (a loss liquidity spiral). As a result, both funding liquidity and asset liquidity are diminished. One implication of Brunnermeier and Pedersen (2009) is that liquidity has commonality across securities because shocks to funding liquidity (capital constraints) affect all securities in which speculators are marginal investors.7 Empirically, there is support for the hypothesis of commonality in liquidity: Chordia, Sarkar, and Subrahmanyam (2005) document commonality in stock and bond market liquidity, Acharya, Schaefer, and Zhang (2008) document an increase in co-movement in CDS spreads during the GM/Ford downgrade period when dealers faced liquidity shocks, and Coughenour and Saad (2004) show that commonality in liquidity across stocks is higher for those handled by a NYSE specialist firm that faces funding constraints. Another implication is that since most market makers have net long positions, liquidity will tend to dry up most quickly when markets perform poorly, and will have a stronger relationship with asset prices during these times than during normal times. To test the implications of Brunnermeier and Pedersen (2009) for hedge fund contagion, we identify six contagion channel variables, that is, variables whose extreme adverse realizations are associated with a tightening of asset and hedge fund funding liquidity, and then test whether large negative shocks to 6 Brunnermeier and Pedersen (2009) note that the liquidity shock can be caused by a shock to liquidity demand, fundamentals, or volatility. 7 In addition, a large literature shows that shocks to liquidity and liquidity risk affect asset returns and that there is co-movement in liquidity and liquidity risk across asset classes (e.g., Amihud and Mendelson (1986), Amihud (2002), Pastor and Stambaugh (2003), Chordia, Sarkar, and Subrahmanyam (2005), and Acharya and Pedersen (2005)). 12 these variables can explain the hedge fund contagion we document. These variables include the Baa–10year Treasury constant maturity yield spread (CRSPRD), the Treasury–Eurodollar (TEDSPRD) spread, the liquidity measure of Chordia, Sarkar, and Subrahmanyam (CSS, 2005) (STKLIQ), a stock index for commercial banks (BANK), a stock index for prime brokers (PBI) and, finally, hedge fund redemptions (FLOW). The construction of these variables, their foundation in the literature, and their relationship to liquidity is detailed in Table IV. Additional proxies could be used. For instance, Dudley and Nimalendran (2011) find that futures margins constitute a useful proxy for funding liquidity. Table V presents summary statistics for the liquidity proxies. For all variables, the number of observations is the same as for the hedge fund indices. The correlations between the variables are positive and significant among the credit spread, TED spread, and stock liquidity variable, and between the bank and prime broker indices. The high correlation between the bank and prime broker indices is not too surprising, since the firms in these indices are all in the financial services industry. The correlation between hedge fund flows and the credit spread is negative and significant, indicating that flows are worse when credit spreads are high, while the credit spread correlation with prime broker stocks is also negative and significant, indicating that prime brokers perform poorly when credit spreads are high. Finally, the CSS liquidity measure (for which high values imply illiquidity) is negatively correlated with bank and prime broker stocks, indicating that these stocks perform poorly when stock markets are less liquid. To test whether large adverse shocks to liquidity can help explain hedge fund contagion, we create indicator variables for each of the six contagion channel variables. These indicators are set to one if the contagion channel variable has a realization in its lowest (highest) quartile over the time series and zero otherwise for variables that are positively (negatively) related to liquidity shocks. Hence, the prime broker index, bank index, and hedge fund flow indicator variables are set to one if the changes in their corresponding contagion channel variables are in the bottom 25% of all respective values, while the CSS 13 stock liquidity measure, the credit spread and the TED spread indicator variables are set to one if the realizations are in the top 25% of all respective values. To perform this analysis, we create a dependent variable COUNT10 defined