The Behavior of Hedge Funds during Liquidity Crises
Itzhak Ben-David
Fisher College of Business, The Ohio State University
Francesco Franzoni
Swiss Finance Institute and University of Lugano
Rabih Moussawi
Wharton Research Data Services, The Wharton School, University of Pennsylvania
April 2010
Abstract
We study hedge funds’ trading patterns in the stock market during liquidity crises. On average at the time
of a crisis, hedge funds reduce their equity holdings by 9% to 11% per quarter (around 0.3% of total
market capitalization). This effect results from large selling by up to a quarter of hedge funds and is not
offset by other hedge funds expanding their positions. Dramatic selloffs took place in the 2008 crisis:
hedge funds sold about 30% of their stock holdings and almost every fourth hedge fund sold more than
40% of its equity portfolio. We identify two main drivers of this behavior. First, we impute about half of
the variation in equity selloffs to a response to lender and investor funding withdrawals. Second, it
appears that hedge funds mobilize capital to other (potentially less liquid) markets in the pursuit of more
profitable investment opportunities.
_____________________
*
We thank Viral Acharya, Giovanni Barone-Adesi, Vyacheslav Fos, José-Miguel Gaspar, Massimo Massa, Ronnie
Sadka, René Stulz, Dimitri Vayanos, and seminar participants at the Ohio State University, 2nd Annual Conference
on Hedge Funds in Paris, and Wharton/FIRS pre-conference for helpful comments.
1
1.
Introduction
It is widely believed that hedge funds provide liquidity to markets (see, e.g., Agarwal,
Fung, Loon, and Naik 2007). According to this view, hedge funds engage in trades when the
demand for risky assets is low, and thus reduce the liquidity premium on assets that are less
desirable and eliminate market mispricing. For example, Brophy, Paige, and Sialm (2009)
present evidence that hedge funds provide liquidity in niche assets when other classes of
investors are reluctant to invest due to the high degree of information asymmetry. Consistent
with this view, Khandani and Lo (2009) document that returns of hedge funds are correlated with
returns of illiquid assets, and Aragon (2007) and Sadka (2009) find that hedge funds earn premia
related to the liquidity level and risk, respectively.
However, theories by Shleifer and Vishny (1997), Gromb and Vayanos (2002), Vayanos
(2004), and Brunnermeier and Pedersen (2009) argue that hedge funds may not be able to
provide liquidity during periods of low aggregate liquidity because trading capital tends to dry up
at these times. 1 These theories, if correct, have policy implications. Since liquidity provision is
an important function in general and crucial in periods of market stress, regulators have
traditionally maintained a low level of supervision on hedge funds (Ackermann, McEnally, and
Ravenscraft 1999). Hedge funds, however, may impose three negative externalities to the
financial system in the form of: (1) counterparty risk to other financial intermediaries, (2) the
ability to move prices further away from fundamentals, (3) synchronized capital erosion which
compromises aggregate liquidity. Among other things, these externalities may be the outcome of
runs on hedge funds’ assets. To prevent this, the regulatory debate has considered limits on
investors’ ability to withdraw their funds (Acharya, Pedersen, Philippon, and Richardson 2009).
The limited empirical evidence on hedge fund trading during a crisis seems to confirm
that hedge funds are affected by funding liquidity. Cao, Chen, Liang, and Lo (2009) find that
hedge funds manage their portfolios to reduce market exposure at times of low liquidity.
Hameed, Kang and Viswanathan (2010) show that stock-level liquidity drops following stock
price declines. They argue that this relation is driven by the financial constraints that arbitrageurs
1
Brunnermeier and Pedersen (2009) suggest that negative shocks to arbitrageurs’ trading capital (“funding
liquidity”) limit their ability to reduce deviations of prices from fundamentals (“market liquidity”). In their model, a
dry-up in funding liquidity and deterioration in market liquidity mutually reinforce each other (“liquidity spirals”).
In the process, arbitrageurs rebalance their portfolios towards more liquid assets, which require less capital for
trading (“flight to quality”).
2
face following price shocks. Aragon and Strahan (2009) document that the liquidity of stocks
held by Lehman-funded hedge funds deteriorated once their funding was cut off following
Lehman’s bankruptcy. Finally, Nagel (2009) shows that the returns from providing liquidity for
Nasdaq stocks increased sharply in the recent financial crisis.
In the light of this theoretical debate and its important policy implications, this paper
intends to provide further empirical evidence on hedge funds’ trading behavior in illiquid
markets and to analyze its driving forces. Our analysis relies on an original dataset. Our main
results are based on stock and option holdings by hedge funds from the 13(f) mandatory
quarterly filings. These data are free from the self-reporting bias that affects commercial data
sets (Agarwal, Fos, and Wang 2010). We are able to identify hedge funds thanks to a proprietary
list of hedge funds provided by Thomson-Reuters and by careful hand matching. Part of our
analysis draws on TASS for hedge fund characteristics and monthly returns. Finally, since 13(f)
filings cover only long stock and option positions, we also use stock-level short interest data in
order to have an indirect view of arbitrageurs’ positions on the short side of trades.
We document that hedge funds significantly reduce their equity holdings during periods
of liquidity dry-ups. In the aggregate, hedge fund participation in the equity market declines by
about 9% to 11% (around 0.3% of the total market capitalization) when market liquidity
deteriorates by two standard deviations. During the crisis of 2008, hedge funds reduced their
holdings more dramatically: in the last two quarter of the year, they cut their equity portfolio by
about 30%, which corresponded to about 1% of the total equity market capitalization.
Consistently throughout our analysis, we find that hedge funds respond to crises defined
according to stock market liquidity factors (Pastor and Stambaugh 2003, Acharya and Pedersen
2005), but not according to measures of aggregate uncertainty (VIX index) or to pure market
crashes.
We find that the decline in equity participation among hedge funds is not spread across
all hedge funds, but rather driven by a limited set of funds which sell off large portions of their
portfolios. In quarters with low aggregate liquidity, 12% of hedge funds sell more than 40% of
their equity portfolios, compared with 4.5% unconditionally. In each of the last two quarters of
2008, 23% of hedge funds sold more than 40% of their equity portfolio.
3
Given that hedge funds overall exit the equity market during a crisis, we explore which
other classes of investors absorb hedge funds’ equity positions at times of aggregate market
illiquidity. We find that mutual funds do not change their equity holdings significantly during a
crisis. Rather, they significantly drop equity when liquidity experiences a mild deterioration.
Other institutional investors (banks, insurance, pension funds, etc.) seem in general to take the
other side of hedge fund trades in the time of crisis. However, in 2008, it appears that firms’
management and retail investors were the ultimate liquidity providers.
Next, we turn to identifying the main channels that cause hedge fund behavior. We
identify two forces that cause hedge funds to decrease their equity holdings: capital outflows and
internal reallocation of funds across asset classes.
First, we document that a tightening of the capital available for arbitrage on the part of
both investors and lenders causes hedge funds to reduce their equity positions. Shleifer and
Vishny (1997) argue that investors may pull their funds if they are concerned that arbitrage
trades may not converge, thus inducing arbitrageurs to avoid taking long-term bets. We find that
investor redemptions explain up to a quarter of the variation in hedge fund selling during
liquidity crises. Moreover, Brunnermeier and Pedersen (2009) suggest that the threat of margin
calls deters arbitrageurs from holding volatile assets. Supporting this explanation, we document
that hedge funds with higher average leverage (who are likely to be subject to greater pressures
by lenders) sell larger portions of their equity portfolio during liquidity crises. Also consistent,
we find that hedge funds are more likely to close positions in high volatility stock (both long and
short) than in low volatility stocks. The “flight-to-quality” effect is also predicted by Vayanos
(2004), who postulates that volatile assets make arbitrageurs’ capital more subject to
redemptions in volatile times. Finally, supporting the link between financial constraints and
equity sales, we report that hedge funds that sell equity are more likely to exit the hedge fund
industry within one year. In our sample, total outflows to capital providers (investors and
lenders) explain about half of the variation in hedge fund selloffs during liquidity crises.
Second, we document that hedge funds that sell equities during liquidity crises reallocate
capital across asset types, and away from the equity market. Specifically, holdings of equity put
options during a crisis are a significant determinant of hedge fund selloffs, consistent with
negative beliefs on the stock market by fund managers. Also, hedge funds that exit the stock
4
market during liquidity dry-ups tend to be familiar with other markets, as they pursue multimarket strategies like global macro, fund of funds, and managed futures. This finding is
consistent with the evidence in Cao, Chen, Liang, and Lo (2009) on liquidity management by
some the hedge funds following these styles. Finally, exiting hedge funds earn high returns on
their non-equity portfolio in the second quarter following liquidity crises, consistent with the idea
that they mobilize capital towards mispriced assets in other markets. To illustrate, two quarters
after a liquidity crisis, hedge funds that sell large portions of their portfolio during the crisis earn
up to 11% more on their non-equity portfolio relative to funds that stay in the equity market.
Our results improve our overall understanding of the role of hedge funds in financial
markets. The activity of hedge funds is limited during market dry-ups by the funding restrictions
of investors. In spite of these limitations, hedge funds appear to identify arbitrage opportunities
and provide liquidity in other corners of the economy. Our findings about the effects of capital
constraints on hedge funds’ trading patterns corroborate previous empirical results on the limits
of arbitrage (Aragon and Strahn 2009, Hombert and Thesmar 2009, Hameed, Kang and
Viswanathan 2010). Also, the evidence on selloffs at times of crisis relates to the literature that
argues that arbitrageurs act as a destabilizing force in financial markets. Consistent with this
view, Khandani and Lo (2007) provide suggestive evidence that the quant crisis in August 2007
was possibly due to the unwinding of large hedge fund positions and to the increased correlation
of hedge fund trades. Similarly, Boyson, Stahel, and Stulz (2008) show significant evidence of
contagion in the hedge fund sector, which is reinforced at times of low liquidity. We see our
result that hedge funds flee equity in bad times as symmetric to the finding that hedge funds ride
stock market bubbles (Brunnermeier and Nagel 2004). However, in contrast to these pessimistic
descriptions of hedge funds as a destabilizing force, our evidence that hedge funds are able to
reallocate capital to more attractive trades in other markets sheds a new, more positive, light on
their role during liquidity crises.
The paper proceeds as following. Section 2 describes the data sources that we use. In
Section 3, we explore the behavior of hedge funds at times of liquidity crises and study the
distribution of hedge fund trades. Section 4 takes a closer look at hedge fund trades and explores
the factors that drive them. Section 5 concludes.
5
2.
Data
2.1.
Data Sources
We use several sources of data in our study. Our primary data source is the 13(f)
mandatory institutional holdings reports that are filed with the SEC on a calendar quarter basis. 2
Thomson-Reuters institutional holdings database (formerly known as the 13(f) CDA Spectrum
34 database) provides institutional holdings as reported on Form 13(f) filed with the SEC. Form
13(f) requires all institutions with investment discretion over $100 million or more to report their
long holdings (mainly publicly traded equity, convertible bonds, and options). 3 Therefore, all
hedge funds with assets under management in such qualified securities of more than $100
million are required to report their holdings in 13(f) filings on a quarterly basis. 4 Also, hedge
funds report their holdings in public equity, convertible bonds and options at the consolidated
management company level.
We then match the list of 13(f) institutions in Thomson-Reuters with a proprietary list of
13(f) hedge fund managing firms and other institutional filers, provided by Thomson-Reuters.
The combined dataset allows us to identify entities in the 13(f) reports which are firms that
manage hedge funds. 5 Before applying the filters that are described below, the number of hedge
funds in the Thomson-Reuters list varies from several dozen in the early 1980s to over 1000 at
its peak in 2007. We cross-checked our list of hedge funds with the FactSet database and we
found it congruent with the FactSet LionShares identification of hedge fund companies.
While Thomson-Reuters collects all the institutional reports filed with the SEC, they only
retain the common equity holdings in their 13(f) institutional holdings database. To be able to
2
According to Lemke and Lins (1987), Congress justified the adoption of Section 13(f) of the Securities Exchange
Act in 1975 because, among other reasons, it facilitates consideration of the influence and impact of institutional
managers on market liquidity: “Among the uses for this information that were suggested for the SEC were to
analyze the effects of institutional holdings and trading in equity securities upon the securities markets, the potential
consequences of these activities on a national market system, block trading and market liquidity…”
3
With specific regard to equity, this provision concerns all long positions greater than 10,000 shares or $200,000
over which the manager exercises sole or shared investment discretion. The official list of Section 13(f) securities
can be found on the following SEC webpage: http://www.sec.gov/divisions/investment/13(f)lists.htm
4
More information about the requirements of Form 13(f) pursuant to Section 13(f) of the Securities Exchange Act
of 1934, can be found at: http://www.sec.gov/divisions/investment/13(f)faq.htm.
5
As a shortcut, from now on we will refer to the observational unit in our data set as a ‘hedge fund’. It should be
clear, however, that 13(f) provides asset holdings at the management firm level. Each firm reports consolidated
holdings for all the funds that it has under management.
6
capture the stock options held by hedge funds with 13(f) institutional holdings reports, we
downloaded and parsed all 13(f) electronic forms available on the SEC website, a method similar
to that of Aragon and Martin (2009). The SEC requires institutions to report separately all call
and put options for a large set of 13(f) securities.6,7 We looked only at the original 13(f) reports
that are filed within forty-five days of the end of the calendar quarter, and mapped the list of our
hedge funds to the CIKs they used in reporting SEC filings. 8
We also follow Griffin and Xu (2009), who use similar datasets, and map hedge funds’
13(f) data to fund characteristics and monthly returns that are collected by Thomson-Reuters’
Lipper-TASS database (drawn in August 2009). The Lipper-TASS database provides hedge fund
characteristics (such as investment style and average leverage) and monthly return information at
the strategy or portfolio level. We aggregate the TASS data at the management company level,
and match it to the 13(f) dataset using the consolidated management company name. 9
The main advantage of our dataset is that it includes all equity hedge funds that are
required to report their holdings in 13(f) filings. Thus our dataset is broader and more
comprehensive than that of Griffin and Xu (2009) and of prior studies. Also, the 13(f) filings are
not plagued by selection and survivorship bias due to the reliance on TASS and other selfreported databases for hedge fund identification (Agarwal, Fos, and Jiang 2010). Finally, the
Thomson-Reuters hedge fund list identifies hedge funds at the disaggregated advisor level, not at
the 13(f) report consolidated level. For example, for Blackstone Group holdings in 13(f) data,
6
The official list of Section 13(f) securities refer to options by their underlying securities:
http://www.sec.gov/divisions/investment/13(f)lists.htm, and requires CALL or PUT designations for options in the
issuer description field. We used such “CALL” and “PUT” strings to identify option positions in 13(f) filings, where
they appear under Item 2 or Item 6 or as a suffix to the company name in the body of the holdings table of the 13(f)
report. Note that some filings used different identifiers for options, such as Goldman Sachs Group, which uses
“CAL” for call options. We were able to capture and identify many such special cases.
