Printer Friendly

Emerging market exposures and the predictability of hedge fund returns.

We examine emerging market and global macro hedge funds and find a significant positive relation between hedge funds' future returns and their exposure to both emerging market equities and emerging market currencies. We present evidence that the strong predictive power of emerging market betas is related to the superior market-timing ability of these fund managers. Results are robust after controlling for commonly used hedge fund factors, the emerging market equity index, lagged fund returns, liquidity risk, and fund characteristics. Our results suggest that hedge funds can earn positive excess returns by timing their exposure to emerging market securities.

**********

It is widely accepted that macroeconomic variables have a strong impact on the prices of risky assets such as stocks, bonds, currencies, and their derivatives. Therefore, it is reasonable to assume that the performance of hedge funds that invest in these risky assets would also be affected by the funds' exposures to these macroeconomic variables. In fact, Asness, Krail, and Liew (2001) illustrate that hedge fund returns are indeed exposed to market (MKT) factors, and Bali, Brown, and Caglayan (2011) show that hedge funds' exposures to default premium and inflation are positively and negatively related to their future fund returns, respectively. In this paper, we examine the relation between emerging market and global macro hedge funds' exposure to emerging market securities and their future performance.

Shortly after the 1997 Asian currency crisis and the 1998 Russian debt crisis, many emerging market economies underwent significant reforms, including the adoption of new flexible exchange rate systems, which led to a period of rapid economic growth. As a result, these new markets attracted the attention of many investors, including hedge funds, who found themselves investing more and more in emerging market securities. In particular, the two hedge fund investment styles, emerging market and global macro, experienced a dramatic increase both in numbers and in assets under management (AUM). (1) Given this tremendous influx of new investment flows to emerging markets in the last decade and a half, this paper explores hedge fund performance in emerging markets and, more importantly, the drivers of that performance. In particular, we investigate emerging market and global macro hedge funds' exposure (betas) to six different emerging market assets and check the predictive power of these betas in explaining the cross-sectional differences in expected future hedge fund returns.

By using parametric Fama and MacBeth (1973) cross-sectional regressions and nonparametric portfolio tests, we find a significant positive relation between hedge funds' future returns and their exposure to both emerging market equities (MSCI beta) and emerging market currencies (EMFX beta). The results are robust after controlling for fund characteristics, lagged fund returns, liquidity risk, the performance of the emerging market equity index, and commonly used hedge fund factors.

In addition, we examine a larger set of hedge funds and divide the fund styles into three broad categories: 1) directional (which includes the emerging market and global marco funds), 2) semi-directional, and 3) nondirectional. We find that emerging market betas explain the cross-sectional differences in hedge fund returns for the directional strategy group only, but not for the other investment strategy groups. We also discover that the variation of emerging market betas through time is much wider for the directional strategies as compared to the other two strategies.

This larger variation and the stronger predictive power of emerging market betas in directional strategies may possibly be explained by the market-timing ability of directional strategy hedge fund managers. To test this, we follow Henriksson and Merton (1981) and measure market-timing ability as the relation between a fund's excess returns and positive moves in the underlying market portfolio. Looking at all three broad categories of hedge funds, and using different proxies of market portfolios, we find that the market-timing measure is positive and significant only in the directional strategy group. This suggests that emerging market and global macro hedge fund managers are able to actively adjust their exposure to emerging market securities in response to macroeconomic conditions and the state of financial markets so as to generate a strong link between their returns and their exposure to emerging market securities. Hence, we conclude that the stronger predictive power of emerging market betas in directional strategies is related to the superior market-timing ability of these fund managers.

Our work is related to a number of studies examining the role of risk exposure in driving hedge fund performance. Agarwal and Naik (2004) discuss the systematic risk exposure of hedge funds using buy-and-hold and option-based risk factors, and note that hedge funds bear significant left-tail risk that is usually ignored by the traditional mean-variance framework. Bali, Gokcan, and Liang (2007) find evidence of a positive and significant link between downside risk measure and future expected hedge fund returns. Brown, Gregoriou, and Pascalau (2012) determine that hedge funds' tail risk exposure may not be diversifiable and indicate that tail risk could explain some portion of future hedge fund returns. Bali, Brown, and Caglayan (2012) introduce a new composite measure of systematic risk for individual hedge funds by dividing total risk into two components: 1) systematic risk and 2) residual (fund-specific) risk. They find a positive and significant link between the composite measure of systematic risk and the cross-section of hedge fund returns. (2) Our paper investigates hedge funds' exposures to emerging market-specific risk factors and checks the predictive power of these exposures in explaining the future returns of hedge funds.

The closest paper to our work is Bali et al. (2011), which examines hedge funds' exposure to major market risk factors through alternative measures of factor betas, and finds that hedge funds that have a greater (lower) exposure to default spread (inflation) generate statistically and economically higher returns in the following month. In this paper, our focus is specifically on emerging markets, as opposed to the major market risk factors tested in Bali et al. (2011). We examine whether hedge funds are able to generate superior returns for their investors by increasing their exposure to emerging market securities and, if they do, how and why. This is a research area that has not yet been explored in the hedge fund literature. We believe that our results will help investors and hedge fund managers tremendously in determining which asset classes in emerging markets are worth considering to produce superior fund returns.

This paper is organized as follows. Section I describes the data and variables used in our empirical analyses. Section II discusses the potential data bias issues in hedge fund studies. Section III presents the main empirical results, while Section IV provides a battery of robustness checks. Section V examines the predictive power of emerging market betas for three broad hedge fund investment strategies and conducts market-timing tests for these strategies separately. Section VI provides our conclusions.

I. Data and Description of Variables

This study uses hedge fund data from the Lipper Trading Advisor Selection System (TASS) database. For our specific analysis on emerging markets, we use information on hedge funds classified as either emerging market or global macro funds in the TASS database. (3) Our emerging market and global macro hedge fund database contains information on a total of 1,453 defunct and live hedge funds with total AUM assets under management as of December 2011 around $90 billion. Of the 1,453 hedge funds that reported monthly returns to TASS during the period January 1999-to-March 2012, we note 760 funds in the defunct/graveyard database and 693 funds in the live hedge fund database. The TASS database, in addition to reporting monthly returns (net of fees) and monthly AUM assets under management, also provides information regarding certain fund characteristics including management fees, incentive fees, redemption periods, minimum investment amounts, and lockup and leverage provisions.

Table I provides summary statistics of the hedge funds' numbers, returns, and assets under management AUM. For each year, Table I reports the number of hedge funds entered into the database, the number of hedge funds dissolved, the total AUM at the end of each year (in billion $), and the mean, median, standard deviation, minimum, and maximum monthly percentage returns on the equal-weighted hedge fund portfolio. We select 1999 as the beginning of our sample period, as emerging markets became an attractive destination for hedge fund investments, especially after the liberalization of these markets following the 1997 Asian currency crisis and the 1998 Russian debt and Long Term Capital Management (LTCM) crises. Our database did not include a sufficient number of emerging market and global macro hedge funds to run (cross-sectional) statistical analyses prior to 1999, largely owing to a lack of interest in emerging markets due to short-sell and other structural constraints that hedge funds faced in investing in these markets prior to this date. Therefore, we select 1999 as the start of our sample and run our analyses on hedge funds during the period January 1999 to March 2012.

Table I reports a sharp reversal in the growth of hedge funds both in number and in AUM since the end of 2007, the beginning of the subprime mortgage financial crisis. From 1999 to 2007, the AUM in our database increased exponentially from $22.2 billion to an eye-opening $185.9 billion, and the number of emerging market and global macro hedge funds performing in the market more than tripled to 785 from 251. However, both of these figures reversed course beginning in 2008. The number of hedge funds performing in the market fell below 450, while the total AUM dropped by more than half to $90.3 billion by the end of December 2011. These two sharp reversals in the data simply explain the severity of the financial crisis that the hedge fund industry went through over the past few years. For instance, in 2008 and 2011, emerging market and global macro hedge funds, on average, lost 2.27% and 0.94% (return) per month, respectively.

In this study, we also utilize six different emerging market financial variables/indicators to measure hedge funds' exposure to emerging markets. These include: 1) the MSCI Emerging Market Equity Index (MSCI), 2) the JPMorgan Emerging Market Bond Index Plus (EMBI+), 3) the JPMorgan Emerging Market Volatility Index (JPEMVOL), 4) the S&P Goldman Sachs Commodity Index (SPGSCI), 5) the S&P Goldman Sachs Precious Metal Index (SPGSPM), and 6) the Emerging Market Currency basket index (EMFX). (4) Each of these six measures is used as a representative/benchmark of a specific asset class in emerging markets. MSCI is the Emerging Markets Equity Total Return Index hedged to US dollar (USD), calculated with the net withholding dividend approach. It is a free-float-adjusted market capitalization index computed in USDs. EMBI+ is a USD denominated benchmark index for emerging markets debt securities, tracking total returns for actively traded external debt instruments in emerging markets. It includes USD denominated Brady bonds, Eurobonds, and traded loans issued by sovereign entities. JPEMVOL is a tradable index that tracks the total returns from the three-month at-the-money forward implied volatility of emerging market FX options. It is used to express views on volatility (in emerging markets) as an asset class, and is regarded as the benchmark for emerging market FX implied volatility. SPGSCI is a tradable USD denominated index for commodity markets, tracking total returns from unleveraged, long-only investments in commodity futures. It is a broadly diversified index containing 24 commodities across the spectrum of all commodity sectors. It is widely recognized as the leading measure for general commodity price movement and inflation in the world economy. SPGSPM is a tradable sub-index of SPGSCI, denominated in USDs. It provides investors a reliable and publicly available benchmark for investment performance in precious metals in commodity markets. Finally, EMFX tracks the performance of an equally weighted emerging market currency basket against the USD, where the basket is composed of seven major emerging market currencies including Brazilian Real (BRL), Mexican Peso (MXN), Turkish Lira (TRY), Polish Zloty (PLN), Hungarian Florin (HUF), South African Rand (ZAR), and Indonesian Rupiah (IDR). For each currency in the basket, the daily moves in the spot exchange rate (the local currency against the USD), as well as the local economy's interest rate difference over the US (derived from the forward points) are included in the computation of the basket performance.