as the number of hedge fund indices that experience worst returns in a given month and estimate a linear probability model. Since contemporaneous as well as lagged liquidity shocks could affect hedge fund returns, we also include lagged realizations of the indicator variables. Regressions are performed separately for each contagion channel variable, for a total of six regressions. The six regressions in Table VI include the relevant contemporaneous and lagged measures of the channel indicator variable. A positive and significant coefficient on a contagion channel indicator variable means that the variable is associated with an increased probability that hedge fund worst returns exhibit contagion. The regressions also include as control variables the returns on the three main market indices, the Treasury Bill return, the small stock factor, 10-year CMT, changes in VIX, the Fung and Hsieh (2004) ABS factors, and extreme return indicators for each of the three main market indices. Consistent with the results from Boyson, Stahel, and Stulz (2010), Table VI provides strong evidence that liquidity shocks are correlated with hedge fund contagion. Notably, contemporaneous shocks to the credit spread, the TED spread, the prime broker stock index, and hedge fund flows are positively linked to hedge fund contagion. Additionally, lagged shocks to the TED spread, stock market liquidity, bank stocks, and prime broker stocks are also positively linked to hedge fund contagion, implying that some liquidity shocks affect hedge fund performance with a lag. Since the liquidity variables are not perfectly correlated, the results also imply that liquidity shocks manifest themselves in a variety of ways that all appear to affect hedge fund contagion.8 To give these statistically significant results an economic magnitude, the 8 We also rerun these analyses adding two more lags of the shock variables, and find that the twice-lagged shocks to the prime broker index, bank stock index, and stock market liquidity are correlated with hedge fund contagion. However, thrice-lagged shocks are never correlated with contagion. These results are available by the authors upon request. 14 number of hedge fund styles experiencing an extreme return increases by 100% when any of the dummy variables indicate tight funding or asset liquidity. Figure 2 overlays the number of contagion channel variables that have extreme realizations in a given period on Figure 1, which shows the number of hedge fund indices experiencing extreme returns each month. The number of contagion channel variables is called LIQCOUNT6, which is the count of the number of liquidity shocks across all 6 contagion channel variables in a given month, with a minimum value of 0 and a maximum value of 6. The number of hedge fund styles that simultaneously experience an extreme return are represented by increasingly larger circles. Consistent with the regression results, hedge fund contagion is exacerbated when liquidity is low, that is, generally the larger LIQCOUNT6, the larger the circles. B. Contagion between Hedge Funds and Main Markets In this section, we attempt to link shocks to liquidity channel variables to the contagion between main markets and hedge funds that we document in Section III. To perform this test we create new variables that measure the strength of the contagion between each main market index and sub-index and hedge funds in general. These variables are created by interacting the COUNT10 variable (which is the number of hedge fund indices that experience worst returns in a given month and ranges from 0 to 10) with indicator variables for each main market index that are set to 1 if the main market index also experiences a worst return and zero otherwise. For example, if during a particular month the COUNT10 variable has a value of 4 and the Russell 3000 index experiences simultaneously a worst return, the value of the interaction variable would be 4. If however the Russell 3000 index does not have a worst return, the interaction variable is zero. Therefore, the interaction variable is zero in the absence of contagion and increases with the extent of the contagion. This approach results in 18 new interaction variables, one for each of the 3 broad-based main market indices, 7 for the stock market sub-indices, 3 for the bond market sub-indices, and 5 for each currency index. The main independent variable is our summary indicator 15 variable for shocks