7
As an accuracy check, we compared the common equity portion of our parsed 13(f) dataset with Thomson-Reuters
13(f) institutional ownership data, and we found a 99% correlation.
8
We noticed that several hedge funds filed subsequent amendments with regard to their confidentially treated
holdings which were excluded from the original 13(f) filings, but reported in the form of amendments only after the
expiration or rejection of the confidential treatment requests. We also noticed that the Thomson-Reuters data usually
excludes such individual holdings from their data as it is published as originally reported and apparently overlooks
subsequent amendments. We rely on Thomson-Reuters 13(f) holdings data as it has better historical coverage from
1980 than the electronic 13(f) filings that have been posted on the SEC website since 1999. See Agarwal, Jiang,
Tang, and Yang (2009) for more information about and statistics on the confidentially treated holdings (Table 2,
Panel B).
9
We used strategy portfolio assets as weights in aggregating fund characteristics and total reported returns.
7
Thomson-Reuters provided us with a classification of each of the advisors within Blackstone that
reported its holdings in the same filing. 10
Because many financial advisors manage hedge-fund-like operations, we need to apply a
number of filters to the data. In order to limit our analysis to hedge funds, we drop institutions
that have many advisors with non-hedge-fund business (e.g. Goldman Sachs Group, JP Morgan
Chase & Co, American International Group Inc.), even though they have hedge funds that are
managed in-house and included with their holdings in the parent management company 13(f)
report. As a further filter, we double-checked the hedge fund classification by Thomson-Reuters
against a complete list of ADV filings by investment advisors since 2006. 11 We matched those
filings by advisor name to our 13(f) data. Then, following Brunnermeier and Nagel (2004), we
kept only the institutions with more than half of their clients classified as “High Net Worth
Individuals” or “Other Pooled Investment Vehicles (e.g. Hedge Funds)” in Item 5.D
(Information About Your Advisory Business) of Form ADV. Therefore, we believe that our final
list of hedge funds contains only institutions with the majority of their assets and reported
holdings in the hedge fund business.
We compare the hedge fund population obtained from 13(f) to the matched population of
hedge funds in TASS. Table 1, Panel D presents the annual number of hedge funds in our sample
that are required to report their holdings through 13(f) filings, and the number of matched hedge
funds who self-reported their total returns, and individual fund characteristics to Lipper-TASS.
The panel suggests that the matched sample contains only about 20% of the universe of hedge
funds filing 13(f) forms. The panel also shows the explosion in the number of hedge funds over
the last decade, and is consistent with the recent patterns of hedge fund liquidations at the end of
2008 and in the first three quarters of 2009. According to Hedge Fund Research Inc., the total
assets managed by hedge funds decreased by around 19% by 2009, due to the market crisis and
10
There are three advisor entities within Blackstone Group L.P. that report their holdings in the same consolidated
Blackstone Group report. Among the three advisors included, GSO Capital Partners and Blackstone Kailix Advisors
are classified by Thomson-Reuters as Hedge Funds (which an ADV form confirms), while Blackstone Capital
Partners V LP is classified as an Investment Advisor. See the “List of Other Included Managers” section in
September 30 2009 Blackstone 13(f) reports filed on November 16 2009:
http://www.sec.gov/Archives/edgar/data/1393818/000119312509235951/0001193125-09-235951.txt
11
All current advisor ADV filings are available on the SEC’s investment advisor public disclosure website:
http://www.adviserinfo.sec.gov/IAPD/Content/Search/iapd_OrgSearch.aspx
8
the record-setting hedge fund closures in 2008 and 2009. 12 This pattern is strongly reflected in
Figure 1, which plots hedge fund equity holdings over time as a fraction of total market
capitalization.
While hedge funds are known for holding both long and short positions, the information
reported in the 13(f) filings includes only long transactions. To complement the long-side data,
we use short interest data provided by the exchanges. These data are reported monthly since
1988 at the stock level (therefore we cannot identify the investors who hold the short positions).
In our analysis we assume that most short sellers are arbitrageurs, and of those, many, if not
most, are hedge funds. Our assumption is supported by the parallel increase in the aggregate
short selling and hedge fund activities over time (compare Figures 1 and 2; the correlation of the
quarterly changes is 0.38). Furthermore, aggregate short selling activity is quite small in
magnitude, even in recent years, suggesting that only a small group of specialized arbitrageurs
engage in it.
We use several widely used datasets for stock-level and market-level information.
Specifically, we use CRSP and Compustat for stock characteristics. We use two popular
aggregate liquidity measures: Pastor and Stambaugh’s (2003) innovations in liquidity and
Acharya and Pedersen’s innovations in illiquidity (2005; see also Acharya, Amihud, and Bharath
2009). Pastor and Stambaugh measure market-wide liquidity from the aggregation of firm-level
OLS slopes of daily returns on signed daily trading volume within a month. Acharya and
Pedersen capture aggregate illiquidity by averaging the stock-level illiquidity as measured by
Amihud’s (2002) ratio. Both liquidity factors are expressed at the quarterly frequency by
summing monthly innovations. In addition, to contrast the effect of liquidity with that of
aggregate uncertainty and returns in the equity market, we also consider the first difference in the
end-of-quarter VIX index (VIX) and quarterly excess market returns (Rm–Rf). We change the
sign to the Acharya and Pedersen and VIX innovations so that an increase in these factors
describes improvements in liquidity and reduction in uncertainty, respectively.
12
See BusinessWeek’s article “Hedge Your Bets like the Big Boys” by Tara Kalwarski, in the December 28, 2009
issue.
9
2.2.
Summary Statistics
Because our analysis focuses on both long and short holdings and since the hedge fund
universe in the 1980s is tiny relative to the explosion that occurred in the following two decades,
we limit our hedge fund holdings and short interest data period to the third quarter of 1989 until
the first quarter of 2009 (this period mostly overlaps with the period for which we have short
interest data). In addition, we winsorize fund flows and changes in hedge fund equity holdings at
the 5th and 95th percentiles within each quarter, as the distributions of these variables have fat
tails. Finally, we verify that our results are not driven by extreme observations.
Table 1 presents summary statistics of the datasets used in the study. In Panel A we
present the summary statistics of the aggregate stock market participation by hedge funds. The
table shows that the selected hedge funds hold, over the two decades, 1.84% of the entire stock
market capitalization on average, peaking at 3.74% (in the second quarter of 2007). The short
interest ratio averages 1.82%, peaking at 3.97% (in the second quarter of 2008).
The dependent variables in the regressions in aggregate data are the change in hedge fund
holdings as a percentage of total equity holdings (1.6% on average) and of total market
capitalization (0.027% on average). To construct these variables we aggregate the quarterly
trades made by each fund evaluated at previous period prices and divide them respectively by
either the total equity holdings by hedge funds in the previous quarter or by the total market
capitalization in the previous quarter. For this construction, we require hedge funds to appear in
two consecutive quarters. When a hedge fund does not report (since it is below the $100 million
assets-under-discretion cutoff), we eliminate the observation (as opposed to reporting a large
drop in holdings). The choice of previous-quarter prices allows us to focus on changes in equity
holdings that are due to trades and not to price changes.
Panel B presents summary statistics for the stock-quarter-level sample. The dependent
variable in the stock-level regressions is the change in the number of shares of a firm held by
hedge funds aggregated across all hedge funds in our sample divided by total number of shares
outstanding for that firm. Across stocks, this figure averages 0.07%. Focusing on the level of
stock ownership, hedge funds hold about 3.1% of firm equity on average. From the comparison
with the aggregate holdings in Panel A, which are weighted by market capitalization, it appears
that hedge funds’ equity holdings are tilted towards smaller stocks, consistent with the evidence
10
in Griffin and Xu (2009). Volatility is computed as the standard deviation of monthly returns
over a two-year window.
Panel C of Table 1 presents summary statistics for the hedge-fund-quarter-level data. The
dependent variable in some of the hedge-fund-level analysis below is the fraction of the fund
equity portfolio that is traded over the quarter. Again, the choice of previous-quarter prices
avoids introducing a bias due to the change in prices over the quarter. To construct this variable
we aggregate the quarterly changes in holdings for all the stocks in the fund portfolio and
evaluate them at the previous quarter prices. Then, the total dollar value of the trades is divided
by the lagged (previous quarter) value of the equity portfolio. The average percentage change in
hedge funds’ equity portfolios is 4.59%. The hedge-fund-quarter data is matched with TASS, as
explained above. We use TASS data to construct total returns, by aggregating returns of funds
within each management company (weighted by asset size). Then, we compute quarter t fund
flows as the quarterly difference in assets under management minus the dollar return on quarter t
– 1 assets. Fund flows are then scaled by the lagged hedge fund equity holdings. We define
categorical variables as equal to one if the fund stops reporting to TASS and 13(f) within one
year, which average 12.8% and 14.1%, respectively. We also define a dummy for whether the
fund holds put options on individual stocks according to the 13(f) filings (about 32% of the
hedge-fund-quarter observations after 1999, on average). Then, we construct variables that
capture the fraction of the firm’s assets that are invested in the different strategies identified by
TASS.
Panel D provides a summary of coverage for the 13(f) and TASS data as well as
summary statistics about hedge funds’ equity portfolios. It shows that, over time, more small
hedge funds entered the industry, holding fewer stocks on average at a higher (annualized)
turnover rate. As in the CRSP mutual fund data base, Wermers (2000), and Brunnermeier and
Nagel (2004), portfolio turnover is defined as the minimum of the absolute values of buys and
sells during a quarter t divided by total holdings at the end of quarter t − 1, where buys and sells
are measured with end-of-quarter t − 1 prices. This definition of turnover captures trading
unrelated to inflows or outflows. Because it is computed from quarterly snapshots, it is
understated, nevertheless it provides an important assessment of the relevance of quarterly
holdings data. The average quarterly turnover in the sample is 27.2% (109% annualized). The
magnitude of turnover in our data is comparable to that found by Brunnermeier and Nagel
11
(2004). While somewhat higher than the 72.8% (annualized) turnover for mutual funds in 1994
found by Wermers (2000), our figure indicates that a substantial part of portfolio holdings
survives on the quarterly horizon. As argued by Brunnermeier and Nagel, this finding legitimates
the use of quarterly snapshots to capture the low frequency component of hedge fund trading.
2.3.
Identifying Crises
For the analysis below, we need to identify periods of stress in the equity market. To this
end, we select quarters of extreme realizations (two standard deviations from the mean) of four
different market condition variables. The Pastor and Stambaugh (2003, PS hereafter) and
Acharya and Pedersen (2005, AP hereafter) variables are based on direct measures of market
liquidity. The PS and AP variables are available from the last quarter of 1989 to the last quarter
of 2008. However, because the 2008 events cause major skewness in the AP variable which
would obfuscate other liquidity events in the sample, we use only the realizations of this variable
in the 1989:Q1-2007:Q4 period to identify financial crises. The evidence for the 2008 crisis is
then presented separately.
We also use other measures for alternative dimensions of market stress. Nagel (2009)
suggests that the returns to providing liquidity increase in periods of aggregate uncertainty, as
captured by the VIX index. In this spirit, we use quarterly changes in the (negative of the) VIX
index as a further variable, which is available over the entire sample. Finally, the return on the
stock market in excess of the risk free rate captures stock market crises Rm–Rf. Because of our
normalization, for all the variables, low variable values reflect poor market conditions. The
correlations (Panel E) among the liquidity variables and between the liquidity variables and the
VIX and the market variable are surprising low. The correlation between the VIX and the market
variable is high (0.71).
Panel F presents the list of identified crises, per market condition variable. We notice that
based on the definition of a crisis quarter as being two standard deviations away from the mean
over the period, crisis quarters are not entirely overlapping across market condition variables.
12
3.
Trades of Hedge Funds during Liquidity Crises
3.1.
Equity Market Participation by Hedge Funds
Our first goal is to characterize hedge fund behavior during crises. In Figure 1, we look at
aggregate hedge fund equity holdings as a fraction of total stock market capitalization. The
vertical lines in the figure denote our selection of events of potential market stress since 1990:
e.g., Summer 1998, September 2001, the period after Summer 2007 (the full list is provided in
the appendix). When examining the time series, it appears that in times of crisis hedge funds
withdraw from the market, especially in 2008.
Next, we use regression analysis to examine the relation between changes in hedge fund
holdings of stocks and the four aggregate market condition variables described in Section 2. The
results of the time-series analysis are presented in Table 2, Panel A. The dependent variable is
the quarter-on-quarter aggregate change in hedge fund equity holdings, and the explanatory
variables include indicators for the changes in the examined market condition variables, and
controls for market returns (also as indicators of standard deviations from the population mean).
Given the design of the regressions, the coefficients measure the average quarterly change in
aggregate equity holdings by hedge funds. The coefficients on the liquidity variables indicators
(Columns (1) and (2)) show that, controlling for market returns, hedge funds decrease their
equity participation almost monotonically, yet non-linearly, as market liquidity decreases. In
particular, hedge funds reduce their participation considerably when liquidity variables are low. 13
The regressions show hedge funds reduce their equity participation by an average of 10% in
quarters in which either the Pastor and Stambaugh factor or the Acharya and Pedersen factor was
two standard deviations below the mean. Interestingly, Columns (3) and (4), where crises are
defined by the VIX and market excess returns, are statistically insignificant and economically
weak. Comparing the coefficients in Columns (1) and (2) to those in Columns (3) and (4)
suggests that hedge funds behavior is associated with illiquidity rather than simply with
uncertainty or poor market returns.