II. Potential Hedge Fund Data Biases

Hedge fund studies are subject to potential data biases. Fung and Hsieh (2000), Liang (2000), and Edwards and Caglayan (2001) cover these well-known data biases extensively in the hedge fund literature. The most common and easily fixable data bias in a hedge fund study is the survivorship bias. Survivorship bias exists if the database does not include the returns of non-surviving (liquidated) hedge funds, causing reported hedge fund performance in that database to appear higher/better than the actual hedge fund performance. In our study, for the sample period from January 1999 to March 2012, we have monthly return histories of 693 funds in the live funds (survivor) database and 760 funds in the graveyard (defunct) database. However, funds in the graveyard database are not necessarily all liquidated funds. As a hedge fund industry practice, when funds stop reporting (arbitrarily) to a specific database vendor, those funds are moved by that database vendor to their graveyard database from their live funds database. Fung and Hsieh (2009) emphasize the importance of differentiating between missing funds and liquidated funds and conclude that only liquidated funds, rather than all funds in the graveyard database, should be considered in survivorship bias estimations.

TASS provides information as to why a fund is dropped from the live database and moved to the graveyard database. Fund liquidation, fund no longer reporting, fund closed to new investments, and fund merged into another entity are some of the possible reasons for being dropped from the live database. Using this information, we find that only 429 of the 760 funds in our graveyard database (56%) are confirmed liquidated funds. This low ratio demonstrates how frequently hedge funds are moved to the graveyard database for reasons unrelated to liquidation. This suggests, as Fung and Hsieh (2009) also note, that survivorship bias estimates should not be calculated based on all of the funds in the graveyard database (as is the case in previous literature), but based on liquidated funds only. Calculating the magnitude of the survivorship bias as the difference between the annualized average return of the combination of live and nonliquidated graveyard funds in the sample and the annualized average return of all (live, liquidated, and nonliquidated) hedge funds in the sample, we find that there would have been a 2.46% upward bias in average annual hedge fund returns if the returns of the liquidated funds had not been included in our analyses.

Another important data bias in a hedge fund study is called the back-fill bias. Once a hedge fund is included in a database, that fund's previous returns are automatically added to that database. This practice suggests only successful hedge funds will report their past returns to the database vendor. As a result, this may generate an upward bias in the returns of newly reporting hedge funds during their early histories. The TASS database provides information as to when a hedge fund was added to the database, as well as the fund's inception date. Aggarwal and Jorion (2010) measure the back-fill period as the difference between a fund's inception date and the date the fund is added to the database. They identify a fund as "non-back-filled" if the back-fill period (the period between the inception date and date added to database) is below 180 days. In other words, they divide the hedge fund sample in two and hedge funds whose inception date and database entry date are in proximity are classified as non-back-filled funds. The rest of funds in the sample (whose back-fill periods are more than 180 days) are classified as back-filled funds. Then, they calculate the average annual return difference between back-filled funds and non-backfilled funds to measure the back-fill bias. Following Aggarwal and Jorion's (2010) procedure, we identify 732 hedge funds as back-filled funds in our sample and estimate a back-fill bias of 2.04% in our study for the sample period January 1999 to March 2012. (5)

The last possible data bias in a hedge fund study is called the multi-period sampling bias. Investors generally ask for a minimum of 36 months of return history before making a decision whether to invest in a hedge fund or not. Therefore, in a hedge fund study, inclusion of hedge funds with return histories shorter than 36 months would be misleading to those investors who seek past performance data to make investment decisions. Also, a minimum 36-month return history requirement makes sense from a statistical perspective to be able to run regressions and get sensible regression estimates, such as betas, for each individual hedge fund in the sample. As such, we require that all hedge funds in the sample have at least 36 months of return history in our study. That is, we drop 290 hedge funds that have return histories shorter than 36 months from our analyses, decreasing our sample size from 1,453 to 1,163 funds. However, this might introduce a new survivorship bias into our analyses as the 290 deleted hedge funds that had return histories less than 36 months were more than likely dissolved due to poor performance. In an effort to find the impact of 290 deleted hedge funds on total hedge fund performance, we compare the performance of hedge funds before and after the 36-month return history requirement. We find that the annual average return of hedge funds that pass the 36-month requirement (1,163 funds) is only 0.57% higher than the return of all hedge funds (1,453 funds) in the sample, a small insignificant percentage difference in annual terms between the two samples in terms of survivorship bias considerations. (6)

III. Empirical Results

A. Cross-Sectional Regressions of Future Fund Returns on Emerging Market Betas

The primary objective of this paper is to determine the capability of emerging market betas to explain the cross-sectional variation in the monthly returns of emerging market and global macro hedge funds. We conduct two separate tests. The first is a parametric regression analysis, while the second is a nonparametric portfolio analysis. In this section, we discuss the parametric tests and initially derive, for each individual hedge fund in our sample, the monthly time-series estimates of betas for six different emerging market financial risk factors estimated over a rolling 36-month window period. In the second stage, for each month in the sample period, we conduct Fama and MacBeth (1973) cross-sectional regressions of one-month-ahead individual hedge fund excess returns (over the risk-free rate) on the emerging market factor betas generated from the first stage. The statistical analysis of the average slope coefficients from the Fama and MacBeth (1973) regressions indicate whether or not the six emerging market betas tested have any significant predictive power for future hedge fund returns.

1. Univariate Emerging Market Betas in Cross-Section Regressions

Table II reports the time-series average intercept and slope coefficients from the Fama and MacBeth (1973) cross-sectional regressions of one-month-ahead hedge fund excess returns on the univariate emerging market betas. In the first stage, univariate monthly emerging market betas are generated for each fund from the time-series regressions of hedge fund excess returns on the emerging market factor over the risk-free rate over a 36-month rolling window period. In the second stage, the cross-section of one-month-ahead funds' excess returns are regressed on the funds' univariate emerging market betas each month during the period January 1999 to March 2012. Put simply, we begin with the first three years of monthly returns from January 1996 to December 1998 to estimate the emerging market betas for each fund in our sample. We then follow a monthly rolling regression approach with a fixed estimation window of 36 months to generate the monthly time-series emerging market beta estimates of individual funds based on the following regression equation:

[R.sub.i,t] = [[alpha].sub.i,t] + [[beta].sup.k.sub.i,t] x [K.sub.t] + [[epsilon].sub.i,t] (1)

where [R.sub.i,t] is the excess return (return over the risk-free rate) on fund i in month t and [K.sub.t] is the emerging market financial risk factory's return (over the risk-free rate) in month t. [[alpha].sub.i,t] and [[beta].sup.k.sub.i,t] are, respectively, the alpha and the emerging market financial risk factor K's beta for fund i in month t. Note that the emerging market financial risk factor K in Equation (1) represents one of the six variables examined in this study: 1) MSCI, 2) EMBI+, 3) JPEMVOL, 4) SPGSCI, 5) SPGSPM, and 6) EMFX. Therefore, Equation (1) is a set of six regression equations where each regression equation is run for each emerging market financial risk factor separately.

Then, in the second stage, beginning in January 1999, we use the Fama and MacBeth (1973) cross-sectional regressions of one-month-ahead individual fund excess returns on the current-month emerging market factor betas:

[R.sub.i,t+1] = [[tau].sub.t] + [[theta]sub.t] x [[beta].sup.K.sub.i,t] + [[epsilon].sub.i,t+1], (2)

where [R.sub.i,t+1] is the excess return on fund i in month t + 1 and [[beta].sup.K.sub.i,t] is the emerging market financial risk factor K's beta for fund i in month t estimated using Equation (1). [[tau].sub.t] and [[theta]sub.t] are, respectively, the monthly intercepts and slope coefficients from the Fama and MacBeth (1973) regressions. As in Equation (1), Equation (2) is also a set of six regression equations where each regression equation is run separately for each emerging market financial risk factor beta. Table II presents the time-series average intercept and slope coefficients from Equation (2) over the sample period January 1999 to March 2012, using as the independent variable, the univariate emerging market betas estimated from Equation (1). The corresponding Newey and West (1987) t-statistics are reported in parentheses. Our results indicate a positive and significant link between the expected returns on hedge funds and the MSCI beta ([[beta].sub.MSCI]), EMBI+ beta ([[beta].sub.EMBI+]), and EMFX beta ([[beta].sub.EMFX]). We find the average slope coefficient from the monthly regressions of one-month-ahead hedge fund excess returns on the previous month's MSCI beta to be 1.439 with a statistically significant Newey-West (1987) t-statistic of 2.12. Similarly, the average slope coefficients from the monthly regression of one-month-ahead hedge fund excess returns on the previous month's EMBI+ beta and EMFX beta (separately) are 0.961 and 0.745, respectively, with statistically significant Newey-West (1987) t-statistics of 2.40 and 2.06. (7) On the other hand, the other three emerging market financial risk factor betas tested (JPEMVOL, SPGSCI, and SPGSPM) do not seem to have any predictive power on one-month-ahead hedge fund excess returns (see the diagonal in Table II).

2. Multivariate Emerging Market Betas in Cross-Section Regressions

In this section, we drop the insignificant emerging market risk factors from our analyses and focus only on the MSCI, EMBI+, and EMFX betas. We analyze the interaction between these emerging market betas themselves, as well as their interaction with certain hedge fund characteristics.

Table III reports the time-series average intercept and slope coefficients from the Fama and MacBeth (1973) cross-sectional regressions of one-month-ahead fund excess returns on the multivariate emerging market betas. In the first stage, we run the following regression with a fixed rolling estimation window of 36 months to generate the monthly time-series of multivariate emerging market betas:

[R.sub.i,t] = [[alpha].sub.i,t] + [[beta].sup.MSCI.sub.i,t] x + [[beta].sup.EMBI+.sub.i,t] x EMBI [+.sub.t] + [[beta].sup.EMFX.sub.i,t] x [EMFX.sub.t] + [[epsilon].sub.i,t], (3)

where [R.sub.i,t] is the excess return on fund i in month t; [MSCI.sub.t], EMBI [+.sub.t], [EMFX.sub.t] are, respectively, the emerging market equity index, the emerging market bond index, and the equally weighted emerging market currency basket index against the USD in month t (all three emerging market index returns are expressed in USD and in excess of the risk-free rate). The coefficients [[alpha].sub.i,t], [[beta].sup.MSCI.sub.i,t], [[beta].sup.EMBI+.sub.i,t] and [[beta].sup.EMPX.sub.i,t] are the intercept, the emerging market equity index beta, the emerging market bond index beta, and the emerging market currency basket beta for fund i in month t, respectively.