to liquidity, called LIQCOUNT6. As noted earlier, this variable tabulates the number of liquidity shocks to the six contagion channel variables in a particular month. A positive and significant coefficient on this variable implies a correlation between liquidity shocks and hedge fund-main market contagion. Results are presented in Table VII. We focus on the main market indices and sub-indices that show evidence of contagion in Table III. These include the Russell 3000 broad-based stock market index, the small-cap and mid-cap domestic stock sub-indices, the Asian and emerging markets equity sub-index, the high yield bond index, and the emerging market bond, the Australian Dollar. Hence, if in Table VII the coefficients on the liquidity shock or lagged liquidity shock indicator variables are positive and significant for the broad-based market indices and market sub-indices that exhibit contagion in Table III, we can interpret this result as evidence that liquidity shocks are linked to contagion between hedge funds and main markets for these main market indices in the sense that if liquidity is scarce, we are more likely to observe contagion between hedge funds and the main markets. The results in Table VII suggest that for six of the eight main market indices and sub-indices that exhibit contagion with hedge funds in Table III, contagion is connected to liquidity shocks, including the Russell 3000 broad-based stock market index, the small- and mid-cap domestic stock market index, the Asian stock market index, and the high yield and emerging bond indices. In addition, the broad-based currency index, the large-cap domestic and the European Stock markets, and the British Pound have positive and significant coefficients indicating that liquidity shocks also exacerbate contagion in these markets.9 The economic relevance of these results can again be directly inferred from the coefficients. For example, for every additional liquidity variable that exhibits a large shock (a one tick increase in 9 We also perform a robustness check where we include two more lags of the liquidity shock indicator variable. For the twice-lagged variable, we find that contagion is connected to liquidity shocks for European stock markets, the small-cap domestic market, domestic mid-cap, and emerging markets. For the thrice-lagged liquidity variable, we find that contagion is connected to liquidity shocks for the Euro and British Pound indices. Detailed results are available from the authors upon request. 16 LIQCOUNT6), the number of hedge fund index worst returns occurring simultaneously with a worst return in the Russell 3000 increases by about 40%. V. Implications and Conclusions In this paper, we use a linear probability regression model to study contagion among hedge funds and between hedge funds and main markets for the period January 1990 to October 2008. We find strong evidence for the existence of contagion both among different styles of hedge funds, confirming the results of Boyson, Stahel, and Stulz (2010), and between hedge funds and main markets, especially for small cap stocks, high yield bonds, the Asian stock market, and the Australian Dollar. We also find evidence that this contagion is linked to liquidity shocks, including shocks to credit spreads and the TED spread, shocks to stock market liquidity, poor performance in bank and prime broker stocks, and finally, shocks to hedge fund flows. These results have important implications for hedge fund managers and hedge fund investors. First, investors in hedge funds that diversify across different hedge fund styles expecting protection against poor performance might be disappointed since all hedge fund styles tend to experience poor performance simultaneously. Second, investors that use hedge funds as diversification tools against poor returns in main markets should also be careful. Specifically, if future crises are similar to past crises, hedge funds do not provide diversification benefits for the broad stock market, for small-cap, mid-cap, Asian, or emerging market stocks, for high yield bonds or Asian bonds, or for Australian Dollar, and in fact, tend to move closely with these markets during crises. That said, certain hedge fund styles do appear to provide some diversification against poor returns in main markets. Notably, the Equity Hedge Market Neutral (results not tabulated) style has no significant exposure to any of the main markets. The results linking