We have a particular interest in quantifying the withdrawal of hedge funds from the
equity market during the crisis of 2008. In Column (5), we add an indicator for the third and
13
The results are almost identical when we restrict the data to the sample of hedge funds that self-report to TASS.
13
fourth quarters of 2008, which are likely to capture the time of most severe stress in 2008 due to
the Lehman collapse. The results show that during the last two quarters of 2008, hedge funds
exited the equity market at a dramatic rate of 16.4% per quarter on average.
In Panel B of Table 2, we explore whether hedge funds’ exit from the equity market is
contemporaneous or whether it happens at a lag or lead. We regress changes in hedge fund
ownership on current, lagged, and lead crisis indicators, as well as on corresponding indicators
for market returns. The results show that the exit of hedge funds is primarily contemporaneous
with respect to liquidity crises.
To assess the economic impact of the exit of hedge funds from the equity market, we
repeat the analyses of Panels A and B where we measure the change in hedge fund holdings as a
percentage of the total market capitalization (in lagged quarter valuations). The results are
presented in Table 2, Panels C and D. The panels show that in quarters of low aggregate
liquidity, hedge funds sell stocks worth 0.25% of the total market capitalization (Column (1)).
During the financial crisis of 2008, hedge funds sold stocks worth 0.6% of the total market
capitalization at each of the third and fourth quarters of 2008. Because pure-play hedge funds,
which are examined here, hold only a small fraction of the market capitalization, their selling
pressure appears to be relatively small in magnitude relative to the total market capitalization,
and thus casting doubt on whether they are able to contribute much to liquidity dry-up in the
market. 14 We investigate the stock-level cross-sectional dimension of selloffs in Section 4.
3.2.
The Distribution of Selling for Stocks and of Hedge Funds
Given the large equity selling on the part of hedge funds during liquidity crises, it is
important to understand whether the effect is driven by large sales on the part of a limited
number of hedge funds or whether it is widespread in small quantities across the hedge fund
universe. To explore this issue, we compute for each hedge fund the fraction of equity bought or
sold at previous-quarter prices. We begin by comparing the distributions of hedge fund trades in
Figures 3a and 3b. Figure 3a presents the unconditional distribution of hedge fund trades and the
distribution of trades conditional on a state of crisis as measured by the PS variable. Figure 3b
shows that same unconditional distribution, and the distribution of hedge fund trades in the
14
One issue with our sample is that we limit the sample to pure-play hedge funds, and therefore, omit a fraction of
large players (e.g., Goldman Sachs) who have a hedge fund business in addition to other types of businesses.
14
second half of 2008. Both charts show that there is a shift to the left in the distribution of trades
during crisis and a non-negligible mass of hedge funds sell significant portions of their
portfolios. This pattern is particularly noticeable for the crisis of 2008. In Figure 3a, there is no
evidence that funds exploit crises to dramatically increase their holdings. However, Figure 3b
shows that in the crisis of 2008, about 5% of hedge funds increase their portfolio holdings by
more than 90%.
We assess these issues also in Table 3, Panels A and B. In Panel A, we present the
distribution of funds with respect to the degree of their buying or selling, and the liquidity state
measured by the PS variable. The table shows that while the distribution of hedge funds that buy
any amounts, or sell moderate amounts is more or less stable across liquidity states, the
distribution becomes skewed for large sellers in extreme events. During normal times about 5%
of the funds sell more than 40% of their equity portfolio in a given quarter. Conversely, during
times of crisis, 13.6% of hedge funds sell more than 40% of their portfolios. When we isolate the
recent crisis of 2008, in the last column, the numbers are staggering. We find that 18.8% of
hedge funds sold between 20% and 40% of their portfolios in each quarter, and an average of
23.4% of funds sold more than 40% of their equity holdings in each quarter of the crisis.
We are also interested in quantifying the economic magnitude of hedge fund selloffs; if
the selling population is composed of small hedge funds, then the economic effect may be
overstated. In Panel B of Table 3, we repeat the analysis of Panel A, while value-weighting the
funds in each buckets using total lagged equity portfolio value. The panel shows that the
distribution of large sellers at illiquid times does not change much, indicating that selloffs are
performed by hedge funds representative of the hedge fund size distribution. However, the
distribution of large buyers shrinks, suggesting that large buyers are small hedge.
3.3.
Which Investor Types Provide Liquidity during Low Liquidity Periods?
Since hedge funds reduce their positions in equity during times of low market liquidity, it
is interesting to find out who buys their shares. In Table 4, we repeat the regression from Table
2, Panel A, for other types of investors: mutual funds, other institutions, and management and
retail investors. We focus on the Pastor and Stambaugh index as the liquidity index. For mutual
funds (Columns (1) and (2)), there is a decrease in holdings in moderate declines in liquidity,
however, there is no effect in liquidity crisis, including the crisis of 2008. For other institutions
15
(Columns (3) and (4)) and for retail investors (Columns (5) and (6)), we do not spot large
changes in equity market participation during liquidity crises. The only noticeable pattern we
identify is that management and retail investors increased their participation in the equity market
in the recent crisis of 2008 by about 1.35%; this corresponds to the amount that hedge funds sold
during the crisis. Nevertheless, this result is not statistically significant (t = 0.94).
4.
Why Do Hedge Funds Sell during Low Market Liquidity Periods?
So far we documented large selling by hedge funds during periods of low market
liquidity. There are two possible (non-mutually exclusive) explanations for this phenomenon.
First, investors and lenders could force hedge funds to liquidate equity positions by tightening
their funding. We test this explanation by examining the effect of investor fund flows on hedge
fund trades during liquidity crises. Also, we explore whether highly leveraged hedge funds are
more likely to sell equities and whether hedge funds in general are more likely to sell high
volatility stocks (which have stricter margin requirements), and whether hedge funds are more
likely to exit the industry following equity sales. Second, hedge funds could sell equity during
liquidity crises as part of a reallocation effort of capital across asset markets as they identify
better opportunities in non-equity markets. To assess this possibility, we infer hedge fund
managers’ beliefs by examining whether there is a correlation between put option holdings and
selloffs during periods of liquidity dry-ups. We further study the types of hedge funds that exit
the equity market during a liquidity crisis and analyze the future returns on their non-equity
portfolios.
4.1.
Financial Constraints during Liquidity Crises
4.1.1. Redemptions by Investors
We conjecture that investor redemptions drive some of the hedge fund selling.
Redemptions may be at their peak during periods of market illiquidity, and may force hedge
funds to sell relatively liquid assets, such as stocks. We explore this idea by analyzing the crosssection of hedge funds’ quarterly trades. The prediction is that hedge funds that experience larger
redemptions would sell more equity. We impute net fund flow data 15 (scaled by lagged equity
15
Net fund flows are computed as TASS variables EstimatedAssets(q) - EstimatedAssets(q-1)(1 + RateOfReturn),
scaled by lagged (q-1) equity portfolio size which is derived from 13(f). EstimatedAssets are aggregated from the
fund/style level to the management company level. RateOfReturns are weighted by the estimated assets.
16
portfolio size) from TASS, and thus need to restrict the 13(f) dataset to the sample matched with
TASS.
We present the results in Table 5, which explores how well the different variables explain
the change in quarterly equity holdings by hedge funds. In all regressions, standard errors are
clustered at the quarter level. Panel A of Table 5 focuses on the Pastor and Stambaugh (2003)
liquidity index (Columns (1) to (6)), and on the 2008 crisis (Column (7)). In Column (1), we
present the baseline regression. We use the universe of firms that appear both on TASS and
13(f), 16 and regress the changes in their equity holdings on the crisis indicator and on hedge fund
controls (size and past returns). We note that based on this sample, hedge funds reduce their
equity portfolio by 7.7% per quarter on average during liquidity crises. In Column (2), we
introduce fund flow variables as well as interactions of the crisis dummy with current and lead
fund flows. The regression shows that flows are positive and statistically significant, while the
coefficients on the interactions with the crisis dummy are statistically insignificant. Yet,
introducing these variables reduced the coefficient on the crisis indicator from 7.7% to 5.7%, a
decline of 26%. Thus hedge funds liquidate their equity positions due to current and future flows
(often known in advance due to the redemption notice period). We focus on the effect of fund
flows in the crisis of 2008 (Column (7)). The results show that contemporaneous flows are
negatively correlated with hedge fund trading, while future fund flows are positively correlated
with trading, suggesting that hedge fund trading during the crisis was driven to some extent by
future investor redemptions.
Panels B, C, and D repeat the regressions in Panel A for the Acharya-Pedersen liquidity
index, the VIX, and market excess returns, respectively. The results in Panel 2 are qualitatively
similar and stronger in magnitude relative to the results in Panel A. In Column (2), the
coefficient on fund flows is positive and reduces the magnitude of the coefficient of the crisis
indicator from 8.7% to 4.7%, a decline of 45%. The regressions in Panel B that use the VIX as a
crisis indicator show no sensitivity to investor redemptions around crisis events. The regressions
in Panel D, where crisis is defined based on market returns, exhibit a significant relation to
investor redemptions.
16
Since the regressions in Table 5 are used as a benchmark across different explanations, we restrict the sample used
to a 1999 start, when put option data is available. The results do not change materially if we begin the sample
earlier.
17
4.1.2. Credit Tightening by Lenders
Next, we look for evidence that hedge funds sell their equity positions because they are
forced to do so by lenders. We carry out two tests. In the first test we examine whether highly
leveraged funds sell larger portions of their portfolios during liquidity crises. In Column (3) of
Table 5, Panel A, we regress the fraction of the equity portfolio traded by hedge funds over the
quarter on crisis indicator (based on the Pastor and Stambaugh index) interacted with hedge
funds’ average leverage. 17 The resulting coefficient on the interaction is negative and statistically
significant, suggesting that highly leveraged hedge funds are more likely to reduce their equity
holdings during crisis. Average leverage is measured as debt over investor equity. The
magnitude of the coefficient is 4.6% and should be multiplied with the leverage in order to get
the economic effect. A 2:1 leveraged hedge fund sells 9.2% more of its equity portfolio than an
unleveraged fund during liquidity crisis. The effect is weaker in magnitude and insignificant in
the crisis of 2008 (Column (7)). When using Acharya-Pedersen as a liquidity indicator, the
results are stronger (Panel B). When crisis is measured as changes in the VIX or market returns,
there is no material effect of hedge fund leverage (Column (4) of Panels C and D). 18
We assess also the total variation in hedge fund equity trading that can be explained by
financial constraints. In Table 5, Panel A, Column (4), we regress the changes in equity holdings
on the crisis indicator interacted with both net fund flows variables, and average leverage.
Comparing the coefficient on the crisis dummy to the coefficient in Column (1), we conclude
that financing constraints account for about half of the variation in equity trading during a
liquidity crisis. In Panel B, where the Acharya and Pedersen index is used, the ratio of the
coefficients leads to the conclusion that financial constraints account for 45% of the variation.
In the second set of results, we explore the cross-section of stocks and examine whether
hedge funds are more likely to close positions in high volatility stocks during liquidity crises. We
focus on return volatility because it is positively correlated with stock margin requirements.
Drawing on Vayanos (2004), Brunnermeier and Pedersen (2009), and Hameed, Kang,
17
The Average Leverage variable is a cross-sectional variable provided by TASS. This variable describes the
general level of leverage of hedge funds, reported by funds’ managers. Like other variables from TASS, we
aggregate this variable from the fund level to the management company level, weighting it by EstimatedAssets.
18
Since about 53% of the TASS hedge funds report a zero average leverage, we repeat the test for the population
that has non-zero average leverage. The coefficient on the interaction between the crisis dummy and average
leverage changes to -5.953 (t = -3.3) and -5.790 (t = -4.5) in Panels A and B, respectively.
18
Viswanathan (2010), we conjecture that at times of low liquidity, hedge funds would have
trouble funding positions in volatile stocks because margin requirements are generally higher for
these stocks. Hence, lenders may force hedge funds to reduce exposure, focusing on high
volatility stocks. Hedge funds may be motivated to sell high volatility stocks for internal reasons
as well. Specifically, many hedge funds (like their capital providers) use value-at-risk (VaR)
models as a tool to monitor risk exposure. Reducing risk exposure could be vital in preempting
redemption by investors. In this test, we cannot distinguish between forced and voluntary selling
of high volatility stocks. 19
In Table 6, we study whether stocks with higher volatility are more heavily sold by hedge
funds during crises. We use stock-quarter level data. The dependent variable is the change in
hedge fund share holdings as percentage points out of the lagged total shares outstanding. The
explanatory variable of interest is the stock level volatility indicator (indicating above–themedian volatility at the end of the preceding quarter) interacted with crises indicators. If hedge
funds reduce their holdings more in high volatility stocks during extreme episodes, then the
expected interaction coefficient is negative. In all regressions, standard errors are clustered at the
quarter level. The results in Table 6, Panel A, Columns (1) and (2) confirm this prediction: hedge
funds are more likely to reduce their positions in high volatility stocks during periods of low
aggregate market liquidity. In Column (1), high volatility stocks have almost double the
likelihood of being sold by hedge funds during liquidity crisis than do low volatility stocks.
Column (2) produces a lower estimate: the likelihood of selling high volatility stocks during a
liquidity crisis is about 50% higher than that of selling low volatility stocks. Column (3) shows
that market uncertainty, as measured by the VIX, does not affect the likelihood of selling stocks
with respect to volatility. Interestingly, Column (4) presents evidence that the likelihood of
selling high volatility stocks in periods of low market returns is only slightly (and statistically
insignificantly) higher than that of low volatility stocks. Finally, Column (5) shows that during
the crisis of 2008, hedge funds reduced their positions more in high volatility stocks.