In the second stage, monthly cross-sectional regressions are run for the following multivariate full specification and its shorter versions:

[MATHEMATICAL EXPRESSION NOT REPRODUCIBLE IN ASCII] (4)

where [R.sub.i,t+1] is the excess return on fund i in month t + 1 and [[beta].sup.MSCI.sb.i,t], [[beta].sup.EMBI+.sub.i,t], and [[beta].sup.EMFX.sub.i,t] are, respectively, the emerging market equity index beta, the emerging market bond index beta, and the emerging market currency basket beta in month / estimated from Equation (3) over the past three years. SIZE, AGE, MANAGEMENTFEE, INCENTIVEFEE, REDEMPTION, MININVEST, DJLOCKUP, and D_LEVERAGE are the fund characteristics. SIZE is measured as the monthly AUM in billions of dollars. AGE is measured as the number of months in existence since the inception of a fund. MANAGEMENTFEE is a fixed percentage fee on AUM, typically ranging from 1% to 2%. INCENTIVEFEE is a fixed percentage fee of the fund's annual net profits above a designated hurdle rate. REDEMPTION is the minimum number of days an investor needs to notify a hedge fund before he/she can redeem the invested amount from the fund. MININVEST is the minimum initial required amount to invest in a hedge fund (measured in millions of dollars). DJLOCKUP is the dummy variable for lockup provisions (one if the fund requires that the investors are unable to withdraw their initial investments for a pre-specified term, usually 12 months, and zero otherwise). D_LEVERAGE is the dummy variable for leverage (one if the fund uses leverage, and zero otherwise). We also include the lagged fund returns ([R.sub.i,t]) to control for potential momentum (MOM) or reversal effects in hedge fund returns.

As demonstrated in Table III, controlling for the other emerging market betas in question ([[beta].sup.MSCI], [[beta].sup.EMBI+], and [[beta].sup.EMFX]), individual hedge fund characteristics and the lagged fund returns do not alter or weaken the statistically significant predictive power of MSCI beta and EMFX beta over future hedge fund returns. There still exists a positive and significant relation between future hedge fund returns and MSCI beta and EMFX beta, whether all of the variables are controlled for simultaneously or in different combinations of groupings. In Table III, the average slope coefficient on MSCI beta is estimated to be between 1.262 and 1.330 (from alternative regression specifications), with the statistically significant Newey-West (1987) t-statistics ranging from 1.98 to 2.18. Similarly, the average slope coefficient on EMFX beta is estimated to be between 0.430 and 0.636, with the statistically significant Newey-West (1987) t-statistics hovering around 2.00 and above. Interestingly, however, our third statistically significant emerging market beta from the univariate regressions, EMBI+ beta, while continues to predict future hedge fund returns after controlling for the effect of EMFX beta, it loses its predictive power over future hedge fund returns when it is combined with the MSCI beta in the same regression equation. That is, EMBI+ beta cannot continue to predict future hedge fund returns strongly when we control for the effect of MSCI beta in the same regression. In sum, we can conclude that both the univariate and multivariate Fama and MacBeth (1973) cross-sectional regressions provide solid evidence of an economically and statistically significant positive link between future hedge fund returns and MSCI beta and EMFX beta. To be exact, hedge funds that increase their exposure, particularly to emerging market equities and emerging market currencies, generate superior returns in the following months. However, the same cannot be said as strongly for EMBI+ beta as its predictive power over future hedge fund returns diminishes in the spotlight of the MSCI beta.

One interesting observation in Table III is the fact that the average slope on lagged fund returns is positive and highly significant when it is included in the regression equation with other hedge fund characteristics. Specifically, the average slope on [R.sub.i,t] is 0.073 with a Newey-West (1987) t-statistic of 4.58. This result indicates strong MOM effects in individual fund returns. That is, winner (loser) funds continue to be winners (losers) in a one month investment horizon. (8)

Another noteworthy point in Table III is that the management fee variable has a positive and statistically significant coefficient (0.205, with a t-statistic of 3.51) in monthly cross-sectional regressions of one-month-ahead funds' excess returns on factor betas when hedge fund characteristics are added to the cross-sectional regression equation as well. This suggests that management fees have strong positive explanatory power over expected hedge fund returns (i.e., funds that charge higher management fees also generate higher future hedge fund returns). Moreover, among other hedge fund characteristics, the minimum investment variable also has a positive and statistically significant coefficient (0.061, with a t-statistic of 2.38) in the multivariate cross section regression. This suggests that hedge funds that require higher initial minimum investment amounts generate higher future hedge fund returns. (9) Finally, another fund characteristic, age, has a negative and statistically significant relation with expected hedge fund returns (a coefficient of -0.003, with a Newey-West (1987) t-statistic of -2.57). Most importantly, however, the existence of statistically significant fund characteristics in the multivariate Fama and MacBeth (1973) cross-sectional regressions does not alter or weaken the predictive power of MSCI beta and EMFX beta over future hedge fund returns.

B. Univariate Quintile Portfolio Analyses of MSCI, EMBI+, and EMFX Betas

In this section, we analyze the relationship between funds' emerging market betas and future hedge fund returns via our second test, the nonparametric portfolio analyses. In this test, each month, we generate quintile portfolios of hedge funds by sorting hedge funds according to their emerging market betas and then we observe each quintile's next month performance to see if there exists a significant difference in the returns of high factor beta quintiles versus low factor beta quintiles.

1. Univariate Portfolio Analysis of MSCI Betas

We conduct our nonparametric quintile portfolio test for the MSCI betas by sorting hedge funds according to their past month's MSCI betas and comparing the performance of the high MSCI beta quintile to the low MSCI beta quintile in the following month. We generate quintile portfolios every month from January 1999 to March 2012 by sorting hedge funds based on their MSCI betas derived from Equation (1), where Quintile 1 contains the hedge funds with the lowest [[beta].sup.MSCI] and Quintiie 5 contains the hedge funds with the highest [[beta].sup.MSCI]. The left panel of Table IV reports the average [[beta].sup.MSCI] and the average next month returns for each of the five MSCI beta sorted quintiles. Moving from Quintile 1 to Quintile 5, we note that the average returns on the MSCI beta portfolios increase monotonically from 0.319% to 1.794% per month. This indicates an average return difference of 1.475% per month (17.7% per year) between Quintiles 5 and 1 (i.e., high [[beta].sup.MSC1] vs. low [[beta].sup.MSCI]) with a Newey-West (1987) t-statistic of 2.00, suggesting that this positive return difference is statistically and economically significant.

Next, we determine whether this significant return difference between high MSCI beta funds and low MSCI beta funds can be explained by commonly used hedge fund factors such as Fama and French's (1993) and Carhart's (1997) four factors of market (MKT), size (SMB), book-to-market (HML), and MOM, as well as Fung and Flsieh's (2004) two bond factors ([DELTA]10[Y.sub.t], [DELTA]CredSpr), and Fung and Hsieh's (2001) three trend-following factors on currencies, bonds, and commodities (FXTF, BDTF, and CMTF). (10) Basically, we regress the monthly time-series of return differences between high MSCI beta and low MSCI beta funds on Fama, French, and Carhart's four factors (MKT, SMB, FIML, MOM), as well as on Fama, French, Carhart, and Fung and Hsieh's combined nine factors (MKT, SMB, HML, MOM, [DELTA]10[Y.sub.t], [DELTA]CredSpr, BDTF, FXTF, CMTF), and we check if the intercepts from these two regressions (i.e., the four-factor alpha and the nine-factor alpha) are statistically significant. The last two rows of Table IV (left panel) report the four-factor and nine-factor alphas from these regressions. The four-factor alpha difference between Quintiles 5 and 1 is 1.207% with a t-statistic of 2.74. Additionally, the nine-factor alpha difference between Quintiles 5 and 1 is 1.270% with a t-statistic of 2.94. This suggests that after controlling for the market, size, book-to-market, MOM, change in treasury yields, change in corporate yields over Treasuries, and trend-following factors, the return difference between high MSCI beta and low MSCI beta funds remains positive and significant. Alternatively, the four and nine factors tested here do not explain the positive relation between [[beta].sup.MSCI] and the cross-section of future hedge fund returns. All in all, these results strengthen our earlier findings of the existence of a positive and significant link between MSCI betas and future hedge fund returns from Fama and MacBeth (1973) cross-sectional regressions.

2. Univariate Portfolio Analysis of EMBI+ Betas

We perform the same nonparametric quintile portfolio analysis this time for EMBI+ betas by sorting hedge funds according to their past month's EMBI+ betas and comparing the performance of the high EMBI+ beta quintile to the low EMBI+ beta quintile in the following month. The middle panel of Table IV presents the average [[beta].sup.EMBI+], as well as the average next month returns for each of the five EMBI+ beta sorted quintiles. As we move from Quintile 1 to Quintile 5, the average returns on the EMBI+ beta portfolios increase monotonically from 0.383% to 1.877% per month. That is, the average return difference between Quintiles 5 and 1 (i.e., high [[beta].sup.EMBI+] vs. low [[beta].sup.EMBI+]) is found to be 1.494% per month (17.9% per year) with a t-statistic of 2.17, suggesting that this positive return difference is statistically and economically significant.

Next, we determine whether the significant positive return difference between high EMBI+ beta and low EMBI+ beta funds can be explained by the commonly used hedge fund factors. We find the four-factor alpha difference between Quintiles 5 and 1 to be 1.201% with a t-statistic of 2.95. Likewise, the nine-factor alpha difference between Quintiles 5 and 1 is 1.246% with a statistically significant t-statistic of 3.20. This indicates that the risk-adjusted return differences between high EMBI+ and low EMBI+ beta funds are also positive and significant, after taking into account the commonly used hedge fund factors.

3. Univariate Portfolio Analysis of EMFX Betas

Finally, we conduct the same nonparametric quintile portfolio analysis for EMFX betas and find a similar positive and significant relation between EMFX betas and future fund returns as in the case for MSCI betas and EMBI+ betas. The right panel of Table IV presents the average [[beta].sup.EMFX] and the average next month returns for each of the five EMFX beta sorted quintiles. Moving from Quintile 1 to Quintile 5, the average returns on the EMFX beta portfolios increase monotonically from 0.369% to 1.832% per month, implying an average return difference between Quintiles 5 and 1 (i.e., high [[beta].sup.EMFX] vs. low [[beta].sup.EMFX]) of 1.463% per month (17.6% per year) with a t-statistic of 2.16, suggesting that this positive return difference is statistically and economically significant.

Finally, we verify whether the significant positive return difference between high EMFX beta and low EMFX beta funds can be explained by commonly used hedge fund factors. The four-factor and nine-factor alpha differences between Quintiles 5 and 1 are 1.200% and 1.253%, respectively, with statistically significant respective t-statistics of 3.03 and 3.24. This implies that after controlling for the commonly used factors, the risk-adjusted return differences between high EMFX and low EMFX beta funds remain positive and significant. Alternatively, these four and nine factors do not explain the positive relation between [[beta].sup.EMFX] and the cross-section of hedge fund returns. In sum, these results reinforce our earlier findings for the existence of a positive and significant link between EMFX betas and future hedge fund returns from univariate Fama and MacBeth (1973) regressions.