liquidity shocks to contagion also have important policy implications. While our tests do not allow us to conclude that liquidity shocks actually cause financial contagion, the correlation is important in that crises that also involve large liquidity shocks are particularly worrisome in terms of 17 financial contagion, both within the hedge fund industry and between hedge funds and main markets. These results are consistent with Bordo and Haubrich (2009), who find that recessions that also involve credit crunches are more severe. Hence, if a central bank is considering intervening in financial markets during an economic downturn, the success of the potential intervention could be affected by the perceived impact on market and funding liquidity. 18 References Acharya, Viral V., and Lasse H. Pedersen, 2005, Asset pricing with liquidity risk, Journal of Financial Economics 77, 375-410. Acharya, Viral V., Stephen M. Schaefer, and Yili Zhang, 2008, Liquidity risk and correlation risk: A clinical study of the General Motors and Ford downgrade of 2005, working paper, London Business School. Adrian, Tobias, and Markus K. Brunnermeier, 2009, CoVar, working paper, Princeton University. Agarwal, Vikas, and Narayan J. Naik, 2004, Risk and portfolio decisions involving hedge funds, Review of Financial Studies 17, 63-98. Bae, Kee Hong, G. Andrew Karolyi, and René M. Stulz, 2003, A new approach to measuring financial contagion, Review of Financial Studies 16, 717-764. Bekaert, Geert, Campbell R. Harvey, and Angela Ng, 2005, Market integration and contagion, Journal of Business 78, 39-69. Billio, Monica, Mila Getmansky, and Loriana Pelizzon, 2010, Crises and hedge fund risk, working paper, University of Massachusetts. Bordo, M. D. and J. G. Haubrich. 2009. Credit Market Turmoil, Monetary Policy and Business Cycles: A Historical Overview. Working paper. Boyson, Nicole M., Christof Y. Stahel, and Rene M. Stulz, 2010. Hedge Fund Contagion and Liquidity Shocks, Journal of Finance 65(5), 1789-1816. Brunnermeier, Markus K., and Lasse Heje Pedersen, 2009, Market liquidity and funding liquidity, Review of Financial Studies 22, 2201-2238. Campbell, John Y., and Glen B. Taksler, 2003, Equity volatility and corporate bond yields, Journal of Finance 58, 2321-2350. Chan, Kevin Nicholas, Mila Getmansky, Shane Haas, and Andrew W. Lo, 2006, Systemic risk and hedge funds, in Mark Carey and René M. Stulz, ed. The Risks of Financial Institutions (University of Chicago Press). Chordia, Tarun, Asani Sarkar, and Avinidhar Subrahmanyam, 2005, An empirical analysis of stock and bond market liquidity, Review of Financial Studies 18, 85-129. Cifuentes, Rodrigo, Gianluigi Ferrucci, and Hyun Song Shin, 2005, Liquidity Risk and Contagion, Journal of the European Economic Association, 3, 556-566. Coughenour, Jay F., and Mohsen Saad, 2004, Common market makers and commonality in liquidity, Journal of Financial Economics 73, 37-70. 19 Dick-Nielsen, Jens, Peter Feldhütter, and David Lando, 2009, Corporate bond liquidity before and after the onset of the subprime crisis, working paper, Copenhagen Business School. Dudley, Evan, and Mahendrarajah Nimalendran, 2011, Margins and hedge fund contagion, Journal of Financial and Quantitative Analysis, forthcoming. Fama, Eugene F., and Kenneth R. French, 1993, Common risk factors in the returns on stocks and bonds, Journal of Financial Economics 33, 3-56. Fung, William, and David A. Hsieh, 2004, Hedge fund benchmarks: A risk-based approach, Financial Analyst Journal 60, 65-80. Goyenko, Ruslan Y., Craig W. Holden, and Charles A. Trzcinka, 2009, Do liquidity measures measure liquidity?, Journal of Financial Economics 92, 153-181. Gupta, Anurag, and Marti G. Subrahmanyam, 2000, An empirical examination of the convexity bias in the pricing of interest rate swaps, Journal of Financial Economics 55, 239-279. Longstaff, Francis A., Sanjay Mithal, and Eric Neis, 2005, Corporate yield spreads: Default risk or liquidity? New evidence from the credit default swap market, Journal of Finance 60, 2213-2253. Taylor, John B., and John C. Williams, 2009, A black swan in the money market, American Economic Journal: Macroeconomics 1, 58-83. 