19
We note, however, that the distinction between internal and external forces is blurred. Consider the external
pressure that lenders and investors may put on hedge funds to liquidate high volatility stock positions. Since hedge
funds can anticipate the demands of risk reduction by external capital providers, they can respond ahead of time by
liquidating high volatility positions. Hence, hedge funds’ reduction of risk according to VaR models can be viewed
as an attempt to preempt forced liquidation. In addition, external pressure to liquidate positions can be observed in
the data as internally motivated sales.
19
We repeat the analyses for changes in stock-level short interest. The motivation for this
analysis is that portfolio volatility can also be reduced by closing short interest positions. We
anticipate similar results in the short interest sample, i.e., that hedge funds more aggressively
reduce positions in high volatility stocks. The evidence in Panel B validates this prediction.
Columns (1) and (2) show that during liquidity crises, stock-level short interest of high volatility
stocks is reduced by a greater amount (measured as change in the short interest, as a fraction of
market capitalization) than the short interest of low volatility stocks. Column (3) shows that
during periods of high market uncertainty as measured by the VIX, short sellers reduce their
overall positions, albeit less than for high volatility stocks. Column (4) presents evidence that
during periods of low market returns, arbitrageurs close short positions (realizing profits),
particularly for high volatility stocks.
4.1.3. Hedge Funds Going Out-of-Business
To further establish the consequences of financial constraints on hedge funds, we
examine the relation between equity sales and future hedge fund bankruptcy. We examine this
relation by examining whether hedge funds that sell large quantities during crises are more likely
to drop from our sample within one year. We use both the 13(f) and TASS datasets. We
construct an indicator of whether hedge funds stop reporting to 13(f) or to TASS within one year.
Each measure has its drawbacks. Although 13(f) filing does not have reporting biases (apart from
minimum portfolio size), dropping the 13(f) means that a hedge fund stopped investing in the
equity market, not necessarily that it went bankrupt. Using the TASS dataset also does not
precisely measure bankruptcies since firms can stop reporting for other reasons (e.g., large hedge
funds may stop reporting as they do not need to attract investors any more).
In Table 6, Panel C, we report the results of the probit regressions (marginal effects for
the average fund are reported) of this variable for both 13(f) (Columns (1) to (4)) and TASS
(Columns (5) and (8)). The independent variables of interest are the change in portfolio holdings
variable, and its interaction with the crisis indicator. Standard errors are clustered at the quarter
level. Both sets of regressions show that the likelihood of dropping out of the 13(f) sample and
TASS is associated with an early sell of equities; however, this association is not stronger for
firms that sell during crises. The magnitude of the effect is large: each percentage increase in the
equity portfolio sold is associated with an increase of 0.2% in the likelihood of dropping of either
20
sample (base probabilities are 14.1% and 12.8% of dropping from the 13(f) and TASS samples,
respectively).
In sum, the results in this section suggest that restrictions on funding by capital providers,
namely investors and lenders, force hedge funds to liquidate equity positions. In particular, we
presented evidence that hedge funds’ selling are induced by investor flows induce equity sales,
that highly leveraged hedge funds are more likely to liquidate equity positions, and that equity
sellers are more likely to go out of business within a year. Further our results show that hedge
funds are more likely to sell high volatility stocks, either due to margin calls or in an attempt to
limit risk exposure and avoid margin calls.
4.2.
Reallocation of Capital across Asset Markets
The results in Section 4.1 suggest that equity and debt redemptions explain about half of
the decrease in equity holdings during a crisis. Thus, we conjecture that selling hedge funds
retain the rest of the funds, potentially to invest them in other, more profitable, asset markets.
After all, negative shocks are usually correlated across asset markets and investment
opportunities may arise in markets that are more illiquid than the equity market. We explore this
hypothesis in different tests. 20 We first assess the beliefs of hedge fund managers during market
illiquidity periods by exploring whether they have defensive portfolio holdings. Then, we study
which types (styles) of funds exit the stock market during liquidity crises. We conjecture that
funds with multi-asset expertise are more likely to exit than equity-specialized hedge funds.
Finally, we explore how hedge funds’ overall returns are related to opportunities open to them in
other markets.
4.2.1. Are Hedge Funds Bearish About the Stock Market during Low Liquidity Periods?
We explore the possibility that hedge fund managers are concerned about the future
returns of equities and therefore exit the market. Specifically, managers of hedge funds who are
bearish about the equity market are expected to sell equity during a liquidity crisis and are more
likely to hold put options. In Table 5, Panel A, Column (5), we interact the Pastor-Stambaugh
20
Ideally, we would like to test directly whether hedge funds increase their non-equity portfolio. However,
combining the TASS and 13(f) datasets does not allow for reconstructing the balance sheets of hedge funds. In our
data we have the size of the equity portfolio (from 13(f)), the size of assets under management (from TASS) and the
average leverage (from TASS). The problem is that the average leverage variable in TASS provides a snapshot at
the time of reporting, that is, it does not have a time-series dimension.
21
crisis dummy with an indicator of whether hedge funds hold put options. The coefficient is
−7.2% suggesting that hedge funds that hold put options sell larger portions of their equity
portfolios by this amount. In the corresponding regression in Table 5, Panel B, Column (5),
which is based on the Acharya-Pedersen index, the magnitude of the coefficient is similar
(−5.3%). Interestingly, there is no relation between hedge fund selling and put positions
following crises based on uncertainty (VIX; see Table 5, Panel C, Column (5)). Following stock
market crashes, however, hedge funds’ equity trading is positively correlated with put options
holding.
In sum, the negative relation in Panels A and B between hedge fund trades during
liquidity crises and put options holding supports the idea that equity selloffs are also driven by
beliefs about further equity price declines.
4.2.2. Which Hedge Funds Exit the Equity Market?
Second, we explore the types of hedge funds that exit the equity market. In particular, we
study how the magnitude of equity sales in a crisis varies with the investment style of hedge
funds. We assume that hedge funds with multi-asset styles have a high likelihood of investing
their selling proceeds in other asset markets. In Table 7, we regress the fraction of hedge funds’
equity portfolios sold on crises indicators interacted with investment strategies, as reported in
TASS. 21 The standard errors are clustered at the fund level. The results in the table show that
hedge funds with market neutral strategies increase their participation in the market during a
liquidity crisis and in crises defined by the VIX index. On the other hand, hedge funds that
follow an event-driven strategy, fixed income arbitrage, global macro, and managed futures,
reduce their holdings significantly during liquidity dry-ups. Interestingly, in the recent crisis of
2008 (Column (9)), we do not observe significant results (potentially due to the small sample
size). In terms of magnitude, it appears that during the recent crisis, fixed income arbitrage hedge
funds were large equity sellers.
Overall, we view these results as additional support to the hypothesis that multi-asset
hedge funds that are familiar with alternative markets (e.g., global macro and managed futures
21
Since TASS reports investment strategies per fund, and our analysis is performed at the hedge fund parent
company level, we aggregate fund styles to the company level, and weight them by the lagged total assets managed
by each fund. Thus, each hedge fund firm can have multiple strategies, each accounts for 0% to 100%, and all add
up to 100%.
22
hedge funds) reallocate equity sales proceeds to those other markets as investment opportunities
arise.
4.2.3. Hedge Fund Returns on Non-Equity Portfolios
Finally, we examine which hedge funds earn high returns following liquidity crises. In
our data we observe the returns on the long equity portfolio of hedge funds in addition to the
returns to the equity investors of the hedge funds (“total returns”). In order to know whether
hedge funds invested successfully reallocated funds across markets, we examine whether future
total returns are higher for hedge funds that sold large fractions of their equity portfolio during
liquidity crises, after controlling for the returns of their long equity portfolios.
In Table 8, we regress future total returns (one- and two-quarters ahead in Panels A and
B, respectively) on changes in equity holdings interacted with crises indicators. The standard
errors are clustered at the fund level. In Panel A, there is no significant effect in the short
horizon, except for in Column (4), in which market crashes are considered as crises. Two
quarters following a crisis, however, exiting hedge funds perform much better than their peers
(Columns (6) and (7)). The economic magnitude is large. Hedge funds that sell 20% of their
equity portfolios have returns that are higher by 1.6% to 2.5% 22 in the second quarter following
the crisis.
In Panel B, we examine the profitability of the different strategies given past crises. We
regress future (one- or two-quarters ahead) returns on crises indicators interacted with investment
strategies. The results in Columns (1) and (2) show that a fixed income arbitrage strategy suffers
serious losses in the quarter following a liquidity crisis, perhaps because this strategy relies on
exploiting a risk premium related to crises. In contrast, global macro, fund of funds, and
managed futures strategies perform well following liquidity crises. In the longer horizon of two
quarters ahead (Columns (6) and (7)), we find that emerging markets, multi-strategy and longshort strategies underperform. The strategies that perform well two quarters following liquidity
crises are short-bias, market neutral, and managed futures strategies.
Overall, these results suggest that hedge funds that exit the equity market invest the
proceeds in other markets. Given that liquidity crises spread swiftly through the economy,
22
In Column (5): -20 * -0.111 + (-20) * 0.031 = 1.6. In Column (6): -20 * -0.149 + (-20) * 0.026 = 2.46.
23
illiquidity in the stock market may mean that other markets will suffer lack of liquidity as well.
The results of the regression corroborate the idea that hedge funds that exit the stock market are
able to exploit investment opportunities outside the equity market during these periods.
5.
Conclusion
The behavior of hedge funds around financial crises has been of interest to academics and
policymakers. In this paper, we present evidence showing that hedge funds exit the equity market
during periods of low market liquidity. The magnitude of the effect is large: during the worst
liquidity crisis in our sample, the crisis of 2008, hedge funds reduce their positions by 18% per
quarter, over two quarters. The economic magnitude is small, however, when computed as a
fraction of market capitalization (about 0.5% per quarter). Although these declines in holdings
are material, it is not clear whether they could cause a market-wide liquidity dry-up, at least in
the equity market at large.
We present evidence suggesting that the outflows to capital providers and the voluntary
reallocation of funds across asset markets are of equal importance in equity selloffs. We show
that some of the proceeds from selling equity during crises are related to capital withdrawals by
investors and lenders. As evidence of outflows, we document that equity selling is associated
with current and future net flows. Also, funds that have higher leverage are more likely to sell
during liquidity crises. In particular, hedge funds sell high volatility stocks (which have higher
margin requirements).
We further find that hedge funds also sell equities for voluntary reasons, and reallocate
the proceeds to other, potentially more depressed, assets. In support of this conclusion, we
document that equity selloffs are correlated with inflows to hedge funds’ non-equity portfolios.
In addition, hedge funds that decrease their equity positions are more likely to take defensive
positions in equity put options, indicating their negative beliefs about future returns in the stock
market. Furthermore, we identify that hedge funds that exit the equity market are likely to have
access to and knowledge about other markets. Finally, we find that hedge funds that exit the
stock market during liquidity crises exhibit significantly higher returns in the second quarter
following the crisis, suggesting that they invested in other, potentially depressed, markets.
Our study provides new evidence about the role of hedge funds as liquidity providers.
First, we show that while hedge funds are not stable as liquidity providers as their capacity to
24
provide liquidity depends on the availability of investor funding. Our results demonstrate that
hedge funds reduce their holdings in the equity market as a response to investor redemptions.
Second, we find that hedge funds are flexible enough to reallocate their capital which is currently
invested in stocks to asset classes which are potentially more mispriced and illiquid. As a result,
hedge funds provide liquidity to investors in the other asset classes. Thus, they may act a
stabilizing force in the other asset markets.
Overall, by looking at the low frequency component of hedge fund portfolios, we
conclude that hedge funds do not seem to stabilize U.S. equity markets during liquidity crises.
Part of the reason is that they are constrained by their capital providers. The complementary
reason is that they mobilize capital to other markets. The bright side is that hedge funds appear to
provide liquidity to other markets, in which mispricing is also likely to be severe, based on the
observation of high subsequent returns for the hedge funds that sell equity during the crisis.
25
References
Acharya, Viral V., Yakov Amihud, and Sreedhar Bharath, 2009, Liquidity Risk of Corporate Bond
Returns, Working Paper.
Acharya, Viral V., and Lasse Pedersen, 2005, Asset Pricing with Liquidity Risk, Journal of Financial
Economics 77, 375-410.
Acharya, Viral V., Lasse Pedersen, Thomas Philippon, and Matthew Richardson, 2009, Regulating
Systemaic Risk, in eds. Viral Acharya and Matthew Richardson, Restoring Financial Stability: How to
Repair a Failed System, John Wiley & Sons Inc., New Jersey.
Ackermann, Carl, Richard McEnally, and David Ravenscraft, 1999, The Performance of Hedge Funds:
Risk, Return, and Incentives, Journal of Finance 54(3), 833-874.
Agarwal, Vikas, Vyacheslav Fos and Wei Jiang, 2010, Inferring Reporting Biases in Hedge Fund
Databases from Hedge Fund Equity Holdings, Columbia University Working Paper.
Agarwal, Vikas, William H. Fung, Yee Cheng Loon, and Narayan Y. Naik, 2007, Liquidity Provision in
the Convertible Bond Market: Analysis of Convertible Arbitrage Hedge Funds, Working Paper.
Agarwal,Vikas, Wei Jiang, Yuehua Tang, Baozhong Yang, 2009, Do Institutional Investors Have an Ace
Up Their Sleeves? Evidence from Confidential Filings of Portfolio Holdings, Working Paper.
Amihud, Yakov, 2002, Illiquidity and Stock Returns: Cross-Section and Time-Series Effects, Journal of
Financial Markets 5, 31–56.
Aragon, George O., 2007, Share restrictions and asset pricing: Evidence from the hedge fund industry,
Journal of Financial Economics 83(1), 33-58.
Aragon, George O., and J. Spencer Martin, 2009, A Unique View of Hedge Fund Derivatives Usage:
Safeguard or Speculation?, Working Paper.
Aragon, George O., Phillip Strahan, 2009, Hedge Funds as Liquidity Providers: Evidence from the
Lehman Bankruptcy, Working Paper.
Boyson, Nicole M., Chrostof W. Stahel, and Rene M. Stulz, 2008, Hedge Fund Contagion and Liquidity,
Ohio State University working paper.
Brophy, David J., Paige P. Ouimet, and Clemens Sialm, 2009, Hedge Funds as Investors of Last Resort?,
Review of Financial Studies 22(2), 541-574.