C. Bivariate Quintile Portfolio Analyses of MSCI, EMBI+, and EMFX Betas

In this section, we continue with our portfolio analysis, but this time, we take into account the interaction between MSCI betas, EMBI+ betas, and EMFX betas. First, we perform bivariate quintile portfolio tests for MSCI betas by controlling for the effect of EMFX betas and EMBI+ betas. Then, we carry out the same test for EMFX betas and EMBI+ betas separately by controlling for the other two emerging market betas in question.

1. Bivariate Portfolios of MSCI Betas After Controlling for EMFX and EMBI+ Betas

We begin our analysis by examining the positive relationship between MSCI betas and future hedge fund returns after controlling for the effect of EMFX betas. To conduct this test, we form quintile portfolios each month from January 1999 to March 2012 by sorting hedge funds into five quintiles based on their 36-month [[beta].sup.EMFX.sub.i,t]. Then, within each [[beta].sup.EMFX.sub.i,t] sorted portfolio, we sort hedge funds further into five sub-quintiles based on their 36-month [[beta].sup.MSCI.sub.i,t]. This produces sub-quintile portfolios of hedge funds with dispersion in MSCI betas with nearly identical EMFX betas, enabling us to control for differences in the EMFX betas among the newly generated MSCI beta sub-quintiles. In sum, this methodology generates 25 sub-quintile portfolios, where [Q.sub.i,j is the jth ranked MSCI beta portfolio within jth ranked EMFX beta portfolio (i = 1,2, ..., 5; j = 1, 2, ..., 5). In Panel A of Table V, "Quintile MSCI,1" represents the lowest MSCI beta ranked hedge fund sub-quintiles within each of the five EMFX beta ranked quintiles. That is, "Quintile MSCI,1" is the average of the following five sub-quintile portfolios: [Q.sub.1,1], [Q.sub.2,1], [Q.sub.3,1], [Q.sub.4,1], and [Q.sub.5,1] j. Similarly, "Quintile MSCI,5" represents the highest MSCI beta ranked hedge fund sub-quintiles within each of the five EMFX beta ranked quintiles and is the average of the following five sub-quintile portfolios: [Q.sub.1,5], [Q.sub.2,5], [Q.sub.3,5], [Q.sub.4,5], and [Q.sub.5,5]. Table V, Panel A presents the next month returns for the quintiles MSCI,1; MSCI,2; MSCI,3; MSCI,4; and MSCI,5. Moving from Quintile MSCI,1 to Quintile MSCI,5, the average returns on the MSCI beta portfolios increase monotonically from 0.414% to 1.665% per month indicating an average return difference between Quintiles MSCI,5 and MSCI,1 (i.e., high [[beta].sup.MSCI] vs. low [[beta].sup.MSCI]) of 1.250% per month with a Newey-West (1987) t-statistic of 2.06. This suggests that the positive relation between MSCI betas and future hedge fund returns remains significant after controlling for EMFX betas, a result that is very similar to the results obtained from our earlier multivariate Fama and MacBeth (1973) regressions.

We also determine whether this significant return difference between Quintile MSCI,5 and Quintile MSCI,1 can be explained by Fama and French's (1993) and Carhart's (1997) four factors of market, size, book-to-market, and MOM, as well as Fung and Hsieh's (2001, 2004) two bond and three trend-following factors on currencies, bonds, and commodities. The four-factor alpha difference between Quintiles MSCI,5 and MSCI,1 is 1.050% with a t-statistic of 2.82. Similarly, the nine-factor alpha difference between Quintiles MSCI,5 and MSCI, 1 is 1.106% with a t-statistic of 3.09. This suggests that after controlling first for EMFX betas, and second for the commonly used hedge fund factors, the return difference between Quintiles MSCI,5 and MSCI,1 remains positive and significant.

Next, we perform the same bivariate portfolio test, but this time, we take into account the interaction between MSCI betas and EMBI+ betas. Each month, we rank hedge funds into quintiles according to their EMBI+ betas first and rank funds further into sub-quintiles based on their MSCI betas. This strategy enables us to check the raw return and risk-adjusted return (alpha) differences between high MSCI beta funds and low MSCI beta funds after controlling for EMBI+ betas. Table V, Panel A also provides the next month returns for the five MSCI beta quintiles, as well as the next month return and alpha differences between high MSCI beta funds and low MSCI beta funds when we control for the effect of EMBI+ betas. The average return difference between Quintiles MSCI,5 and MSCI,1 (i.e., high [[beta].sup.MSCI] vs. low [[beta].sup.MSCI]) is 1.299% per month with a Newey-West (1987) t-statistic of 2.00, suggesting that the positive relation between MSCI betas and future hedge fund returns remains significant after controlling for EMBI+ betas, a result that is in line with the results obtained from our earlier parametric Fama and MacBeth (1973) regressions. In addition, the four-factor and nine-factor alpha differences between Quintiles MSCI,5 and MSCI,1 are also positive (1.058% and 1.111%, respectively) and significant (with respective t-statistics of 2.61 and 2.81) indicating that these well-known hedge fund factors used in the literature are not capable of explaining the positive relation between [[beta].sup.MSCI] and the cross-section of future hedge fund returns after controlling for the effect of EMBI+ betas.

2. Bivariate Portfolios of EMFX Betas After Controlling for MSCI and EMBI+ Betas

We now examine the positive relationship between EMFX betas and future hedge fund returns after controlling first for MSCI betas and then for EMBI+ betas. For the first test, each month, we rank hedge funds into quintiles according to their MSCI betas and rank funds further into sub-quintiles based on their EMFX betas. This strategy enables us to check the raw return and risk-adjusted return (alpha) differences between high EMFX beta funds and low EMFX beta funds after controlling for MSCI betas. Table V, Panel B presents the next month returns for the five EMFX beta quintiles, as well as the next month return and alpha differences between high EMFX beta funds and low EMFX beta funds when we control for the effect of MSCI betas. The average return difference between Quintiles EMFX,5 and EMFX,1 is 0.466% per month with a Newey-West (1987) t-statistic of 2.03 suggesting that the positive association between EMFX betas and future hedge fund returns remains significant after controlling for MSCI betas. Similarly, the four-factor and nine-factor alpha differences between Quintiles EMFX,5 and EMFX,1 are also positive and significant. The four-factor alpha difference is 0.376% with a t-statistic of 2.09, and the nine-factor alpha difference is 0.329% with a t-statistic of 1.99. These results confirm our earlier findings of a positive and significant relation between EMFX betas and future hedge fund returns after controlling for MSCI betas and the well-known hedge fond factors.

Following the same sequence of analyses, we now explore the relationship between EMFX betas and future hedge fond returns, but this time after controlling for the effect of EMBI+ betas. To do this, each month, we rank hedge funds into quintiles according to their EMBI+ betas and rank funds further into sub-quintiles based on their EMFX betas. This sequence of ranking enables us to check the raw return and risk-adjusted return (alpha) differences between high EMFX beta funds and low EMFX beta funds after controlling for EMBI+ betas. Table V, Panel B also provides the next month return and alpha differences between high EMFX beta funds and low EMFX beta funds after controlling for the effect of EMBI+ betas. The average return difference between Quintiles EMFX,5 and EMFX,1 is 0.919% per month with a Newey-West (1987) t-statistic of 2.06 suggesting that the positive relation between EMFX betas and future hedge fond returns remains significant after controlling for EMBI+ betas. Moreover, the four-factor and nine-factor alpha differences between Quintiles EMFX,5 and EMFX, 1 are also positive (0.728% and 0.708%, respectively) and significant (with respective t-statistics of 2.55 and 2.43) indicating that these well-known hedge fund factors used in the literature are not capable of explaining the positive relationship between [[beta].sup.EMFX] and future hedge fond returns after controlling for EMBI+ betas.

All in all, these new results from bivariate portfolio analyses reinforce our major findings (from multivariate Fama and MacBeth (1973) cross-sectional regressions) that both MSCI betas and EMFX betas do not lose their predictive power over future hedge fund returns when their interaction with each other, as well as their interaction with EMBI+ betas, are taken into consideration.

3. Bivariate Portfolios of EMBI+ Betas After Controlling for MSCI and EMFX Betas

In multivariate Fama and MacBeth (1973) cross-sectional regressions, we have seen that the EMBI+ beta, although maintaining its predictive power when we control for the effect of EMFX beta, was losing its predictive power over future hedge fond returns when we controlled for the effect of MSCI beta in the same regression. In this subsection, we conduct bivariate portfolio tests to determine whether we will get similar results. Basically, we examine the relationship between EMBI+ betas and future hedge fond returns after controlling first for MSCI betas and then for EMFX betas by forming quintile portfolios. To perform this test, each month, we rank hedge funds according to their MSCI betas and rank funds further into sub-quintiles based on their EMBI+ betas. This methodology enables us to test the raw return and risk-adjusted return (alpha) differences between high EMBI+ beta funds and low EMBI+ beta funds after controlling for MSCI betas. Table V, Panel C indicates that the average return difference between Quintiles EMBI+,5 and EMBI+,1 (i.e., high [[beta].sup.EMBI+] Vs. low [[beta].sup.EMBI+]) is only 0.293% per month with a statistically insignificant Newey-West (1987) t-statistic of 0.91 suggesting that the positive and significant relation between EMBI+ betas and future hedge fond returns disappears after controlling for the effect of MSCI betas. Conducting the analyses in risk-adjusted returns (alphas) also generates very similar results. The four-factor and nine-factor alpha differences between high EMBI+ beta funds and low EMBI+ beta funds are all positive, but not significant.

Next, we examine the raw return and risk-adjusted return (alpha) differences between high EMBI+ beta funds and low EMBI+ beta funds after controlling for the effect of EMFX betas. Table V, Panel C shows that the average return difference between Quintiles EMBI+,5 and EMBI+,1 is 0.843% per month with a statistically significant t-statistic of 2.02. This suggests that the positive and significant relation between EMBI+ betas and future hedge fund returns continues after controlling for EMFX betas. Conducting the analyses in risk-adjusted returns also yield similar results. The four-factor and nine-factor alpha differences between high EMBI+ beta funds and low EMBI+ beta funds are all positive and significant.

In sum, our bivariate portfolio tests on EMBI+ betas support our earlier findings of a significant link between EMBI+ betas and future fund returns after controlling for the effect of EMFX betas, but also reinforce our earlier conclusion of a no significant link between EMBI+ betas and future fund returns after controlling for the effect of MSCI betas.