20 Table I Summary Statistics of Monthly Returns on HFR Indices and Market Factors: January 1990 to October 2008 Summary statistics for monthly returns for ten HFR hedge fund indices and eighteen market factors used in the paper are reported below. Panel A reports data on hedge fund indices, which include Event Driven Distressed Securities (EDDS), Event Driven Merger Arbitrage (EDMA), Equity Hedge Market Neutral (EHMN), Equity Hedge Quantitative Directional (EHQD), Equity Hedge Short Bias (EHSB), Emerging Markets (EMKT), Macro Systematic Diversified (MACS), Relative Value Fixed Income Convertible Arbitrage (RVCA), Relative Value Fixed Income Corporate (RVCI), and Relative Value Multi-Strategy (RVMS), and are described more fully in Section II. Panel B reports data on the main market indices which include three broad-based indices: the Russell 3000 (R3K), the Lehman Brothers Bond Index (LBD), the change in the trade-weighted U.S. dollar exchange rate index (DLR), seven stock indices: the Russell 2000 (SML), the Wilshire U.S. mid-cap index (MID), the S&P 500 index (LGE), the Datastream total return Asia index (ASA), the Datastream total return Europe index (ERS), the MSCI Emerging Markets index (EMS), the MSCI Emerging Markets Latin America Index (LAT), three bond indices: the Datastream investment grade corporate bond index (IGR), the Bank of America/Merrill Lynch high yield bond index (HYD), the JP Morgan emerging markets bond index (EMB), and 5 currency indices: the Australian dollar index (AUS), the Swiss Franc index (SFR), the Euro index (EUR), the British Pound index (PD), and the Japanese Yen index (YEN). All market indices are from Datastream. Correlations between the variables, the autocorrelations, as well as Jarque-Bera test statistics for normality are reported below the summary statistics. A * indicates significance at the 5% level. Panel A EDDS EDMA EHMN Median Standard deviation 1.019 1.850 0.755 1.251 0.664 0.922 EDDS EDMA EHMN EHQD EHSB EMKT MACS RVCA RVCI RVMS 1.000 0.585* 1.000 0.309* 0.319* 1.000 EHQD EHSB 1.150 0.383 3.977 5.684 Correlations 0.662* 0.577* 0.274* 1.000 -0.481* -0.402* -0.120 -0.864* 1.000 21 EMKT MACS RVCA RVCI RVMS 1.138 4.261 1.057 2.115 0.590 1.762 0.618 1.936 0.671 1.263 0.684* 0.498* 0.199* 0.735* -0.569* 1.000 0.295* 0.302* 0.251* 0.660* -0.409* 0.464* 1.000 0.647* 0.500* 0.331* 0.442* -0.504* 0.489* 0.096 1.000 0.806* 0.572* 0.208* 0.602* 0.010 0.610* 0.220* 0.643* 1.000 0.755* 0.469* 0.360* 0.610* -0.560* 0.649* 0.221* 0.760* 0.797* 1.000 Table I Summary Statistics of Monthly Returns on HFR Indices and Market Factors: January 1990 to October 2008, continued Panel B Main Market Indices R3K LBD DLR SML MID LGE ASA ERS EMS LAT IGR HYD EMB AUS SFR EUR PD YEN Median Std. Dev. 0.72 4.25 0.55 1.10 0.03 1.91 0.66 5.41 0.86 5.03 0.54 4.18 0.13 6.13 0.75 4.74 0.68 6.90 1.37 8.85 1.36 6.79 0.58 2.23 0.56 1.10 0.18 2.71 0.09 2.44 0.05 2.30 0.20 2.74 0.18 2.71 R3K LBD DLR SML MID LGE ASA ERS EMS LAT IGR HYD EMB AUS SFR EUR PD YEN 1.00 0.17* 1.00 0.15* 0.24* 1.00 0.83* 0.06 0.11* 1.00 0.84* 0.10 0.09 0.87* 1.00 0.99* 0.18* 0.15* 0.77* 0.79* 1.00 0.51* 0.07 0.40* 0.45* 0.46* 0.50* 1.00 0.76* 0.13 0.42* 0.66* 0.67* 0.75* 0.62* 1.00 0.62* 0.02 0.12 0.61* 0.59* 0.60* 0.48* 0.59 0.88* 1.00 0.16 0.99* 0.24* 0.05 0.08 0.17* 0.06 0.11 -0.01 0.01 1.00 0.61* 0.33* 0.17* 0.60* 0.53* 0.59* 0.37* 0.58* 0.56* 0.51* 0.32* 1.00 0.38* 0.16* 0.03 0.36* 0.33* 0.36* 0.13 0.29* 0.41* 0.41* 0.15* 0.34* 1.00 0.21* -0.02 0.46* 0.22* 0.17* 0.20* 0.32* 0.39* 0.40* 0.30* -0.02 0.36* 0.13* 1.00 -0.04 0.18* 0.60* -0.09 -0.06 -0.03 0.09 0.12* -0.03 -0.09 0.17* -0.06 -0.01 0.28* 1.00 0.04 0.21* 0.64* -0.02 -0.01 0.05 0.15* 0.24* 0.08 0.01 0.20* 0.08 0.03 0.42* 0.92* 1.00 0.07 0.21* 0.52* -0.05 0.00 0.09 0.19* 0.28* 0.11* 0.06 0.20* 0.07 0.07 0.37* 0.71* 0.80* 1.00 -0.03 0.05 0.42* -0.07 -0.04 -0.02 0.24* 0.02 -0.02 -0.08 0.05 -0.17* -0.12* 0.12* 0.49* 0.39* 0.27* 1.00 Correlations 0.69* 0.00 0.19* 0.67* 0.65* 0.66* 0.63* 0.70 1.00 22 Table II Contagion Among Hedge Funds The event of a worst return in each hedge fund style is modeled as a linear probability model. The dependent variable is an indicator set to 1 if the hedge fund index has a worst return and zero otherwise. The regressions include a number of control variables, described in Section II. COUNT9 is the independent variable of interest, and represents the number of hedge funds, other than the hedge fund represented by the dependent variable, that have worst returns in a given month. The indices modeled include Event Driven Distressed Securities (EDDS), Event Driven Merger Arbitrage (EDMA), Equity Hedge Market Neutral (EHMN), Equity Hedge Quantitative Directional (EHQD), Equity Hedge Short Bias (EHSB), Emerging