Brown, Stephen, Marcin Kacperczyk, Alexander Ljungqvist, Anthony Lynch, Lasse Pedersen, and
Matthew Richardson, 2009, Hedge Funds in the Aftermath of the Financial Crisis, in Eds. Viral Acharya
and Matthew Richardson, Restoring Financial Stability: How to Repair a Failed System, John Wiley &
Sons Inc., New Jersey.
Brunnermeier, Markus K., Stefan Nagel, 2004, Hedge Funds and the Technology Bubble, Journal of
Finance 59(5) 2013-2040.
Brunnermeier, Markus K., Lasse H. Pedersen, 2009, Market Liquidity and Funding Liquidity, Review of
Financial Studies 22, 2201-2238.
Cao, Charles, Yong Chen, Bing Liang, and Andrew W. Lo, 2009, Can Hedge Funds Time Market
Liquidity?, Working Paper.
Griffin, John, Jin Xu, 2009, How Smart Are the Smart Guys? Unique View from Hedge Fund Stock
Holdings, Review of Financial Studies 22(7), 2531-2570.
Gromb, Denis, Dimitri Vayanos, 2002, Equilibrium and Welfare in Markets with Financially Constrained
Arbitrageurs, Journal of Financial Economics 66, 361-407.
26
Hameed, Allaudeen, Wenjin Kang, and S. Viswanathan, 2010, Stock Market Declines and Liquidity,
Journal of Finance 65(1), 257-293.
Hombert, Joahn, and David Thesmar, 2009, Limits of Limits of Arbitrage: Theory and Evidence, HEC
Working Paper.
Khandani, Amir E., and Andrew W. Lo, 2007, What Happened to the Quants in August 2007?, MIT
Working Paper.
Khandani, Amir E., and Andrew W. Lo, 2009, Illiquidity Premia in Asset Returns: An Empirical Analysis
of Hedge Funds, Mutual Funds, and U.S. Equity Portfolios, Working Paper.
Nagel, Stefan, 2009, Evaporating Liquidity, Stanford University Working Paper.
Pastor, Lubos, and Robert F. Stambaugh, 2003, Liquidity Risk and Expected Stock Returns, Journal of
Political Economy 111(3), 642-685.
Sadka, Ronnie, 2006, Momentum and Post-Earnings-Announcement Drift Anomalies: The Role of
Liquidity Risk, Journal of Financial Economics 80, 309-349.
Sadka, Ronnie, 2010, Liquidity Risk and the Cross-Section of Hedge-Fund Returns, Journal of Financial
Economics, forthcoming.
Shleifer, Andrei, and Robert W. Vishny 1997, The Limits of Arbitrage, Journal of Finance 52(1), 35-55.
Vayanos, Dimitri, 2004, Flight to Quality, Flight to Liquidity, and the Pricing of Risk, London School of
Economics Working Paper.
Wermers, Russ, 2000, Mutual Fund Performance: An Empirical Decomposition into Stock-Picking
Talent, Style, Transactions Costs, and Expenses, Journal of Finance 55, 1655–1695.
27
Appendix
List of liquidity events (for Figures 1 and 2):
1. Iraq Invasion of Kuwait - 08/1990
2. Asian Crisis - 4/1997 and 12/1997
3. Russian Default and LTCM Crisis - 6/1998 and 10/1998
4. Internet Stocks Crisis - 03 - 04/2000
5. 9/11 Terrorist Attacks - 09/2001
6. Market Confidence Crisis 09 - 10/2002
7. Quant Liquidity Shock - 08/2007
8. Bear Stearns Collapse - 03/2008
9. Lehman Brothers' Bankruptcy - 09/2008
28
19
89
q
19 3
90
q
19 3
91
q
19 3
92
q
19 3
93
q
19 3
94
q
19 3
95
q
19 3
96
q
19 3
97
q
19 3
98
q
19 3
99
q
20 3
00
q
20 3
01
q
20 3
02
q
20 3
03
q
20 3
04
q
20 3
05
q
20 3
06
q
20 3
07
q
20 3
08
q3
19
89
q
19 3
90
q
19 3
91
q
19 3
92
q
19 3
93
q
19 3
94
q
19 3
95
q
19 3
96
q
19 3
97
q
19 3
98
q
19 3
99
q
20 3
00
q
20 3
01
q
20 3
02
q
20 3
03
q
20 3
04
q
20 3
05
q
20 3
06
q
20 3
07
q
20 3
08
q3
Figure 1. Time Series Hedge Funds Holdings (% of Total Market Capitalization)
HF Holdings (% of mkt cap)
4.00
3.00
2.00
1.00
Quarter
Figure 2. Time Series Aggregate Short Interest (% of Total Market Capitalization)
Aggregate SIR (% of mkt cap)
4.00
3.00
2.00
1.00
0.00
Quarter
29
be
l
-7 ow
0% -7
0
-6 to %
0 % -6
0
-5 to %
0 % -5
0
-4 to %
0 % -4
0
-3 to %
0 % -3
0
-2 to %
0 % -2
0%
t
- 1 o -1
0% 0
%
t
0% o 0
%
10 to 1
% 0%
20 to 2
% 0%
30 to 3
% 0%
40 to 4
% 0%
50 to 5
% 0%
60 to 6
% 0%
70 to 7
% 0%
80 to 8
% 0%
ab to 9
ov 0%
e
90
%
be
l
-7 ow
0% -7
0
-6 to %
0 % -6
0
-5 to %
0 % -5
0
-4 to %
0 % -4
0
-3 to %
0 % -3
0
-2 to %
0 % -2
0%
t
- 1 o -1
0% 0
%
t
0% o 0
%
10 to 1
% 0%
20 to 2
% 0%
30 to 3
% 0%
40 to 4
% 0%
50 to 5
% 0%
60 to 6
% 0%
70 to 7
% 0%
80 to 8
% 0%
t
ab o 9
ov 0%
e
90
%
Figure 3a. Distribution of Hedge Fund Trades, Unconditionally and During Crises
Distribution of HF trades (% of equity holdings)
.2
Unconditional distribution
Distribution in crisis
.15
.1
.05
0
Figure 3b. Distribution of Hedge Fund Trades, Unconditionally and During the 2008 Crisis
Distribution of HF trades (% of equity holdings)
.2
Unconditional distribution
Distribution in 2008 (Q3/Q4)
.15
.1
.05
0
30
Table 1. Summary Statistics
The table presents summary statistics for the data used in the study. Panel A presents summary statistics for the
hedge fund holdings sample, aggregated at the calendar quarter level. Panel B similarly presents summary
statistics for the hedge fund holdings sample, aggregated at the stock-calendar quarter level. Panel C
provides the same statistics, aggregated at the hedge-fund-calendar quarter level. Panel D presents timeseries summary statistics for hedge funds, aggregated at the hedge-fund-year level. Panel E shows correlations
between the market-condition variables used in the study. Panel F lists the crises for the market condition variables.
A crisis quarter is defined as a quarter in which the market condition variable is two standard deviations or more
from the sample mean.
Panel A: Summary Statistics for Aggregate Sample (Quarterly Frequency)
N
78
78
78
78
78
78
78
78
78
78
78
77
73
78
78
HF holdings over mkt cap (%)
∆ HF Holdings (%, share of equity holdings)
∆ HF Holdings (%, share of mkt cap)
MF holdings over mkt cap (%)
∆ MF Holdings (%, share of mkt cap)
Other institutional holdings over mkt cap (%)
∆ Other institutional holdings (%, share of mkt cap)
Retail holdings over mkt cap (%)
∆ Retail holdings (%, share of mkt cap)
Short interest ratio (SIR) (%)
∆ Short interest ratio (∆ SIR) (%)
Pastor-Stambaugh (PS)
Minus Acharya-Pedersen (AP)
Minus ∆VIX (VIX)
Rm- Rf
Mean
1.840
1.610
0.027
11.900
0.105
40.000
0.157
43.300
-0.317
1.750
0.035
0.000
0.000
0.000
0.010
St.Dev.
0.686
6.880
0.145
3.500
0.232
4.410
0.900
7.660
1.680
0.738
0.167
1.000
1.000
1.000
0.084
Min
1.060
-16.600
-0.583
4.920
-0.852
31.900
-2.740
28.500
-5.170
0.589
-0.683
-2.770
-2.800
-3.490
-0.240
Median
1.650
0.652
0.011
13.000
0.108
38.300
0.233
45.600
-0.532
1.650
0.035
0.217
-0.015
0.044
0.016
Max
3.750
37.000
0.499
17.000
0.725
49.800
2.730
54.900
10.000
3.770
0.653
2.740
2.250
2.710
0.203
Panel B: Summary Statistics for Stock-Level Sample (Quarterly Frequency)
Hedge fund equity holdings (%)
∆ Hedge fund equity holdings (%)
Short interest ratio (SIR) (%)
∆ Short interest ratio (SIR) (%)
Mktcap ($bn)
Volatility
Past ret 12
N
449256
431438
573099
543633
470817
421236
453907
31
Mean
St.Dev.
3.096
5.883
0.070
1.214
1.990
4.663
0.043
2.455
1602.761 10083.571
0.151
0.095
0.147
0.862
Min
0.000
-5.419
0.000
-92.309
0.011
0.000
-0.999
Median
Max
0.780
100.000
0.000
6.570
0.341
100.000
0.000
89.516
111.994 602432.938
0.126
0.500
0.03
58.68
Panel C: Summary Statistics for Hedge-Fund-Level Sample (Quarterly Frequency)
N
18091
4586
4949
4688
16759
15815
4547
18091
5328
16032
12278
5042
5328
5328
5328
5328
5328
5328
5328
5328
5328
5328
5328
∆ HF equity holdings (%, out of lagged equity holdings)
Fund flows (%, out of lagged equity holdings)
Fund return next quarter (%, ret(q+1))
Fund return in two quarters (%, ret(q+2))
Equity holdings return next quarter (%, eqret(q+1))
Equity holdings return in two quarters (%, eqret(q+2))
Return past 12 months (%, Past ret 12)
Equity portfolio size (log(assets))
Out of TASS sample in 1 year (%)
Out of 13-F sample in 1 year (%)
Put dummy
Average leverage
Convertible arbitrage strategy
Short bias strategy
Emerging markets strategy
Market neutral strategy
Event driven strategy
Fixed income arbitrage strategy
Fund of funds strategy
Global macro strategy
Long-short strategy
Futures strategy
Multi strategy
32
Mean St.Dev.
Min
Median Max
4.590 27.100 -66.100 1.210 122.000
-3.360 130.000 -2810.000 0.304 302.000
2.240
8.880
-56.200 2.150 77.500
2.200
8.970
-56.200 2.170 77.500
0.881 14.400 -81.400 1.920 246.000
0.798 14.500 -81.400 1.930 246.000
11.200 21.800 -84.900 9.770 225.000
19.200 1.590
9.730
19.200 24.700
12.800 31.500
0.000
0.000 100.000
14.100 34.800
0.000
0.000 100.000
0.320
0.467
0.000
0.000 1.000
0.471
0.799
0.000
0.000 8.350
0.054
0.216
0.000
0.000 1.000
0.002
0.029
0.000
0.000 0.729
0.020
0.127
0.000
0.000 1.000
0.043
0.190
0.000
0.000 1.000
0.198
0.388
0.000
0.000 1.000
0.014
0.112
0.000
0.000 1.000
0.028
0.150
0.000
0.000 1.000
0.040
0.188
0.000
0.000 1.000
0.520
0.488
0.000
0.837 1.000
0.017
0.119
0.000
0.000 1.000
0.065
0.221
0.000
0.000 1.000
Panel D: Summary Statistics for Hedge-Fund-Level, by Year
Year
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
2007
2008
2009
Total Assets
Under Mngmt
in TASS ($bn)
2.2
1.6
2.6
4.6
8.8
14.3
17.3
22.8
30.7
40.0
29.0
39.0
41.5
52.2
65.3
93.1
112.2
147.4
189.4
149.2
77.1
Number of Mgrs.
13F
TASS match
38
3
46
5
50
8
59
10
68
16
75
22
88
23
101
27
113
31
158
52
171
59
220
72
258
83
272
87
295
97
366
109
441
132
520
140
595
145
629
119
545
76
Avg. Equity portfolio ($m)
Mean Median St. dev.
1,264
278
3,406
1,076
245
2,904
1,315
238
3,518
1,326
207
3,588
1,323
219
3,581
869
180
2,114
961
252
2,090
1,101
270
2,848
971
283
2,381
902
263
2,222
925
250
2,547
848
261
2,638
599
163
1,812
496
137
1,530
572
176
1,672
653
240
1,599
819
271
2,013
891
250
2,386
977
269
2,804
648
146
2,010
418
87
1,300
Number of Stocks per manager
Mean
Median
St. dev.
179.0
87.5
271
184.0
79.0
277
194.0
82.0
295
184.0
79.0
289
184.0
76.0
363
133.0
74.0
247
135.0
72.0
177
137.0
58.0
204
125.0
55.0
204
119.0
53.0
218
104.0
49.0
196
92.3
46.0
179
91.9
41.0
190
85.5
41.0
165
94.5
43.0
186
91.9
42.0
193
96.5
40.0
216
98.3
35.0
238
93.1
34.0
229
72.1
25.0
204
67.0
21.0
197
Portfolio turnover
Mean Median St. dev.