IV. Robustness Checks

In this section, we examine the predictive power of emerging market betas after we control for the performance of emerging market equities and the effect of liquidity risk. In addition, we analyze the predictive power of emerging market betas over future hedge fund returns for longer holding periods with investment horizons beyond one month.

A. Portfolio Alphas with Alternative Emerging Market Specific Factor Models

In our earlier portfolio analyses, we analyzed the performance difference between high beta funds and low beta funds in terms of raw returns and risk-adjusted returns. For risk-adjusted returns, we regressed the monthly time-series of return differences between high beta funds and low beta funds on commonly used hedge fund factors using Fama, French, and Carhart's four factors (MKT, SMB, HML, MOM), as well as Fama, French, Carhart, Fung, and Hsieh's combined nine factors (MKT, SMB, HMF, MOM, [DELTA]10[Y.sub.t], [DELTA]CredSpr, BDTF, FXTF, CMTF), and we determined whether the intercepts from these two regressions (i.e., the four-factor and nine-factor alpha) were statistically significant. One potential criticism for this method of calculating alphas between high beta funds and low beta funds is that the two factor models that we utilize do not include any emerging market specific factors. As such, the significant return difference between high beta funds and low beta funds could be due to the funds' simple passive exposure to emerging market equities since the emerging market equities outperformed the developed markets during the specific time period that we analyzed.

In order to overcome this potential criticism, we extend and modify the factor models that we utilize and include an emerging market specific factor in our analyses. Specifically, we regress the monthly time-series of return differences between high beta funds and low beta funds on the MSCI Emerging Market Equity Total Return Index (over the risk-free rate) only, and obtain the single factor alpha difference between high beta funds and low beta funds. Next, we modify the four-factor and nine-factor models slightly and replace the excess return on the Center for Research in Security Prices (CRSP) value-weighted index (the market factor, MKT) with the excess return on the MSCI Emerging Market Equity Total Return Index. Table VI reports the single factor alpha differences, as well as the modified four-factor and nine-factor alpha differences between high beta funds and low beta funds for the univariate ranked portfolios of MSCI betas, EMBI+ betas, and EMFX betas separately. For instance, the single factor alpha difference between high MSCI beta funds and low MSCI beta funds is 0.553% per month (6.64% per year) with a statistically significant Newey-West (1987) t-statistic of 1.99, suggesting that the positive return difference between high MSCI beta funds and low MSCI beta funds is still statistically and economically significant after controlling for the outperformance of the equities in emerging markets during our sample period. Similarly, the single factor alpha differences between high EMBI+ beta funds and low EMBI+ beta funds and high EMFX beta funds and low EMFX beta funds are, respectively, 0.646% and 0.595% per month (7.75% and 7.14% per year) with respective t-statistics of 2.19 and 2.46. This implies that the outperformance of equity markets in emerging markets cannot explain the statistically significant return differences between high beta funds and low beta funds. Finally, utilizing the modified four-factor and nine-factor models yields the same statistically positive and significant results for the MSCI, EMBI+, and EMFX beta portfolios with similar magnitudes of alpha differences and Newey-West (1987) t-statistics as in the single factor model, indicating that the outperformance of high beta funds is not due to the outperformance of equities in emerging markets (see the last two rows in Table VI). Finally, although not reported here to save space, we also analyze the predictive power of emerging market betas over future hedge fund returns for various subsample periods. We find the positive and significant relationship between the MSCI betas and future hedge fund returns and the EMFX betas and future hedge fund returns to continue at all times, suggesting that our results are not sample period specific.

B. Predictive Power of Emerging Market Betas After Controlling for Liquidity Risk

Sadka (2010) suggests that liquidity risk is an important determinant in the cross-section of hedge fund returns and finds that hedge funds that significantly load on liquidity risk subsequently earn 6% more average annual returns when compared to low liquidity risk loading funds. Therefore, it seems reasonable to test whether our main results from this paper (i.e., the positive and significant relation between emerging market betas and future hedge fund returns) will still hold after controlling for liquidity risk. In other words, the positive link between emerging market betas and future fund returns could be due to hedge funds' exposure to liquidity risk and, after controlling for the effect of liquidity risk, the positive relation between emerging market betas and hedge fund returns could disappear. In this section, we examine whether the predictive power of emerging market betas continues after we control for hedge funds' exposure to liquidity risk. Specifically, in the first stage, we regress individual hedge fund excess returns on Sadka's (2010) liquidity factor (LIQ) and three emerging market indices (MSCI, EMBI+, and EMFX) simultaneously, on a 36-month rolling window basis, to generate monthly time-series estimates of MSCI Betas, EMBI+ Betas, EMFX Betas, and LIQ Betas. (11) Next, in the second stage, we regress one-month-ahead individual hedge fund excess returns on the past month's aforementioned four betas (in different groupings) each month during the period January 1999 to December 2010, and determine whether the average slope coefficients on the emerging market betas from these Fama-MacBeth (1973) cross-sectional regressions are still positive and significant. (12) Table VII reports the average intercept and slope coefficients from the second stage Fama-MacBeth (1973) cross-sectional regressions. Each row in Table VII represents a different regression specification tested in our analyses.

In Table VII, in line with Sadka's (2010) findings, we find the average slope coefficient on the liquidity beta ([[beta].sub.LIQ]) to always be positive and significant, ranging in between 0.030 and 0.049, with statistically significant Newey-West (1987) t-statistics ranging from 1.88 to 2.62, no matter how the cross-sectional regression equation is specified. More importantly, however, we observe that controlling for the effect of liquidity beta ([[beta].sub.LIQ]) does not alter or weaken the statistically significant predictive power of MSCI beta and EMFX beta over future hedge fund returns. There still exists a positive and significant relationship between future hedge fund returns and MSCI beta and EMFX beta, whether the liquidity beta and hedge fund characteristics are controlled simultaneously or in different combinations of groupings. In Table VII, the average slope coefficient on MSCI beta is estimated to be between 1.407 and 1.533 (from alternative regression specifications), with the statistically significant Newey-West (1987) t-statistics ranging from 1.96 to 2.14. Similarly, the average slope coefficient on EMFX beta is estimated to be between 0.510 and 0.881, with the statistically significant Newey-West (1987) t-statistics ranging from 2.13 to 2.25. Similar to our earlier results in Table III, EMBI+ beta, again loses its predictive power over future hedge fund returns when it is combined together with the MSCI beta in the same regression equation. Overall, we can conclude that although liquidity beta is a strong determinant of the cross-sectional variation in hedge fund returns, its significance does not alter or weaken the positive and statistically significant relation between future hedge fund returns and MSCI beta and EMFX beta.

Next, although not reported here as a separate table to save space, we conduct a similar analysis examining the relationship between emerging market betas and future hedge fund returns after controlling for the effect of liquidity beta, via our nonparametric bivariate portfolio test. That is, each month, we form quintile portfolios of hedge funds double sorted first by liquidity beta and then by emerging market beta, and then calculate the next month return performance difference between high emerging market beta funds and low emerging market beta funds within each liquidity beta sorted quintile portfolio. During the period January 1999 to December 2010, we find the average return difference between high [[beta].sub.MSCI] and low [[beta].sub.MSCI] quintile funds to be 1.372% per month with a Newey-West (1987) t-statistic of 2.40 suggesting that the positive relation between MSCI betas and future fund returns remains significant after controlling for the liquidity beta. Similarly, we find the average return difference between high [[beta].sub.EMBI+] and low [[beta].sub.EMBI+] quintile funds to be 1.306% per month with a t-statistic of 2.38 indicating that the positive relation between EMBI+ betas and future fund returns also remains significant after controlling for the liquidity beta. Finally, we find the average return difference between high [[beta].sub.EMFX] and low [[beta].sub.EMFX] funds to be 1.436% per month with a t-statistic of 2.68 indicating that the positive relationship between EMFX betas and future fund returns remains significant after controlling for the liquidity beta as well. All in all, we conclude that both Fama-MacBeth (1973) regressions and bivariate portfolio tests indicate that the positive and significant relation between emerging market betas and future fund returns persist even after we control for the effects of liquidity beta.

C. Long-Term Return Predictability of MSCI Betas, EMBI+ Betas, and EMFX Betas

Our analyses thus far have only focused on the one-month-ahead return predictability of emerging market betas. However, from a practical standpoint, it would make sense to analyze the return predictability of emerging market betas for long-term periods. In this analysis, using our parametric test of Fama-MacBeth (1973) cross-sectional regressions, we examine the predictive power of emerging market betas over longer periods to see whether the return predictability of emerging market betas persists beyond the one-month investment horizon.

Table VIII reports the average slope coefficients from the cross-sectional regressions of single month future hedge fund excess returns (not compounded) on the current-month MSCI, EMBI+, and EMFX betas separately. That is, we regress t+n month ahead hedge fund excess returns (n = 2, 3, 4, ..., 12) on the current-month emerging market betas (separately) each month during the period January 1999 to March 2012, and test whether the average slope coefficients on the emerging market betas are still positive and significant. For a comprehensive analysis of the future return predictability of hedge funds with emerging market betas, we begin with the t 4- 2 month-ahead return predictability test and examine all future return predictabilities up to t + 12 months. As expected, and can be seen clearly in Table VIII, the predictive power of the emerging market betas decreases as we move from t + 2 month-ahead predictability to t + 12 month ahead predictability. Flowever, all of the average slope coefficients, even in the t + 12 month-ahead return predictability test, manage to remain statistically significant at least at the 10% level. Specifically, the average slope coefficients on the MSCI beta range from 1.248 to 1.398, with Newey-West (1987) t-statistics ranging from 1.82 to 2.18. Similarly, the average slope coefficients on the EMBI+ beta range from 0.890 to 1.022, with /-statistics ranging from 2.14 to 2.61. Finally, the average slope coefficients on the EMFX beta range from 0.580 to 0.775, with t-statistics ranging from 1.74 to 2.34. In sum, these results indicate that the predictive power of emerging market betas persists well beyond the one-month-ahead returns and, in fact, extends to even 12-month-ahead returns.

V. Predictive Power of Emerging Market Betas by Hedge Fund Strategies

Our analyses thus far have only focused on emerging market and global macro funds as these two hedge fund styles have the highest betas in magnitude, as well as the greatest percentage of significant betas with respect to emerging market securities. In this section, we examine whether our results would hold for other hedge fund investment strategies, as well as for the universe of all hedge funds, and speculate as to why these results may be unique to emerging market and global macro funds.