Markets (EMKT), Macro Systematic Diversified (MACS), Relative Value Fixed Income Convertible Arbitrage (RVCA), Relative Value Fixed Income Corporate (RVCI), and Relative Value Multi-Strategy (RVMS). Below the coefficients are the p-values in parentheses. Coefficients with ***, ** , and * are statistically significant at the 1%, 5%, and 10% levels, respectively. EDDS Intercept COUNT9 Adj. R2 Controls? -0.032 (0.444) 0.153*** (0.000) 0.584 Yes EDMA 0.160*** (0.004) 0.039** (0.016) 0.404 Yes EHMN 0.139** (0.015) 0.083*** (0.000) 0.244 Yes EHQD -0.077** (0.045) 0.055*** (0.000) 0.653 Yes EHSB -0.066 (0.260) 0.047** (0.013) 0.215 Yes 23 EMKT 0.023 (0.651) 0.076*** (0.000) 0.357 Yes MACS -0.025 (0.674) 0.042** (0.029) 0.181 Yes RVCA 0.230*** (0.000) 0.046*** (0.008) 0.317 Yes RVCI -0.041 (0.373) 0.094*** (0.000) 0.495 Yes RVMS 0.088* (0.063) 0.086*** (0.000) 0.459 Yes Table III Contagion Between Hedge Funds and Main Markets The event of a worst return in each hedge fund style is modeled as a pooled cross-sectional time series linear probability model. The dependent variable is an indicator set to 1 if the hedge fund index has a worst return and zero otherwise. The fixed effects regressions include a number of control variables, described in Section III, as well as a separate intercept for each of the hedge fund indices. Specification 1 includes as independent variables worst return indicator variables for the broad-based stock, bond, and currency markets. Specification 2 includes as independent variables worst return indicator variables for the seven stock market subindices. Specification 3 includes as independent variables worst return indicator variables for the three bond market subindices. Finally, Specification 4 includes as independent variables worst return indicator variables for the five currency subindices . All subindices are described in detail in Section III. Below the coefficients are the p-values in parentheses. Coefficients with ***, **, and * are statistically significant at the 1%, 5%, and 10% levels, respectively. Spec.1: Broad Indices Spec. 2: Stock Subindices Spec. 3: Bond Subindices Spec. 4: Curr. Subindices Worst Return Indicator Variables IND: Russell 3000 Index IND: Lehman Brothers Bond Index IND: Δ in FRB Dollar Index 0.216*** (0.000) -0.018 (0.508) 0.042 (0.125) IND: Russell 2000 Index IND: Dow Jones Mid-Cap Equity Index IND: S&P 500 Index IND: Datastream Asian Equity Index IND: Datastream Europe Equity Index IND: MSCI Em. Markets Equity Index 0.097*** (0.005) 0.144*** (0.000) 0.004 (0.915) 0.056** (0.020) 0.030 (0.351) 0.070*** (0.010) IND: Inv. Grade Corp Bond -0.020 (0.457) 0.188*** (0.000) 0.041* (0.077) IND: BOA/ML Corp. High Yield Index IND: JPM Emerging Market Bond Index IND: Australian Dollar IND: Swiss Franc IND: Euro IND: British Pound IND: Japanese Yen R2 Includes Controls? Fixed Effects? 0.193 Yes Yes 24 0.212 Yes Yes 0.215 Yes Yes 0.040* (0.089) 0.015 (0.546) 0.002 (0.933) -0.007 (0.771) -0.045** (0.031) 0.196 Yes Yes Table IV Contagion Channel Variables This table presents detail on the contagion channel variables from Section IV. Prior Literature Using this Variable Longstaff, Mithal, and Neis (2005), Dick-Nielsen, Feldhütter, and Lando (2009) Relationship to Liquidity Inverse Increased spreads imply higher borrowing costs and/or higher credit risk. Gupta and Subrahmanyam (2000), Campbell and Taksler (2003), Taylor and Williams (2009) Inverse Higher trading costs imply lower liquidity. Other common liquidity measures include Amihud (2002), Pastor and Stambaugh (2003), and Acharya and Pedersen (2005); we choose the measure based on recent work by Goyenko, Holden, and Trzcinka (2009) suggesting that bid-ask spreads are the most appropriate measure of liquidity. Chordia, Sarkar, and Subrahmanyam (2005), Goyenko, Holden, and Trzcinka (2009) Inverse Shocks that decrease the financial strength of hedge fund intermediaries could be transmitted to hedge funds through increased margin requirements as they curtail their lending. Chan, et. al. Direct PBI: Monthly change in the equallyweighted stock price index of prime broker firms including Goldman Sachs, Morgan Stanley, Bear Stearns, UBS AG, Bank of America, Citigroup, Merrill Lynch, Lehman Brothers, Credit Suisse, Deutsche Bank, and Bank of New York Mellon, adjusted for mergers and including bankruptcy returns. Data are from CRSP. Shocks that decrease the financial strength of hedge fund intermediaries could