0.62
0.48
0.56
0.70
0.52
0.64
0.71
0.58
0.57
0.71
0.55
0.61
0.80
0.63
0.66
0.80
0.62
0.68
0.87
0.74
0.67
0.88
0.75
0.65
0.96
0.78
0.74
0.91
0.75
0.74
1.05
0.90
0.79
1.15
1.02
0.84
1.15
1.02
0.89
1.19
1.07
0.87
1.23
1.12
0.86
1.22
1.10
0.81
1.16
1.04
0.77
1.12
1.03
0.75
1.15
1.03
0.79
1.06
0.93
0.81
1.16
0.96
0.99
Panel E: Correlations of Market Condition Variables
Pastor-Stambaugh (PS)
Minus Acharya-Pedersen (AP)
Minus ∆VIX (VIX)
Rm- Rf
PS
1.00
0.23
0.35
0.35
Factor correlations
AP
VIX
1.00
0.15
0.24
1.00
0.71
Rm - Rf
1.00
Panel F: Crisis Quarters (Quarters in which Market Condition Factor < -2σ)
Market Condition Factor
PS
AP
VIX
1998q3
1998q3
1998q3
2000q2
2007q3
2001q3
2007q3
2008q1
2008q3
Rm - Rf
1990q3
2001q3
2002q3
2008q3
33
Table 2. Hedge Fund Trades and Aggregate Liquidity
The table presents time-series OLS regressions. The explanatory variable is the change in aggregate hedge fund
dollar holdings between two quarters. To be included in the sample, a hedge fund must have equity holdings in both
quarters. Panel A and C regress the changes in aggregate hedge fund holdings on market condition variables
dummies. Panels B and D regress the changes in aggregate hedge fund holdings on contemporaneous, lagged, and
lead market condition variables. Panels A and B express the changes in hedge fund holdings as a percentage of total
hedge fund holdings. Panels C and D express the changes in hedge fund holdings as a percentage of total equity
market cap, using lagged quarter valuations. In Columns (1) and (2), the market condition variable is the PastorStambaugh liquidity index. In Columns (3), (4), and (5), the market condition variables are the Acharya-Pedersen
liquidity index, the VIX index, and the market’s excess returns, respectively. All regressions have a constant which
is not presented. t-statistics are presented in parentheses. ***, **, and * represent statistical significance at the 1%,
5%, and 10% levels, respectively.
Panel A: Changes in aggregate hedge fund holdings and non-parametric liquidity measures
Mkt Cond ≥ 2σ
σ ≤ Mkt Cond < 2σ
Dependent variable: ∆ HF total holdings (%)
PS
AP
VIX
Rm - R f
PS
(1)
(2)
(3)
(4)
(5)
3.889
-4.512
3.598
-5.910
4.167
(0.579)
(-0.713)
(0.802) (-1.179)
(0.659)
-0.689
2.303
-5.054
-1.177
-0.442
(-0.273)
(1.060)
(-1.187) (-0.422) (-0.186)
-σ ≤ Mkt Cond < σ
-2σ ≤ Mkt Cond < -σ
Mkt Cond < -2σ
-2.159
-1.166
(-0.677)
(-0.439)
-9.863*** -10.768**
(-2.802)
(-2.277)
-0.002
(-0.000)
-5.293
(-1.144)
-0.016
(-0.006)
-2.637
(-0.731)
0.467
(0.150)
-6.881*
(-1.994)
-16.443***
(-3.116)
dummy(Q3/Q4-2008)
FE for market returns
Yes
Yes
Yes
No
Yes
Observations
77
73
78
78
77
2
0.035
0.024
-0.034
-0.027
0.145
Adj R
34
Table 2. Hedge Fund Trades and Aggregate Liquidity (Cont.)
Panel B: Changes in aggregate hedge fund holdings and lead/lags of liquidity measures
lag(Mkt Cond < -2σ)
Mkt Cond < -2σ
lead(Mkt Cond < -2σ)
dummy(Q3/Q4-2008)
lag(Mkt ret)
Dependent variable: ∆ HF total holdings (%)
PS
AP
VIX
Rm - Rf
PS
(1)
(2)
(3)
(4)
(5)
-2.915
1.297
-6.243
0.771
-3.063
(-0.758)
(0.191)
(-1.610) (0.215)
(-0.820)
-9.159*** -10.980**
-5.859
-2.472
-9.575**
(-2.910)
(-2.368)
(-1.541) (-0.689) (-2.563)
1.610
3.727
5.482
-2.767
1.109
(0.470)
(0.805)
(1.424) (-0.772)
(0.332)
-19.533***
(-2.852)
-7.495
(-0.742)
-11.594
(-1.177)
9.520
(1.021)
-3.981
(-0.403)
-12.028
(-1.184)
7.097
(0.721)
-12.021
(-1.172)
-1.514
(-0.152)
23.996**
(2.458)
Observations
76
72
78
78
76
2
0.065
0.015
0.079
-0.025
0.119
Mkt ret
lead(Mkt ret)
Adj R
-6.439
(-0.657)
-12.228
(-1.298)
4.508
(0.464)
Panel C: Changes in aggregate hedge fund holdings (measured as % of total market
capitalization) and non-parametric liquidity measures
Mkt Cond ≥ 2σ
σ ≤ Mkt Cond < 2σ
Dependent variable: ∆ HF holdings (% of total mktcap)
PS
AP
VIX
Rm - Rf
PS
(1)
(2)
(3)
(4)
(5)
0.026
-0.068
0.030
-0.088
0.034
(0.191) (-0.604) (0.317) (-0.832)
(0.289)
0.013
0.044
-0.084
-0.025
0.021
(0.262)
(1.131) (-0.939) (-0.422)
(0.464)
-σ ≤ Mkt Cond < σ
-2σ ≤ Mkt Cond < -σ
Mkt Cond < -2σ
-0.075
-0.024
-0.013
(-1.150) (-0.511) (-0.139)
-0.247*** -0.279*** -0.166*
(-3.447) (-3.298) (-1.709)
-0.037
(-0.673)
-0.101
(-1.325)
0.002
(0.035)
-0.160**
(-2.475)
-0.480***
(-4.859)
dummy(Q3/Q4-2008)
FE for market returns
Yes
Yes
Yes
No
Yes
Observations
77
73
78
78
77
2
0.108
0.088
-0.017
-0.018
0.331
Adj R
35
Table 2. Hedge Fund Trades and Aggregate Liquidity (Cont.)
Panel D: Changes in aggregate hedge fund holdings (measured as % of total market
capitalization) and lead/lags of liquidity measures
lag(Mkt Cond < -2σ)
Mkt Cond < -2σ
lead(Mkt Cond < -2σ)
dummy(Q3/Q4-2008)
lag(Mkt ret)
Dependent variable: ∆ HF holdings (% of total mktcap)
PS
AP
VIX
Rm - Rf
PS
(1)
(2)
(3)
(4)
(5)
-0.068
-0.001
-0.147*
0.022
-0.077
(-0.901) (-0.008) (-1.895) (0.301)
(-1.135)
-0.223*** -0.283*** -0.184** -0.099 -0.222***
(-3.612) (-3.542) (-2.419) (-1.344) (-3.288)
0.098
0.169**
0.123
-0.151**
0.079
(1.461)
(2.120)
(1.597) (-2.051)
(1.304)
-0.617***
(-4.990)
-0.158
(-0.797)
-0.123
(-0.637)
0.351*
(1.922)
-0.061
(-0.360)
-0.149
(-0.852)
0.217
(1.279)
-0.250
(-1.217)
0.060
(0.302)
0.639***
(3.272)
Observations
76
72
78
78
76
2
0.166
0.139
0.180
0.037
0.334
Mkt ret
lead(Mkt ret)
Adj R
36
-0.137
(-0.773)
-0.170
(-0.997)
0.151
(0.859)
Table 3. The Distribution of Hedge Fund Trades
The table presents results about the distribution of hedge fund trades. Panel A presents the distribution of hedge fund
trades with respect to innovations in the Pastor and Stambaugh liquidity index. Panel B repeats the analysis in Panel
A, but weighs the buckets by hedge fund dollar holdings.
Panel A: Equally weighted distribution of hedge fund trades
% HFs that trade:
Buy 20%+
Buy 10%-20%
Buy 5%-10%
Unchanged ±5%
Sell 5%-10%
Sell 10%-20%
Sell 20%-40%
Sell 40%+
N
Unconditional
24.01
10.48
7.83
21.76
7.92
10.33
10.38
7.30
17,546
PS < -2σ
20.59
10.21
6.70
18.31
6.33
11.66
12.62
13.58
2,195
Equally-Weighted
-2σ ≤ PS < -σ -σ ≤ PS < 0 0 ≤ PS < σ
25.70
25.10
23.53
8.97
9.84
10.98
5.53
8.01
8.68
17.18
22.38
23.24
5.73
8.37
8.58
9.48
10.06
10.56
15.27
9.87
9.12
12.15
6.38
5.32
1,572
4,156
6,886
σ ≤ PS < 2σ
25.10
11.52
7.70
22.20
8.18
9.67
9.79
5.84
2,482
PS ≥ 2σ
27.84
9.02
6.67
25.49
7.45
8.63
9.02
5.88
255
Q3-Q4/2008
18.32
6.84
4.65
13.58
4.38
9.99
18.84
23.40
1,141
σ ≤ PS < 2σ
16.90
14.31
9.92
29.38
PS ≥ 2σ
11.94
6.21
17.65
46.68
Q3-Q4/2008
9.30
6.16
5.85
19.14
7.63
11.14
8.37
2.35
2,482
6.86
3.98
4.58
2.11
255
6.77
13.45
15.93
23.39
1,141
Panel B: Value-weighted distribution of hedge fund trades
% HFs that trade:
Buy 20%+
Buy 10%-20%
Buy 5%-10%
Unchanged ±5%
Sell 5%-10%
Sell 10%-20%
Sell 20%-40%
Sell 40%+
N
Unconditional
13.78
11.03
10.52
31.50
PS < -2σ
10.61
9.85
10.26
20.88
9.67
10.90
8.07
4.54
17,546
12.49
11.10
12.72
12.08
2,195
Value-Weighted
-2σ ≤ PS < -σ -σ ≤ PS < 0 0 ≤ PS < σ
17.08
14.20
12.99
7.45
9.08
12.19
3.97
9.27
12.55
27.31
36.41
33.43
7.82
11.79
14.12
10.45
1,572
8.86
11.19
7.94
3.06
4,156
37
10.29
10.60
5.38
2.57
6,886
Table 4. Who Buys What Hedge Funds Sell?
The table presents results about trades during crises by investor type. The table regresses changes in aggregate
holdings by investor type on crisis indicators. Crisis is measured based on the Pastor and Stambaugh index. In
Columns (1) and (2), the dependent variable is the change in aggregate holdings by mutual funds. In Columns (3)
and (4), the dependent variable is the change in aggregate holdings by institutions which are not hedge funds or
mutual funds. In Columns (5) and (6), the dependent variable is the change in aggregate holdings by firms’
management and retail investors. All regressions include a constant, which is not presented. t-statistics are presented
in parentheses. ***, **, and * represent statistical significance at the 1%, 5%, and 10% levels, respectively.
PS ≥ 2σ
σ ≤ PS < 2σ
Dependent variable: ∆ total holdings (%) of
Mutual funds
Other institutions
Management & retail investors
(1)
(2)
(3)
(4)
(5)
(6)
0.088
0.089
0.527
0.532
-0.586
-0.608
(0.380)
(0.378)
(0.597)
(0.599)
(-0.338)
(-0.351)
0.059
0.059
0.207
0.212
-0.688
-0.708
(0.670)
(0.668)
(0.625)
(0.635)
(-1.058)
(-1.087)
-σ ≤ PS < σ
-2σ ≤ PS < -σ
-0.237**
(-2.145)
0.023
(0.185)
-0.235**
(-2.027)
0.026
(0.200)
-0.017
(-0.087)
-0.320
(-0.764)
0.745
(1.613)
-0.271
(-0.619)
0.801
(1.655)
-0.307
(-0.415)
0.472
(0.574)
-0.875
(-0.964)
0.256
(0.300)
-1.120
(-1.185)
1.353
(0.935)
FE for market returns
Yes
Yes
Yes
Yes
Yes
Yes
Observations
77
77
77
77
77
77
2
0.022
0.008
-0.003
-0.015
-0.045
-0.047
PS < -2σ
dummy(Q3/Q4-2008)
Adj R
38
Table 5. What Explains Hedge Funds’ Equity Market Participation During Crises?
The table presents results about the relation between hedge fund trades and net fund flows, hedge fund leverage,
flows into other assets, and investment in put options. In Panel A, the market condition variable is the PastorStambaugh liquidity index. The sample used in Column (7) is restricted to observations in the last two quarters of
2008. In Panel B, the market condition variable is the Acharya-Pedersen liquidity index. In Panel C, the market
condition variable is the VIX index. In Panel D, the market condition variable is the market’s excess returns. All
regressions include a constant, which is not presented. t-statistics are presented in parentheses. ***, **, and *
represent statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are clustered by
calendar quarter, except for Column (7), in which there is no clustering.
Panel A: Explaining changes in hedge fund equity holdings (crisis defined by Pastor and
Stambaugh)
PS < -2σ
PS < -2σ x Fund flows
PS < -2σ x lead(Fund flows)
Dependent variable: ∆ HF equity portfolio (%)
Q3-Q4/2008
Full sample
(1)
(2)
(3)
(4)
(5)
(6)
(7)
-7.559** -5.527** -4.892
-3.679
-4.786
-1.088
(-2.449) (-2.378) (-1.506) (-1.407) (-1.664) (-0.498)
-0.033
-0.031
-0.034*
(-1.596)
(-1.581)
(-1.760)
-0.041*
-0.041*
-0.041*
PS < -2σ x lead2(Fund flows)
(-1.839)
0.002
(0.222)
(-1.852)
0.001
(0.178)
-4.592*** -3.306
(-2.825) (-1.636)
0.032*
(1.688)
0.063***
(2.838)
-0.000
(-0.016)
0.032*
(1.688)
0.063***
(2.850)
-0.000
(-0.017)
-0.105
(-0.117)
PS < -2σ x Leverage
PS < -2σ x Put dummy
Fund flows
lead(Fund flows)
lead2(Fund flows)
Hedge fund leverage
0.083
(0.089)
Put option dummy
Past ret 12
log(Portfolio size)
10.187**
(2.366)
-6.957***
(-6.869)
(-1.829)
0.001
(0.158)
-3.082
(-1.543)
-7.380*** -7.164***
(-3.850) (-3.379)
1.509
(1.004)
5.079
9.890**
4.836 10.180**
(1.222) (2.312) (1.164) (2.370)
-6.689*** -6.985*** -6.712*** -6.960***
(-7.271) (-6.894) (-7.221) (-6.787)
0.032*
-0.032***
(1.703)
(-4.862)
0.063*** 0.006***
(2.838)
(2.756)
-0.000
(-0.008)
-0.215
-1.673
(-0.223)
(-0.532)
1.535
-7.488
(0.990)
(-1.196)
4.826
56.450***
(1.165)
(3.074)
-6.724***
-0.640
(-7.113)
(-0.295)
Observations
2283
2283
2283
2283
2283
2283
135
2
0.078
0.105
0.078
0.105
0.078
0.105
0.196
Adj R
39
Table 5. What Explains Hedge Funds’ Equity Market Participation During Crises? (Cont.)