A. Variation in Emerging Market Betas and Portfolio Tests for Hedge Fund Strategies

Hedge funds have various trading strategies. Bali et al. (2011, 2012) categorize hedge fund investment styles under three broad investment strategies: 1) directional strategies, 2) semi-directional strategies, and 3) nondirectional strategies. Directional strategies willingly take direct market exposure and risk, while nondirectional strategies try to minimize the market risk altogether. Semi-directional strategies try to diversify the market risk by taking both long and short, diversified positions. Following the methodology of Bali et al. (2011, 2012), we also sort the different hedge fund investment styles (in our TASS database) into three broad hedge fund strategies. Accordingly, we classify the emerging market and global macro funds, the focal point of this study, under the directional strategy group, the equity market neutral, fixed income arbitrage, and convertible arbitrage funds under the nondirectional strategy group, and the long-short equity hedge, event driven, and multi-strategy funds under the semi-directional strategy group.

Given these three broad hedge fund investment strategies, it is not surprising to see varying degrees of exposure to a specific emerging market index by different hedge fund categories. Even within the same investment strategy, one can see varying degrees of exposure to the same emerging market index through time, as hedge fund managers adjust their exposures dynamically in response to changing market conditions. In order to understand the variation in emerging market betas among different hedge fund investment strategies clearly, Table IX Panel A presents, for each of the three broad hedge fund investment strategies separately, the cross-sectional average of individual hedge funds' time-series standard deviations of emerging market betas. Moreover, in Panel B of Table IX, we also report the cross-sectional average of individual funds' maximum minus minimum (max-min) emerging market beta differences for the same three broad hedge fund strategies. For comparison purposes, the cross-sectional averages of these two statistics across all hedge funds (irrespective of the hedge fund strategies) are also reported in the last row of each panel of Table IX. As illustrated in both panels, the standard deviation and max-min differences of emerging market betas increase monotonically as we move from the nondirectional strategy to the directional strategy. In other words, directional strategies, which include the emerging market and global macro hedge funds, have very high standard deviations and max-min differences of emerging market betas when compared to nondirectional and semi-directional strategies. Also, nondirectional strategies' standard deviations and max-min differences of emerging market betas are considerably smaller when compared to directional and semi-directional strategies. Finally, semi-directional strategies have standard deviations and max -min differences of emerging market betas that are very similar to the all hedge fund group.

Based on our finding of significant differences in the variation of emerging market betas through time across different investment strategies, we expect a larger variation in emerging market betas for a given strategy to improve the cross-sectional relation between emerging market betas and future hedge fund returns for that strategy if hedge fund managers have market-timing abilities. That is, if a larger variation in betas for a given strategy can translate into a larger variation in hedge fund returns and can, in turn, improve the predictive power of emerging market betas over future hedge fund returns, this may be an indication for the existence of the market-timing ability of some fund managers in that strategy.

We now investigate the predictive power of emerging market betas over future hedge fund returns for the three aforementioned broad hedge fund investment strategies separately and test if indeed a larger variation in betas through time is associated with a stronger predictive power of the emerging market betas. We perform this test by forming univariate quintile portfolios of emerging market betas for each hedge fund investment strategy separately and by analyzing the next month return and alpha differences between the high beta and low beta quintiles. Table X reports, for each of the three investment strategies and the all hedge funds category separately, the next month average return spreads, as well as the nine-factor alpha differences between the high emerging market beta and the low emerging market beta quintiles. As presented in the table, for all three emerging market beta portfolios tested, (i.e., MSCI beta, EMBI+ beta, and EMFX beta portfolios), the return and nine-factor alpha spreads between high beta (Quintile 5) and low beta (Quintile 1) funds increase monotonically as we move from nondirectional strategies to directional strategies. For instance, while the return spread between high beta funds and low beta funds ranges from 0.212% to 0.285% per month (among the three emerging market beta portfolios) for the nondirectional strategies, it ranges from 0.372% to 0.502% per month for the semi-directional strategies, and from 1.463% to 1.494% per month for the directional strategies. More importantly, the return and alpha spreads between the high beta funds and the low beta funds are statistically significant only for the directional strategies (i.e., emerging market and global macro funds), the focal point of our previous analysis. This finding is consistent for all of the three emerging market beta portfolios tested as well. On the other hand, the return and alpha spreads between Quintiles 5 and 1 are not statistically significant for the nondirectional and semi-directional strategies or for the all hedge fund category.

Combining these new set of results with the results we obtained earlier on the variation of betas through time across different investment strategies, we find a much stronger economic and statistical relation between emerging market betas and future returns for funds with sizeable and greater variation in emerging market betas. One possible explanation for this could be the market-timing ability of some fund managers. Many fund managers, especially those that pursue directional strategies, can actively vary their exposure to emerging market securities up and down in a timely fashion according to the macroeconomic conditions and the state of the financial markets and, as a result, can generate superior returns. In light of this conjecture, our results suggest that the stronger predictive power of emerging market betas in directional strategies is linked to the superior market-timing ability of these fund managers.

B. Market-Timing Tests for Three Broad Hedge Fund Strategies

While the results from previous analysis suggest the existence of a possible market-timing ability by some hedge fund managers in the directional strategy, the analysis conducted is not a direct market-timing test. Therefore, the results are not conclusive. In this section, we apply the direct market-timing test of Henriksson and Merton (1981), also utilized in Jagannathan and Korajczyk (1986), for three broad categories of hedge fund strategies separately to determine whether market-timing ability is specific to a group of hedge funds. We test for the market-timing ability of hedge funds with pooled panel regressions of individual hedge funds using Henriksson and Merton's (1981) model as follows:

[R.sub.i](t) = [alpha] + [[beta].sub.1] x X(t) + [[beta].sub.2] x Y(t) + [[epsilon].sub.i](t), (5)

where [R.sub.i](t) is the excess return (over the risk-free rate) on fund i, X(t) is the excess return on the market portfolio, Y(t) = max{0, -X(t)} is the factor implying market-timing ability, and [[epsilon].sub.i](t) is the residual, [alpha], [[beta].sub.i], and [[beta].sub.2] are, respectively, the intercept, the market beta, and the measure for market-timing ability. In this regression specification, a positive and significant value of [[beta].sub.2] implies superior market-timing ability.

Table XI Panel A reports the [[beta].sub.2] (and respective t-statistics) from Equation (5), where individual hedge fund excess returns are regressed on the market premium, as well as on the factor implying market-timing ability using pooled panel regressions for the sample period January 1999 to March 2012. Equation (5) is run separately for each of the three broad hedge fund categories (directional, semi-directional, and nondirectional), as well as for each of the three emerging market indices (MSCI, EMBI+, and EMFX) tested in our previous analyses. The t-statistics reported in parentheses are estimated using clustered robust standard errors, accounting for two dimensions of cluster correlation (fund and year). This approach allows for correlations among different funds in the same year, as well as correlations among different years in the same fund (see Petersen, 2009, for an estimation of clustered robust standard errors). Numbers in bold in Panel A denote statistical significance of the [[beta].sub.2] coefficients from Equation (5). Using the MSCI index as a proxy for the market portfolio, we find a positive and significant [[beta].sub.2] for the directional strategy only (a [[beta].sub.2] coefficient of 0.178 with a t-statistic of 2.40) suggesting that hedge funds following directional strategies, in general, possess market-timing ability. On the other hand, [[beta].sub.2] is negative and insignificant for the nondirectional strategy, and negative and significant for the semi-directional strategy indicating no evidence of market-timing ability for the other two hedge fund strategies. In fact, semi-directional strategy exhibits perverse market-timing activity. Moreover, we find that conducting the same regression analyses using EMBI+ or EMFX as a proxy for the market portfolio generate very similar results. In all of the different proxies of market portfolios tested, [[beta].sub.2] is always positive and significant in the directional strategy only, indicating that market-timing ability is specific to directional strategies only. Overall, we believe these results make sense in the real world setting of hedge funds as directional strategies (emerging market and global macro hedge funds), willingly take direct market exposure and risk, relying primarily on their market-timing ability for generating superior returns rather than engaging in individual stock selection activity. That is, since they are highly exposed to market risk, for these investment strategies, timing the switch in market trends is essential to their existence.

As a final analysis, in addition to pooled panel regressions, we run Henriksson and Merton's (1981) market-timing model with time-series ordinary least squares (OLS) regressions for each individual hedge fund separately, estimating one [[beta].sub.2] for each fund in the sample. Table XI Panel B reports the percentage of funds with positive and significant [[beta].sub.2]'s from individual OLS regressions for the three broad hedge fund investment strategies. Using the MSCI index as a proxy for the market portfolio in Equation (5), we find that the percentage of funds that exhibit market-timing ability (percentage of funds with positive and significant [[beta].sub.2]) is significantly higher for those funds following directional strategies when compared to funds following nondirectional and semi-directional strategies (e.g., 22% for the directional strategy vs. 6% and 4% for the nondirectional and semi-directional strategies, respectively). We obtain similar results when EMBI+ and EMFX are utilized as proxies for the market portfolio in our regression analyses as well. In sum, these results strengthen our earlier findings from pooled panel regressions in the sense that the market-timing ability of funds in the directional strategy is not an artifact of a couple of star performers in that strategy.

Finally, we can conclude that our previous results, which demonstrate a statistically significant link between emerging market betas and future fund returns in the directional strategy group, can be attributed, to some extent, to the evidence of the superior market-timing ability found only among directional strategy hedge fund managers.

VI. Conclusions

In response to the influx of new investment flows to emerging markets in the last decade and a half, this study contributes to the literature on hedge funds in a major way by providing an answer to the question of whether hedge funds are better off by increasing their exposures to emerging market securities. We assess hedge funds' exposures to various emerging market assets through univariate and multivariate estimates of betas, and then examine the predictive power of these betas over future hedge fund returns during the sample period January 1999 to March 2012. Until now, no study in the hedge fund literature has analyzed the cross-sectional variation in hedge fund returns in relation to funds' sensitivities (factor loadings) to emerging market securities.

After conducting Fama and MacBeth (1973) cross-sectional regressions and portfolio tests, we find strong, clear, and consistent results indicating a positive and significant relation between the emerging market equity beta (MSCI beta) and future hedge fund returns, as well as a positive and significant link between the emerging market currency beta (EMFX beta) and future hedge fund returns. Moreover, controlling for the commonly used hedge fund factors, hedge fund characteristics, and lagged fund returns do not alter or reduce the statistically significant predictive power of MSCI betas and EMFX betas over future hedge fund returns.