be transmitted to hedge funds through increased margin requirements as they curtail their lending. N/A Direct FLOW: Monthly change in hedge fund outflows as a percentage of assets under management calculated from individual hedge fund data from Lipper TASS and matched to HFR index data based on style description. We use both contemporaneous FLOWt and one-month-ahead FLOWt+1 since many hedge funds have redemption notice periods. Redemption requests force hedge funds to liquidate more assets than required to meet redemptions if they are levered, and make it harder for hedge funds to borrow. Redemption requests may come about from poor performance, shifts in sentiment, or other reasons. An alternative to accepting redemption requests is to put up redemption gates, which are unpopular. N/A Direct Variable and Source CRSPRD: Change in BAA-10-year Constant Maturity Treasury credit spread from the Federal Reserve Board’s website. Basis for Inclusion Increased spreads imply higher borrowing costs and/or counterparty risk. TEDSPRD: Change in Treasury-Eurodollar (TED) spread from the Federal Reserve Board’s website. STKLIQ: Change in average round-trip cost of a trade on the NYSE within a month; calculated as the monthly average of daily changes of the NYSE stock market liquidity after removing deterministic day-of-theweek effects and effects related to changes in tick size. The daily changes are calculated from daily cross-sectional valueweighted averages of individual stock proportional bid-ask spreads. BANK: Monthly change in the equally weighted stock price index of large commercial banks from Datastream. 25 Table V Summary Statistics for Contagion Channel Variables: January 1990 to October 2008 Summary statistics for monthly data on six contagion channel variables are described below. The variables include: the monthly percent change in the Baa-10-year CMT spread, the monthly percent change in the Treasury-Eurodollar (TED) spread, the monthly percent change in the Chordia, Sarkar, and Subrahmanyam (2005) liquidity measure, the monthly percent change in hedge fund flows as a percentage of assets (contemporaneous), the monthly returns from the Datastream Bank Index, and the monthly returns from the Prime Broker Index. Further description of these variables is in Section III. The number of observations is 226. Correlations between the variables are reported below the summary statistics. A * indicates significance at the 5% level. Mean Standard deviation Baa-10y Treasury CMT Spread TED Spread CSS Liquidity Measure Contemp. Hedge Fund Flows Bank Index Prime Broker Index Baa-10-year CMT Spread 1.073 10.731 1.000 CSS Liquidity Measure TED Spread 0.016 0.183 Correlations 0.359* 1.000 26 Contemporaneous Hedge Fund Flows Bank Index 0.001 0.761 1.067 2.249 1.023 6.061 0.365* 0.255* 1.000 -0.244* -0.126 -0.105 1.000 -0.130 -0.011 -0.291* 0.065 1.000 Prime Broker Index 1.660 7.426 -0.262* -0.070 -0.367* 0.065 0.866 1.000 Table VI Liquidity Shocks and Contagion Across Hedge Funds The co-occurrence of extreme monthly negative returns in hedge fund style indices is modeled as the outcome of a variable (COUNT10) that takes a value of 0 to 10 and represents the number of hedge fund index returns that experience worst returns in a given month. A monthly return is classified as a “worst return” if it belongs to the bottom 10% of all returns of that style. The regressions also include the continuous contagion channel variables and indicator variables corresponding to contemporaneous negative quartile realizations of the contagion channel variables. The contagion channel variables and their corresponding indicator variables include: the change in the Baa-10-year CMT Credit Spread (CRSPRD), the change in the Treasury-Eurodollar spread (TEDSPRD), the change in the Chordia, Sarkar, Subramanyam (2005) Liquidity Measure (STKLIQ), flows from other hedge funds (FLOW), the monthly change in the Datastream Bank Index (BANK), and the monthly change in the Prime Broker Index (PBI). The p-values are shown below the coefficients in parentheses. Coefficients with ***, **, and * are statistically significant at the 1%, 5%, and 10% levels, respectively. CRSPRD Constant Liquidity Shock Indicator Variable Liquidity Shock Indicator Variable, lagged Adj. R2 Includes Controls? Fixed Effects? TEDSPRD STKLIQ FLOW BANK PBI 0.809*** (0.003) 0.933*** (0.000) 0.763*** (0.006) 0.898*** (0.000) 0.858*** (0.001) 0.745*** (0.004) 0.752*** (0.006) 