Panel B: Explaining changes in hedge fund equity holdings (crisis defined by Acharya and
Pedersen)
AP < -2σ
AP < -2σ x Fund flows
AP < -2σ x lead(Fund flows)
Dependent variable: ∆ HF equity portfolio (%)
Full sample
(1)
(2)
(3)
(4)
(5)
(6)
-8.675*** -4.695*** -4.839*** -2.388* -6.373*** -0.301
(-7.803) (-4.370) (-4.074) (-1.994) (-6.293) (-0.283)
-0.069***
-0.072***
-0.074***
(-2.781)
(-2.894)
(-2.985)
0.174***
0.163***
0.162***
(7.265)
-0.036**
(-2.163)
AP < -2σ x lead2(Fund flows)
AP < -2σ x Leverage
(6.800)
-0.025
(-1.475)
-6.371*** -4.205***
(-6.964) (-4.443)
AP < -2σ x Put dummy
Fund flows
0.029
(1.148)
0.065**
(2.678)
0.019
(1.077)
lead(Fund flows)
lead2(Fund flows)
Hedge fund leverage
0.092
(0.100)
Put option dummy
Past ret 12
log(Portfolio size)
8.313*
(1.965)
-7.426***
(-6.933)
(6.742)
-0.022
(-1.299)
-4.162***
(-4.051)
-5.663*** -5.229***
(-3.805) (-3.334)
0.029
(1.149)
0.065**
(2.683)
0.019
(1.074)
-0.057
(-0.062)
1.704
(1.117)
3.144
8.103*
2.950
8.349*
(0.740) (1.924) (0.694) (1.978)
-6.850*** -7.418*** -6.845*** -7.455***
(-6.651) (-6.888) (-6.543) (-6.891)
0.029
(1.150)
0.065**
(2.667)
0.019
(1.082)
-0.172
(-0.176)
1.539
(0.954)
2.948
(0.696)
-6.877***
(-6.482)
Observations
2080
2080
2080
2080
2080
2080
2
0.082
0.108
0.083
0.108
0.082
0.107
Adj R
40
Table 5. What Explains Hedge Funds’ Equity Market Participation During Crises? (Cont.)
Panel C: Explaining changes in hedge fund equity holdings (crisis defined by VIX)
VIX < -2σ
VIX < -2σ x Fund flows
VIX < -2σ x lead(Fund flows)
Dependent variable: ∆ HF equity portfolio (%)
Full sample
(1)
(2)
(3)
(4)
(5)
(6)
-4.311
-1.493
-3.432
-1.632
-3.998
-1.608
(-0.556) (-0.228) (-0.488) (-0.265) (-0.631) (-0.341)
-0.006
-0.006
-0.005
(-0.275)
(-0.295)
(-0.211)
-0.034
-0.034
-0.034
(-1.463)
0.003
(0.491)
VIX < -2σ x lead2(Fund flows)
VIX < -2σ x Leverage
-1.486
(-0.807)
(-1.454)
0.003
(0.496)
0.228
(0.132)
VIX < -2σ x Put dummy
-0.892
(-0.138)
Fund flows
0.029
(1.598)
0.053**
(2.323)
-0.001
(-0.121)
lead(Fund flows)
lead2(Fund flows)
Hedge fund leverage
-0.397
(-0.423)
Put option dummy
Past ret 12
log(Portfolio size)
10.209**
(2.507)
-7.043***
(-6.964)
0.029
(1.609)
0.053**
(2.312)
-0.001
(-0.131)
-0.456
(-0.502)
0.596
(0.414)
5.727 10.049** 5.571 10.244**
(1.413) (2.446) (1.364) (2.517)
-6.817*** -7.072*** -6.847*** -7.057***
(-7.362) (-7.003) (-7.336) (-6.887)
(-1.457)
0.003
(0.490)
0.301
(0.173)
-0.109
(-0.016)
0.029
(1.613)
0.053**
(2.315)
-0.001
(-0.125)
-0.517
(-0.542)
0.810
(0.555)
5.603
(1.374)
-6.873***
(-7.248)
Observations
2283
2283
2283
2283
2283
2283
2
0.074
0.100
0.073
0.100
0.073
0.099
Adj R
41
Table 5. What Explains Hedge Funds’ Equity Market Participation During Crises? (Cont.)
Panel D: Explaining changes in hedge fund equity holdings (crisis defined by Rm-Rf)
Rm-Rf < -2σ
Rm-Rf < -2σ x Fund flows
Rm-Rf < -2σ x lead(Fund flows)
Dependent variable: ∆ HF equity portfolio (%)
Full sample
(1)
(2)
(3)
(4)
(5)
(6)
5.291
3.136
5.334
2.900
2.025
-1.110
(0.769) (0.451) (1.008) (0.528) (0.285) (-0.195)
0.180***
0.180***
0.211***
(8.647)
(6.371)
(8.349)
-0.160*
-0.159*
-0.182**
Rm-Rf < -2σ x lead2(Fund flows)
(-1.884)
0.063***
(9.688)
(-1.714)
0.063***
(8.395)
-0.156
0.399
(-0.046) (0.139)
(-2.473)
0.061***
(7.909)
0.430
(0.128)
12.860*** 14.577***
(4.430) (4.557)
0.033*
(1.981)
0.029***
(3.595)
0.001
(0.437)
0.033*
(1.990)
0.029***
(3.598)
0.001
(0.401)
-0.477
-0.435
(-0.528) (-0.493)
0.033*
(1.993)
0.029***
(3.600)
0.001
(0.408)
-0.473
(-0.512)
0.516
(0.357)
7.304
(1.637)
-6.896***
(-7.320)
Rm-Rf < -2σ x Leverage
Rm-Rf < -2σ x Put dummy
Fund flows
lead(Fund flows)
lead2(Fund flows)
Hedge fund leverage
Put option dummy
Past ret 12
log(Portfolio size)
11.740**
(2.422)
-7.003***
(-6.981)
0.309
(0.216)
7.354 11.568** 7.199 11.830**
(1.666) (2.372) (1.616) (2.438)
-6.834*** -7.036*** -6.864*** -7.030***
(-7.410) (-6.986) (-7.368) (-6.940)
Observations
2283
2283
2283
2283
2283
2283
2
0.074
0.100
0.073
0.099
0.074
0.099
Adj R
42
Table 6. Hedge Fund Trades and Stock Volatility
The table tests whether hedge funds sell more high volatility stocks during crises. Panels A and B are at the stockcalendar quarter level. Panel A regresses changes in stock-level aggregate hedge fund holdings on crisis indicators
interacted with a high volatility indicator (above median volatility, within the quarter). Panel B regresses changes in
stock-level aggregate short interest on crisis indicators interacted with high volatility indicator (above-median
volatility, within the quarter). In Column (1), the market condition variable is the Pastor-Stambaugh liquidity index.
In Column (2), it is the Acharya-Pedersen liquidity index, and the VIX index in Column (3). In Column (4), the
market condition variable is the market’s excess returns. In Column (5), the sample is restricted to the last two
quarters of 2008. Panel C presents probit regressions (marginal effects are presented) of whether hedge funds
dropped out of the sample of 13(f) and TASS, as a function of their trades during crises. In Panel C, Columns (1)
and (5), (2) and (6), (3) and (7), and (4) and (8), the market condition variables are the Pastor-Stambaugh liquidity
index, the Acharya-Pedersen liquidity index, the VIX index, and the market’s excess returns, respectively. All
regressions include a constant, which is not presented. t-statistics are presented in parentheses. ***, **, and *
represent statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors are clustered by
calendar quarter, except for Column (5).
Panel A: Changes in hedge fund holdings and stock volatility
Mkt Cond < -2σ x High volatility
Mkt Cond < -2σ
High volatility
log(mktcap)
Past ret 12
Observations
Adj. R
2
Dependent variable: ∆ HF holdings (%)
PS
AP
VIX
Rm - Rf Q3-Q4/2008
(1)
(2)
(3)
(4)
(5)
-0.192*** -0.161*** -0.006
-0.080
(-6.486) (-4.556) (-0.387) (-1.229)
-0.187
-0.271** -0.020
-0.246
(-1.490) (-2.367) (-0.932) (-1.462)
0.040** 0.049***
0.023
0.027
-0.325***
(2.083)
(2.881)
(1.148)
(1.440)
(-5.975)
0.012
0.018***
0.010
0.010
-0.103***
(1.594)
(2.745)
(1.250)
(1.313)
(-7.562)
0.034*** 0.019*
0.036** 0.032*** 0.169***
(2.746)
(1.993)
(2.640)
(2.732)
(2.680)
385364
369469
389379
389379
7844
0.002
0.002
0.001
0.002
0.009
43
Table 6. Hedge Fund Trades and Stock Volatility (Cont.)
Panel B: Changes in short interest ratio and stock volatility
Mkt Cond < -2σ x High volatility
Mkt Cond < -2σ
High volatility
log(mktcap)
Past ret 12
Dependent variable: ∆ short interest ratio (SIR) (%)
PS
AP
VIX
Rm - Rf Q3-Q4/2008
(1)
(2)
(3)
(4)
(5)
-0.287*** -0.264 0.079*** -0.226
(-3.839) (-1.661)
(3.632) (-1.353)
0.086
-0.153** -0.123*** -0.323
(0.453) (-2.002) (-4.297) (-0.868)
-0.003
0.014
-0.027
-0.014
-0.839***
(-0.106) (0.719) (-0.816) (-0.523)
(-10.494)
0.017
0.021**
0.017
0.018
-0.207***
(1.299)
(2.622)
(1.303)
(1.445)
(-10.318)
0.106*** 0.089*** 0.106*** 0.100*** 0.576***
(4.805)
(6.210)
(4.685)
(5.663)
(6.528)
Observations
368877
352068
373041
373041
8370
2
0.004
0.004
0.004
0.006
0.022
Adj. R
Panel C: Survival of Hedge Funds Following Liqudity Crises
Exits in one year (13(f))
PS
AP
VIX
Rm - Rf
(1)
(2)
(3)
(4)
Mkt Cond < -2σ
0.041**
0.021 0.051*** 0.039***
(2.352) (1.034) (3.444) (4.059)
Mkt Cond < -2σ x max(Δ Equity portfolio,0)
-0.000
0.000* -0.000* -0.000*
(-0.854) (1.857) (-1.849) (-1.817)
Mkt Cond < -2σ x min(Δ Equity portfolio,0)
0.000
0.001
0.001
0.001
(0.769) (1.031) (1.571) (1.084)
max(Δ Equity portfolio,0)
-0.000*** -0.000*** -0.000*** -0.000***
(-2.947) (-2.882) (-2.943) (-2.932)
min(Δ Equity portfolio,0)
-0.002*** -0.002*** -0.002*** -0.002***
(-7.461) (-7.564) (-8.733) (-8.575)
log(total assets)
-0.045*** -0.043*** -0.045*** -0.045***
(-16.176) (-17.245) (-16.608) (-16.579)
Exits in one year (TASS)
PS
AP
VIX
Rm - Rf
(5)
(6)
(7)
(8)
0.131*
0.156
-0.039 -0.113***
(1.888) (1.550) (-0.532) (-4.475)
0.000
-0.000
0.001 0.002***
(0.354) (-1.359) (1.036) (6.047)
0.000
-0.000
0.000
-0.002*
(0.019) (-0.444) (0.495) (-1.747)
0.000* 0.000**
0.000
0.000
(1.867) (2.337) (1.513) (1.229)
-0.002*** -0.002*** -0.003*** -0.003***
(-3.672) (-3.054) (-5.063) (-5.047)
0.010** 0.016*** 0.010*
0.010*
(2.007) (3.749) (1.852) (1.811)
Observations
15779
15150
15779
15779
5244
4800
5328
5328
Pseudo R2
0.0903
0.0849
0.0897
0.0894
0.0193
0.0150
0.0101
0.0119
44
Table 7. Hedge Fund Behavior during Crises, by Hedge Fund Style
The table presents results about the relation between hedge fund trades and hedge fund style. In Columns (1) and
(2), the market condition variable is the Pastor-Stambaugh liquidity index. In Columns (3) and (4), it is the AcharyaPedersen liquidity index, and the VIX index in Columns (5) and (6). In Columns (7) and (8), the market condition
variable is the market’s excess returns. The sample used in Column (9) is restricted to observations in the last two
quarters of 2008. All regressions include a constant, which is not presented. t-statistics are presented in parentheses.
***, **, and * represent statistical significance at the 1%, 5%, and 10% levels, respectively. Robust standard errors
are clustered by calendar quarter, except for Column (9), in which there is no clustering.