In addition, we show that incorporating an emerging market specific factor in the estimation of the alphas does not alter our portfolio results suggesting that the outperformance of equities in emerging markets cannot fully explain the statistically significant return differences between high beta funds and low beta funds. Furthermore, in a separate test controlling for the effect of liquidity risk, we find that the positive and significant relation between emerging market betas and future fund returns persists even after we control for the effect of the liquidity beta. Finally, we examine the long-term predictive power of emerging market betas over future hedge fund returns, and find that the predictive power of emerging market betas continues well beyond the one-month-ahead returns and, in fact, extend to even 12-month-ahead returns.

Ultimately, by dividing hedge fund styles into three broad categories (directional, semi-directional, and nondirectional) and conducting an analysis of the predictive power of emerging market betas for these three hedge fund investment strategies separately, we discover that the return and alpha spreads between high beta funds and low beta funds are statistically significant for the directional strategies only, but not for the nondirectional and semi-directional strategies. This result is further supported by the findings from market-timing tests that provide evidence of market-timing ability only among directional strategy hedge fund managers. In sum, the results from these two tests suggest that the outperformance of directional funds is related to the superior market-timing ability of these fund managers.

References

Agarwal, V and N.Y. Naik, 2004, "Risks and Portfolio Decisions Involving Hedge Funds," Review of Financial Studies 17, 63-98.

Agarwal, V and N.Y. Naik, 2000, "Multi-Period Performance Persistence Analysis of Hedge Funds," Journal of Financial and Quantitative Analysis 35, 327-342.

Aggarwal, R.K. and P. Jorion, 2010, "The Performance of Emerging Hedge Funds and Managers," Journal of Financial Economics 96, 238-256.

Aragon, G.O., 2007, "Share Restrictions and Asset Pricing: Evidence from the Hedge Fund Industry," Journal of Financial Economics 83, 33-58.

Asness, C., R. Krail, and J. Liew, 2001, "Do Hedge Funds Hedge?" Journal of Portfolio Management 28, 6-19.

Bali, T.G., S.J. Brown, and M.O. Caglayan, 2011, "Do Hedge Funds' Exposures to Risk Factors Predict Their Future Returns?" Journal of Financial Economics 101, 36-68.

Bali, T.G., S.J. Brown, and M.O. Caglayan, 2012, "Systematic Risk and the Cross Section of Hedge Fund Returns," Journal of Financial Economics 106, 114-131.

Bali, T.G., S. Gokcan, and B. Liang, 2007, "Value at Risk and the Cross-Section of Hedge Fund Returns," Journal of Banking and Finance 31, 1135-1166.

Brown, S.J., G. Gregoriou, and R. Pascalau, 2012, "Is it Possible to Overdiversify? The Case of Funds of Hedge Funds," Review of Asset Pricing Studies 2, 89-110.

Carhart, M.M., 1997, "On Persistence in Mutual Fund Performance," Journal of Finance 52, 57-82.

Edwards, F.R. and M.O. Caglayan, 2001, "Hedge Fund Performance and Manager Skill," Journal of Futures Markets 21, 1003-1028.

Fama, E.F. and K.R. French, 1993, "Common Risk Factors in the Returns of Stocks and Bonds," Journal of Financial Economics 33, 3-56.

Fama, E.F. and J.D. MacBeth, 1973, "Risk and Return: Some Empirical Tests," Journal of Political Economy 81,607-636.

Fung, W. and D.A. Hsieh, 1997, "Empirical Characteristics of Dynamic Trading Strategies: The Case of Hedge Funds," Review of Financial Studies 10, 275-302.

Fung, W. and D.A. Hsieh, 2000, "Performance Characteristics of Hedge Funds and CTA Funds: Natural versus Spurious Biases "Journal of Financial and Quantitative Analysis 35, 291-307.

Fung, W. and D.A. Hsieh, 2001, "The Risk in Hedge Fund Strategies: Theory and Evidence from Trend Followers," Review of Financial Studies 14, 313-341.

Fung, W. and D.A. Hsieh, 2004, "Hedge Fund Benchmarks: A Risk-Based Approach," Financial Analysts Journal 60, 65-80.

Fung, W. and D.A. Hsieh, 2009, "Measurement Biases in Hedge Fund Performance Data: An Update," Financial Analysts Journal 65, 1-3.

Fung, W. D.A. Hsieh, N.Y. Naik, and T. Ramadorai, 2008, "Hedge Funds: Performance, Risk, and Capital Formation," Journal of Finance 63, 1777-1803.

Getmansky, M., A.W. Lo, and 1. Makarov, 2004, "An Econometric Model of Serial Correlation and Illiquidity in Hedge Fund Returns," Journal of Financial Economics 74, 529-610.

Gupta, A. and B. Liang, 2005, "Do Hedge Funds Have Enough Capital? A Value at Risk Approach," Journal of Financial Economics 77, 219-253.

Henriksson, R.D. and R.C. Merton, 1981, "On Market Timing and Investment Performance. II. Statistical Procedures for Evaluating Forecasting Skills," Journal of Business 54, 513-533.

Jagannathan, R. and R.A. Korajczyk, 1986, "Assessing the Market Timing Performance of Managed Portfolios," Journal of Business 59, 217-235.

Jagannathan, R., A. Malakhov, and D. Novikov, 2010, "Do Hot Hands Exist Among Hedge Fund Managers? An Empirical Evaluation," Journal of Finance 65, 217-255.

Jegadeesh, N., 1990, "Evidence of Predictable Behaviour of Security Returns," Journal of Finance 45, 881-898.

Jegadeesh, N. and S. Titman, 1993, "Returns to Buying Winners and Selling Losers: Implications for Stock Market Efficiency," Journal of Finance 48, 65-91.

Jegadeesh, N. and S. Titman, 2001, "Profitability of Momentum Strategies: An Evaluation of Alternative Explanations," Journal of Finance 56, 699-720.

Lehmann, B., 1990, "Fads, Martingales, and Market Efficiency," Quarterly Journal of Economics 105, 1-28.

Liang, B., 2000, "Hedge Fund: The Living and the Dead," Journal of Financial and Quantitative Analysis 35, 309-326.

Liang, B. and H. Park, 2007, "Risk Measures for Hedge Funds: A Cross-Sectional Approach," European Financial Management 13, 333-370.

Newey, W.K. and K.D. West, 1987, "A Simple, Positive Semi-Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix," Econometrica 55, 703-708.

Patton, A. J., 2009, "Are 'Market Neutral' Hedge Funds Really Market Neutral?" Review of Financial Studies 22, 2495-2530.

Petersen, M.A., 2009, "Estimating Standard Errors in Finance Panel Data Sets: Comparing Approaches," Review of Financial Studies 22, 435-480.

Sadka, R., 2006, "Momentum and Post-Earnings Announcement Drift Anomalies: The Role of Liquidity Risk," Journal of Financial Economics 80, 309-349.

Sadka, R., 2010, "Liquidity Risk and the Cross-Section of Hedge Fund Returns," Journal of Financial Economics 98, 54-71.

Titman, S. and C. Tiu, 2011, "Do the Best Hedge Funds Hedge?" Review of Financial Studies 24, 123-168.

We are grateful to Marc Lipson (Editor) and an anonymous referee for their extremely helpful comments and suggestions. We also benefited from discussions with Turan Bali and seminar participants at the 2012 International Finance Congress, Ozyegin University, Federal Reserve Board, and Norwegian School of Economics. Finally, we thank Kenneth French, David Hsieh, and Ronnie Sadka for making a large amount of historical data publicly available in their online data library. All errors remain our responsibility.

Mustafa Onur Caglayan and Sevan Ulutas *

* Mustafa Onur Caglayan is an Assistant Professor of Finance in the Faculty of Economics and Administrative Sciences at Ozyegin University in Alemdag, Cekmekoy, Istanbul, Turkey. Sevan Ulutas is a Finance Ph D. student at Ozyegin University in Istanbul, Turkey.

(1) In our Trading Advisor Selection System (TASS) database, emerging market hedge funds and global macro hedge funds more than tripled in number from 1999 to 2007, and the assets under management (AUM) for these two styles increased eightfold over the same period. This big jump in both numbers and AUM is by far the largest observed in any hedge fund investment style reported to TASS.

(2) For other studies analyzing the risk return characteristics of hedge funds, see Fung and Hsieh (1997, 2000, 2001, 2004), Getmansky, Lo, and Makarov (2004), Gupta and Liang (2005), Liang and Park (2007), Fung et al. (2008), Patton (2009), Jagannathan, Malakhov, and Novikov (2010), Aggarwal and Jorion (2010), and Titman and Tiu (2011).

(3) In the TASS database, among all hedge fund investment styles, only emerging market and global macro hedge funds (i.e., directional hedge fund strategies) have high betas in magnitude, as well as high percentages of significant betas with respect to emerging market securities. The percentage of funds that have statistically significant emerging market betas (with respect to emerging market stocks, bonds, and currencies) is around 40% for emerging market and global macro hedge fund investment styles, whereas the magnitude of betas and the percentages of significant betas for the remaining hedge fund investment styles are quite low. Thus, our focus on emerging market and global macro hedge funds only (based on TASS classification) truly reflects the fund sample size of hedge funds that primarily invest in emerging market securities.

(4) All of the six emerging market index returns utilized in our analyses are expressed in USD and in excess of the risk-free rate.

(5) This finding is comparable to earlier studies of hedge funds. For example, Bali et al. (2011,2012) report a back- fill bias estimate of 2.09% and 2.03%, respectively.

(6) Fung and Hsieh (2000) also impose a 36-month return history requirement and find the survivorship bias estimate to be 0.60%.

(7) In addition to generating the monthly time-series estimates of emerging market betas from 36-month rolling window regressions, we also estimate emerging market betas using 24-month rolling window regressions. Using the 24-month rolling window regression beta estimates in Equation (2) generates similar results to those reported in Table II, suggesting that the results are not sensitive to the choice of the number of months utilized to generate betas. Moreover, in a separate test, we divide our full hedge fund sample into two sub-categories and examine the predictive power of MSCI betas, EMBI+ betas, and EMFX betas for emerging market hedge funds and global macro hedge funds separately. The cross-sectional relation between future fund returns and emerging market betas is found to be positive and significant for both sub-components of the full hedge fund sample as well.

(8) Agarwal and Naik (2000) and Jagannathan et al. (2010) also detect evidence of short-term persistence in hedge fund returns. In the stock literature, Jegadeesh and Titman (1993, 2001) find momentum in stock returns for three-month to 12-month horizons. On the other hand, Jegadeesh (1990) and Lehmann (1990) provide evidence of a short-term reversal in individual stock returns from a one-week to one-month horizon.