0.763*** (0.000) 0.304 (0.198) 0.542** (0.015) 0.400 (0.114) 0.435* (0.090) 0.156 (0.460) 0.512** (0.011) 0.479** (0.025) 0.258 (0.238) 0.778*** (0.000) 0.791*** (0.000) 0.506 Yes No 0.535 Yes No 0.504 Yes No 0.512 Yes No 0.527 Yes No 0.527 Yes No 27 Table VII Liquidity Shocks and Contagion Between Hedge Funds and Main Markets Contagion between hedge funds and main markets is linked to liquidity shocks using a linear probability model. The dependent variable is COUNT10 interacted with an indicator variable set to one if a main market or main market sub-index has a worst return, and zero otherwise. These variables are denoted as “i” followed by the name of the index. For example, iRUS3000 is the interaction of COUNT10 and an indicator variable for a worst return in the Russell 3000 index. The regressions include a number of control variables, described in Section IV. The index and sub-index interaction variables include three broad-based market indices: the Russell 3000 (iR3000), the Lehman Brothers Bond Index (iLBBD), the change in the trade-weighted U.S. dollar exchange rate index (iDLR), seven stock market indices: the Russell 2000 (iSML), the Wilshire U.S. mid-cap index (IMID), the S&P 500 index (iLARGE), the Datastream total return Asia index (iASST), the Datastream total return Europe index (iEUST), the MSCI Emerging Markets index (iEMST), the MSCI Emerging Markets Latin America Index (iLATIN), three bond market indicies: the Datastream investment grade corporate bond index (iIGRD), the Bank of America/Merrill Lynch high yield bond index (iHYLD), the JP Morgan emerging markets bond index (iEMBD), and five currency indices: the Australian dollar index (iAUST), the Swiss Franc index (iSWFR), the Euro index (iEURO), the British Pound index (iPOUND), and the Japanese Yen index (iYEN). The independent variable of interest is LIQCOUNT6, which is a variable that takes a value from 0 to 6 and counts the number of contagion channel variables that experience a liquidity shock in a particular period. Both contemporaneous and one-month lagged measures of LIQCOUNT6 are included in the regressions. Below the coefficients are the p-values in parentheses. Coefficients with ***, **, and * are statistically significant at the 1%, 5%, and 10% levels, respectively. Panel A: Broad-Based Markets and Equity Subindices Broad-Based Indices iR3000 iLBBD iDLR Intercept LIQCOUNT6 LIQCOUNT6, lagged Adj. R2 Includes Controls? Fixed Effects? iSML iMID iLGE Equity Subindices iASST iEUST iEMST iLATIN 0.381* (0.091) 0.124* (0.086) 0.116* 0.432*** (0.007) 0.015 (0.773) -0.005 0.191 (0.196) 0.081* (0.097) 0.061 0.317 (0.173) 0.070 (0.342) 0.175*** 0.200 (0.426) 0.136* (0.089) 0.054 0.413* (0.055) 0.050 (0.462) 0.117** 0.1354 (0.561) 0.025 (0.734) 0.140** 0.294 (0.193) 0.055 (0.445) 0.152** 0.176 (0.444) 0.109 (0.138) 0.060 0.588*** (0.003) 0.043 (0.493) 0.000 (0.054) 0.522 Yes (0.907) 0.411 Yes (0.117) 0.343 Yes (0.005) 0.503 Yes (0.414) 0.373 Yes (0.041) 0.521 Yes (0.025) 0.387 Yes (0.012) 0.500 Yes (0.325) 0.450 Yes (0.999) 0.494 Yes No No No No No No No No No No 28 Table VII Liquidity Shocks and Contagion Between Hedge Funds and Main Markets, continued Panel B: Bond and Currency Subindices Fixed Income Subindices iIGRD iHYLD iEMBD Intercept LIQCOUNT6 LIQCOUNT6, lagged Adj. R2 Includes Controls? Fixed Effects? 0.432*** (0.007) 0.015 (0.773) -0.005 (0.907) 0.411 Yes No 0.232 (0.291) 0.105 (0.150) 0.212*** (0.000) 0.524 Yes No 0.188 (0.357) 0.035 (0.600) 0.107* (0.054) 0.430 Yes No 29 iAUST 0.221 (0.224) -0.012 (0.844) 0.052 (0.279) 0.500 Yes No Currency Subindices iSWFR iEURO iPD 0.054 (0.616) -0.007 (0.832) 0.038 (0.187) 0.075 Yes No 0.238 (0.128) -0.037 (0.475) 0.048 (0.248) 0.384 Yes No 0.355** (0.049) -0.078 (0.184) 0.096** (0.045) 0.368 Yes No iYEN -0.019 (0.807) -0.006 (0.806) -0.006 (0.767) 0.015 Yes No Figure 1. Number of Hedge Fund Styles that have Worst Returns by Month. The number of hedge fund styles that experience a worst return each month is plotted, by month, for the period January 1990 to October 2008. 30 Figure 2. Hedge Fund Contagion and Liquidity Shocks. The number of contagion channel variables experiencing extreme realizations (LIQCOUNT6) is plotted, by month, for the period January 1990 to October 2008, on the left axis. The number of hedge fund style indices that experience a worst return month are depicted circles. The larger a circle the more styles experience such an event simultaneously. 31