PS
Mkt Cond < -2σ
(1)
-6.582***
(-3.007)
Mkt Cond < -2σ x Convertible arb
-9.481
(-1.221)
-32.391
(-0.725)
-6.645
(-1.208)
16.987***
(3.274)
-11.693**
(-2.484)
-20.053*
(-1.970)
-14.276
(-1.193)
-27.611***
(-3.048)
-1.872
(-0.733)
-19.782
(-1.191)
-14.758
(-1.403)
Mkt Cond < -2σ x Short bias
Mkt Cond < -2σ x Emerging
Mkt Cond < -2σ x Market neutral
Mkt Cond < -2σ x Event driven
Mkt Cond < -2σ x Fixed income arb
Mkt Cond < -2σ x Fund of funds
Mkt Cond < -2σ x Global macro
Mkt Cond < -2σ x Long-short
Mkt Cond < -2σ x Futures
Mkt Cond < -2σ x Multistrategy
Convertible arbitrage strategy
Short bias strategy
Emerging markets strategy
Market neutral strategy
Event driven strategy
Fixed income arbitrage strategy
Global macro strategy
Long-short strategy
Futures strategy
Multi strategy
Past ret 12
log(Portfolio size)
Observations
Adj R2
(2)
6.701
(1.234)
-6.604
(-0.177)
-7.948
(-1.233)
-3.421
(-0.710)
10.516**
(2.398)
6.291
(1.508)
-1.990
(-0.431)
-6.188
(-1.482)
-6.789
(-1.072)
5.178
(0.912)
14.623*** 16.195***
(4.938)
(5.786)
-6.109*** -5.494***
(-8.775) (-7.954)
Dependent variable: ∆ HF equity portfolio (%)
AP
VIX
Rm - Rf
(3)
(4)
(5)
(6)
(7)
(8)
-5.820***
-4.957
-2.248
(-5.540)
(-1.105)
(-0.542)
-4.391
-3.479
-12.369
(-0.535)
(-0.214)
(-0.463)
38.115
106.579***
106.039***
(1.046)
(2.989)
(2.973)
-11.152**
3.396
64.257***
(-2.588)
(0.319)
(4.389)
20.690*
33.981***
0.995
(1.778)
(2.895)
(0.084)
-14.403***
-14.130**
-5.019
(-6.142)
(-2.430)
(-1.341)
-25.754**
13.064
21.666
(-2.466)
(0.652)
(0.756)
-12.594***
-31.062***
-2.012
(-2.674)
(-4.378)
(-0.292)
-36.574**
-8.471**
-16.939***
(-2.563)
(-2.395)
(-3.039)
0.503
-4.219
-1.734
(0.210)
(-1.640)
(-0.778)
-50.830***
-14.765**
44.757***
(-9.707)
(-2.568)
(2.715)
-4.964*
13.474*
-7.859
(-1.731)
(1.712)
(-0.356)
7.457
(1.282)
-10.120
(-0.289)
-9.405
(-1.435)
-4.996
(-0.985)
10.866**
(2.175)
6.018
(1.298)
-1.294
(-0.251)
-6.330
(-1.322)
-9.686
(-1.441)
8.819
(1.517)
11.276*** 12.893***
(4.547)
(5.526)
-6.249*** -5.547***
(-8.478) (-7.721)
6.347
(1.288)
-22.825
(-0.638)
-9.383
(-1.620)
-2.872
(-0.670)
11.046***
(2.819)
4.921
(1.368)
-3.766
(-0.844)
-5.523
(-1.488)
-9.300
(-1.594)
2.105
(0.390)
14.076*** 15.648***
(4.828)
(5.613)
-6.419*** -5.829***
(-8.919) (-8.176)
9.096*
(1.749)
-20.664
(-0.577)
-9.439*
(-1.716)
0.969
(0.185)
12.719***
(2.994)
7.342*
(1.713)
-1.363
(-0.282)
-3.369
(-0.799)
-9.297
(-1.635)
5.918
(1.010)
14.647*** 16.268***
(4.785)
(5.620)
-6.416*** -5.778***
(-8.790) (-8.086)
Q3-Q4/2008
(9)
-17.937
(-0.767)
18.503
(0.722)
19.325
(0.897)
14.200
(0.870)
-155.324
(-0.496)
-0.560
(-0.022)
8.806
(0.570)
15.206
(0.583)
-3.726
(-0.188)
31.746
(1.502)
-7.857***
(-3.759)
3953
3953
3570
3570
4022
4022
4022
4022
178
0.068
0.102
0.068
0.111
0.070
0.103
0.069
0.100
0.068
45
Table 8. Hedge Fund Behavior During Crises and their Future Returns
The table explores the future total returns of hedge funds with respect to their equity trades during crises and their
investment styles. Panel A explores the total returns of hedge funds in the following quarter. Panel B explores the
total returns of hedge funds two quarters ahead. The sample is based on TASS and 13(f), and consists of hedgefund-quarters from the third quarter of 1989 through the second quarter of 2009. In Columns (1) and (6), the market
condition variable is the Pastor-Stambaugh liquidity index. In Columns (2) and (7), it is the Acharya-Pedersen
liquidity index, and the VIX index in Columns (3) and (8). In Columns (4) and (9), the market condition variable is
the market’s excess returns. In Columns (5) and (10), the sample is restricted to the last two quarters of 2008.
Investment style variables (main effects) are included in Panel B. All regressions include a constant, which is not
presented. t-statistics are presented in parentheses. ***, **, and * represent statistical significance at the 1%, 5%,
and 10% levels, respectively. Robust standard errors are clustered by calendar quarter, except for Columns (5) and
(10).
Panel A: Future hedge fund returns and equity sales during crises
Mkt Cond < -2σ
Mkt Cond < -2σ x max(Δ Equity portfolio,0)
Mkt Cond < -2σ x min(Δ Equity portfolio,0)
max(Δ Equity portfolio,0)
min(Δ Equity portfolio,0)
Equity portfolio return (same period)
Fund flows
lead(Fund flows)
lead2(Fund flows)
Past ret 12
log(Portfolio size)
PS
(1)
0.376
(0.593)
0.024
(1.007)
-0.010
(-0.225)
-0.006
(-1.106)
0.015
(1.058)
0.418***
(15.761)
-0.003
(-1.520)
-0.004***
(-3.807)
0.001***
(4.739)
0.045**
(2.291)
-0.118
(-1.178)
Total portfolio ret(q+1) (%)
AP
VIX
Rm - Rf Q3-Q4/2008
(2)
(3)
(4)
(5)
0.433
0.241
-0.509
(0.391)
(0.309) (-1.122)
-0.001
-0.020
-0.021
(-0.023) (-0.762) (-0.855)
-0.055
0.070
0.131*
(-0.945) (1.247)
(1.903)
Total portfolio ret(q+2) (%)
PS
AP
VIX
Rm - Rf Q3-Q4/2008
(6)
(7)
(8)
(9)
(10)
-2.275** -1.708
0.507
-0.395
(-2.097) (-1.111) (0.684) (-0.473)
0.018
0.012
0.015
0.016
(0.557)
(0.234)
(0.786)
(1.540)
-0.110*** -0.149*** -0.026
0.023
(-4.521) (-8.591) (-0.858) (0.519)
-0.009*
-0.003
-0.004
(-1.844) (-0.550) (-0.594)
0.019
0.006
0.006
(1.440)
(0.408)
(0.472)
0.417*** 0.417*** 0.422***
(15.961) (16.434) (16.866)
-0.007** -0.002
-0.003
(-2.157) (-1.135) (-1.382)
-0.007 -0.005*** -0.004***
(-1.341) (-3.643) (-4.232)
0.016*** 0.001*** 0.001***
(4.580)
(4.788)
(4.302)
0.034*
0.043** 0.042**
(1.716)
(2.110)
(2.149)
-0.106
-0.113
-0.122
(-0.950) (-1.098) (-1.193)
0.005
0.005
0.007
0.007
(0.947)
(0.898)
(1.158)
(1.226)
0.031** 0.025**
0.016
0.012
(2.326)
(2.077)
(0.995)
(0.761)
0.414*** 0.413*** 0.419*** 0.418***
(16.500) (14.165) (16.640) (16.475)
-0.004* -0.007** -0.004*
-0.004
(-1.742) (-2.319) (-1.678) (-1.482)
0.001
0.002
0.000
-0.000
(1.113)
(0.749)
(0.325) (-0.161)
-0.000**
0.000
-0.000* -0.000**
(-2.135) (0.075) (-1.833) (-2.384)
0.007
0.002
0.011
0.009
(0.418)
(0.122)
(0.660)
(0.519)
-0.061
-0.028
-0.062
-0.061
(-0.615) (-0.263) (-0.617) (-0.616)
0.026
(0.268)
-0.127
(-1.398)
0.671***
(4.412)
-0.001
(-0.097)
-0.001
(-0.366)
-0.000
(-0.089)
0.450***
(4.097)
0.043
(0.043)
0.051
(0.781)
0.000
(0.000)
0.371***
(3.984)
0.005
(0.577)
-0.002
(-0.798)
0.001
(1.537)
0.071
(0.950)
-1.015
(-1.455)
Observations
3322
3101
3322
3322
67
3322
3101
3322
3322
67
Adj R2
0.418
0.410
0.418
0.420
0.443
0.409
0.383
0.406
0.405
0.236
46
Table 8. Returns of Hedge Funds that Exit the Market (Cont.)
Panel B: Future hedge fund returns and investment styles
Mkt Cond < -2σ x max(Δ Equity portfolio,0)
Mkt Cond < -2σ x min(Δ Equity portfolio,0)
Mkt Cond < -2σ x Convertible arb
Mkt Cond < -2σ x Short bias
Mkt Cond < -2σ x Emerging markets
Mkt Cond < -2σ x Market neutral
Mkt Cond < -2σ x Event driven
Mkt Cond < -2σ x Fixed income arb
Mkt Cond < -2σ x Fund of funds
Mkt Cond < -2σ x Global macro
Mkt Cond < -2σ x Long-short
Mkt Cond < -2σ x Futures
Mkt Cond < -2σ x Multi strategy
max(Δ Equity portfolio,0)
min(Δ Equity portfolio,0)
Equity portfolio return (same period)
Fund flows
lead(Fund flows)
lead2(Fund flows)
Past ret 12
log(Portfolio size)
Total portfolio ret(q+1) (%)
Total portfolio ret(q+2) (%)
PS
AP
VIX
R m - Rf
PS
AP
VIX
Rm - R f
(1)
(2)
(3)
(4)
(5)
(6)
(7)
(8)
0.031
0.022
-0.009
-0.029**
-0.007
0.009
-0.003
0.008
(1.083)
(0.483) (-0.338) (-2.051)
(-0.227) (0.204) (-0.234) (1.098)
0.004
-0.037
0.068
0.133*
-0.087*** -0.116*** -0.025
0.022
(0.084) (-0.502) (1.082)
(1.878)
(-3.220) (-4.458) (-0.893) (0.554)
-4.721
-5.129** -5.948
-2.314**
-1.629
-2.543
6.058*
1.129
(-1.425) (-2.499) (-1.536) (-2.409)
(-0.406) (-0.991) (1.709)
(1.062)
10.451
-6.504 -38.233***-35.961*** 19.346** 48.835*** 6.549
5.651
(1.009) (-0.738) (-4.624) (-4.414)
(2.034)
(5.691)
(0.713)
(0.616)
3.873
8.237** 12.523*
9.328*
-20.071***-11.658*** -2.163
-2.490
(1.126)
(2.310)
(1.808)
(1.783)
(-6.912) (-3.469) (-0.140) (-0.146)
2.595
0.029 13.112*** 66.015*** 6.253***
1.931
7.109
6.711***
(0.633)
(0.009)
(2.701) (29.456)
(2.757)
(0.593)
(1.566)
(4.729)
-0.430
-0.376
-0.417
0.237
-0.890
-0.352
1.264
-0.286
(-0.445) (-0.219) (-0.376) (0.544)
(-0.607) (-0.218) (1.295) (-0.519)
-18.058***-19.228*** -5.413
2.690
2.463
5.975*** 4.050**
2.741
(-9.860) (-18.147) (-0.990) (1.388)
(0.832)
(4.768)
(2.247)
(1.631)
4.691***
2.390
8.096***
1.148
-2.107
0.183
1.389
0.814
(2.731)
(0.920)
(4.350)
(0.998)
(-1.427) (0.202)
(1.511)
(0.637)
7.229** 4.004*** 17.018***
0.998
-3.121
-1.606
(2.071)
(4.201)
(5.673)
(0.348) (-0.752) (-0.382)
0.544
1.187
-0.264
-0.834
-3.199** -2.170
-0.171
-0.701
(0.515)
(0.676) (-0.198) (-1.408)
(-2.637) (-1.178) (-0.200) (-0.794)
5.864*
6.207
-9.144
6.578* 9.970*** -3.722
(1.963)
(1.480) (-1.394)
(1.796)
(3.785) (-0.654)
-2.408
-1.269* -4.080*
-0.546
-2.426* -3.575*** 0.384
0.848
(-1.192)
-0.007
(-1.407)
0.017
(1.400)
0.422***
(15.412)
-0.003
(-1.377)
-0.004***
(-3.537)
0.001***
(3.340)
0.041**
(2.147)
-0.165
(-1.664)
(-1.969) (-1.845) (-0.803)
-0.010** -0.004
-0.004
(-2.293) (-0.751) (-0.666)
0.023*
0.010
0.009
(1.906)
(0.702)
(0.721)
0.419*** 0.424*** 0.421***
(16.108) (16.681) (16.817)
-0.007** -0.002
-0.003
(-2.058) (-0.817) (-1.273)
-0.006 -0.006*** -0.004***
(-1.297) (-3.111) (-4.171)
0.016*** 0.001*** 0.001***
(4.943)
(2.974)
(4.318)
0.030
0.039*
0.038*
(1.532)
(1.958)
(1.953)
-0.134
-0.180* -0.179*
(-1.225) (-1.835) (-1.774)
(-1.712) (-4.845) (0.293)
(0.407)
0.008
0.006
0.009
0.009
(1.432)
(1.106)
(1.635)
(1.642)
0.031** 0.028**
0.017
0.013
(2.545)
(2.350)
(1.079)
(0.846)
0.414*** 0.414*** 0.419*** 0.417***
(16.527) (14.204) (16.689) (16.486)
-0.003
-0.006** -0.003
-0.003
(-1.284) (-2.020) (-1.246) (-1.201)
0.001
0.003
-0.000
-0.000
(1.440)
(0.875) (-0.165) (-0.216)
-0.001**
0.000
-0.000
-0.000**
(-2.309) (0.161) (-0.681) (-2.261)
0.000
-0.004
0.005
0.004
(0.019) (-0.241) (0.306)
(0.217)
-0.096
-0.069
-0.092
-0.097
(-0.922) (-0.636) (-0.841) (-0.907)
Observations
3322
3101
3322
3322
3322
3101
3322
3322
2
0.426
0.415
0.430
0.427
0.419
0.390
0.409
0.408
Adj R
47