(9) This result is also consistent with Aragon's (2007) findings.

(10) The MKT (market) factor is the excess return on the stock market portfolio proxied by the value-weighted Center for Research in Security Prices (CRSP) index. The small minus big (SMB) factor is the difference between the returns on the portfolio of small size stocks and the returns on the portfolio of large size stocks. The high minus low (HML) factor is the difference between the returns on the portfolio of high book-to-market stocks and the returns on the portfolio of low book-to-market stocks. The MOM (winner minus loser) factor is the difference between the returns on the portfolio of stocks with higher past two-month to 12-month cumulative returns (winners) and the returns on the portfolio of stocks with lower past two-month to 12-month cumulative returns (losers). [DELTA]10Y is the monthly change in the US Federal Reserve 10-year constant maturity yield. [DELTA]CredSpr is the monthly change in the difference between Moody's BAA yield and the 10-year constant maturity yield. FXTF is a currency trend-following factor measured as the return on the Primitive Trend Following Strategy (PTFS) Currency Lookback Straddle. BDTF is a bond trend-following factor measured as the return on the PTFS Bond Lookback Straddle. CMTF is a commodity trend-following factor measured as the return of the PTFS Commodity Lookback Straddle.

(11) LIQ is the permanent variable component of Sadka's (2006) liquidity factor.

(12) Note that all previous analyses thus far are conducted for the full sample period January 1999 to March 2012. However, this particular analysis regarding the predictive power of emerging market betas after controlling for the effect of liquidity beta is conducted for the shorter sample period of January 1999 to December 2010 as the Sadka (2010) liquidity factor is only available through December 2010.
Table I. Descriptive Statistics

There are a total of 1,453 hedge funds that reported monthly returns
to TASS at some period between January 1999 and March 2012 in this
database, of which 760 are defunct funds and 693 are live funds. For
each year from 1999 to 2011, this table reports the number of hedge
funds entered into the database, the number of hedge funds dissolved,
the total AUM at the end of each year by all hedge funds
(in billion $s), and the mean, median, standard deviation, minimum,
and maximum monthly percentage returns on the equal-weighted hedge
fund portfolio.

Summary Statistics Year by Year

Year   Year    Entries   Dissolved   Year     Total AUM
       Start                         End      (billion $s)

1999    251      47         31        267            22.2
2000    267      48         56        259            10.5
2001    259      27         34        252            11.3
2002    252      55         16        291            14.8
2003    291      84         17        358            32.0
2004    358      123        22        459            55.8
2005    459      179        35        603            89.7
2006    603      158        50        711           130.4
2007    711      233        159       785           185.9
2008    785      110        124       771           179.9
2009    771      88         155       704            98.4
2010    704      35         165       574           110.2
2011    574      15         144       445            90.3

Summary Statistics Year by Year

Year   Equal-Weighted Hedge Fund (EWHF)
       Portfolio Monthly Returns (%)

       Mean    Median     Std. Dev.   Minimum    Maximum

1999    2.63    1.42        3.89      -1.54        10.07
2000   -0.14    0.82        3.27      -4.43         4.47
2001    1.05    2.03        2.82      -4.30         4.78
2002    0.65    1.04        2.04      -3.18         2.91
2003    2.42    2.60        1.81      -0.81         4.99
2004    0.95    1.56        1.78      -2.61         3.23
2005    1.32    1.49        1.99      -2.00         3.90
2006    1.58    1.57        2.15      -2.85         5.04
2007    1.70    1.86        1.90      -1.61         4.35
2008   -2.27   -2.42        3.59      -9.16         2.96
2009    2.07    1.36        2.54      -1.07         7.09
2010    0.80    0.61        2.19      -4.05         3.95
2011   -0.94   -0.84        2.51      -5.24         3.21

Table II. Univariate Fama-MacBeth Cross-Sectional Regressions of
One-Month-Ahead Hedge Fund Excess Returns on the Univariate
Factor Betas

This table reports the average intercept and slope coefficients
from the Fama-MacBeth (1973) cross-sectional regressions of
one-month-ahead hedge fund excess returns on the univariate
factor betas. In the first stage, monthly factor betas are
estimated for each fund from the univariate time-series
regressions of hedge fund excess returns on the factor over
a 36-month rolling window period. In the second stage, the
cross-section of one-month-ahead funds' excess returns are
regressed on the funds' factor betas each month for the period
January 1999 to March 2012. Newey-West (1987) t-statistics
are reported in parentheses to determine the statistical
significance of the average intercept and slope coefficients.
Numbers in bold denote statistical
significance of the average slope coefficients.

Emerging Market and Global Macro Hedge Funds

Intercept   [[beta].sup.MSCI]      [[beta].sup.EMBI+]

0.182              1.439#
(1.866)           (2.124)#
0.228                                    0.961#
(1.705)                                 (2.397)#
0.357
(2.265)
0.632
(2.919)
0.509
(2.116)
0.265
(2.335)

Intercept   [[beta].sup.JPEMVOL]   [[beta].sup.SPGSCI]

0.182
(1.866)
0.228
(1.705)
0.357              -2.275
(2.265)           (-1.469)
0.632                                     0.332
(2.919)                                  (0.398)
0.509
(2.116)
0.265
(2.335)

Intercept   [[beta].sup.SPGSPM]    [[beta].sup.EMFX]

0.182
(1.866)
0.228
(1.705)
0.357
(2.265)
0.632
(2.919)
0.509              0.333
(2.116)           (0.489)
0.265                                    0.745#
(2.335)                                 (2.063)#

Note: Numbers in bold denote statistical
significance of the average slope coefficients indicated with #.

Table III. Multivariate Fama-MacBeth Cross-Sectional Regressions of
One-Month-Ahead Fund Excess Returns on the Multivariate Factor Betas

This table reports the average intercept and slope coefficients from
the Fama-MacBeth (1973) cross-sectional regressions of one-month-ahead
hedge fund excess returns on the multivariate factor betas. In the
first stage, the MSCI, EMBI+, and EMFX factor betas ([[beta].sup.MSCI],
[[beta].sup.EMBI+] [[beta].sup.EMFX) are estimated for each fund from
the time-series regressions of hedge fund excess returns on the MSCI,
EMBI+, and EMFX factors using a 36-month rolling window period. In the
second stage, the cross-section of one-month-ahead funds' excess
returns are regressed on the funds' factor betas each month for the
period January 1999 to March 2012 with and without controlling for the
individual hedge fund characteristics (size, age, management fee,
incentive fee, redemption period, minimum investment amount, dummy
for lockup, and dummy for leverage) and the past month's individual
hedge fund returns. Newey-West (1987) t-statistics are given
in parentheses to determine the statistical significance of the
average intercept and slope coefficients. Numbers in bold denote
statistical significance of the average slope coefficients.

Emerging Market and Global Macro Hedge Funds

Intercept   [[beta].sup.MSCI]   [[beta].sup.EMBI+]   [[beta].sup.EMFX]

0.268          1.262#              0.292
(2.207)        (2.176)#            (1.191)
0.230          1.288#                                   0.445#
(2.202)        (2.096)#                                 (2.356)#
0.230                              0.778#               0.636#
(2.275)                            (2.201)#             (1.972)#
0.193          1.330#              0.475                0.472#
(2.108)        (1.976)#            (1.617)              (1.935)#
-0.113         1.271#              0.574#               0.430#
(-0.719)       (2.097)#            (2.180)#             (2.042)#

Intercept   Lagged       Size          Age
            Return

0.268
(2.207)
0.230
(2.202)
0.230
(2.275)
0.193
(2.108)
-0.113      0.073#       -0.059        -0.003#
(-0.719)    (4.582)#    (-0.358)      (-2.568)#

Intercept   Management    Incentive    Redemption
              Fee           Fee          Period

0.268
(2.207)
0.230
(2.202)
0.230
(2.275)
0.193
(2.108)
-0.113       0.205#        -0.001         0.002
(-0.719)    (3.513)#      (-0.214)       (1.156)

Intercept    Minimum        Dummy       Dummy
            Investment      Lockup     Leverage

0.268
(2.207)
0.230
(2.202)
0.230
(2.275)
0.193
(2.108)
-0.113       0.061#          0.235       -0.002
(-0.719)    (2.375)#       (1.528)      (-0.031)

Note: Numbers in bold denote statistical significance of the
average slope coefficients indicated with #.

Table IV. Univariate Quintile Portfolios of Hedge Funds Sorted by
[[beta].sup.MSCI], [[beta].sup.EMBI+], and [[beta].sup.EMFX]

Quintile portfolios are formed every month from January 1999 to March
2012 by sorting hedge funds based on their [[beta].sup.MSCI],
[[beta].sup.EMBI+], and [[beta].sup.EMFX] separately. Quintile 1 is
the portfolio of hedge funds with the lowest [[beta].sup.MSCI],
[[beta].sup.EMBI+], and [[beta].sup.EMFX], while Quintile 5 is the
portfolio of hedge funds with the highest [[beta].sup.MSCI],
[[beta].sup.EMBI+, and [[beta].sup.EMFX]. The table reports the
average [[beta].sup.MSCI], average [[beta].sup.EMBI+, average
[[beta].sup.EMFX]; and the average next month returns for each
quintile. The last three rows represent the differences between
Quintile 5 and Quintile 1, the monthly returns; the alphas with
respect to the Four-Factor Model of Fama-French-Carhart; and the
alphas with respect to the combined Nine-Factor Model of
Fama-French-Carhart and Fung-Hsieh. Average returns and alphas are
defined in monthly percentage terms. Newey-West (1987) t-statistics
are reported in parentheses. Numbers in bold denote statistical
significance.

Quintiles                            Average          Next Month
                                [[beta].sup.MSCI]       Average
                                                        Returns

Low [[beta].sup.MSCI]                 -0.030             0.319
2                                     0.142              0.549
3                                     0.323              0.748
4                                     0.588              1.101
High [[beta].sup.MSCI]                1.042              1.794

High [[beta].sup.MSCI]                                   1.475#
  -Low [[beta].sup.MSCI]
Return Diff.                                            (1.997)#
High [[beta].sup.MSCI]                                   1.207#
  -Low [[[beta].sup.MSCI]
4-Factor Alpha Diff.                                    (2.737)#
High [[beta].sup.MSCI]                                   1.270#
  -Low [[[beta].sup.MSCI]
9-Factor Alpha Diff.                                    (2.937)#

Quintiles                            Average          Next Month
                                [[beta].sup.EMBI+]      Average
                                                        Returns

Low [[beta].sup.EMBI+]                -0.112             0.383
2                                     0.195              0.639
3                                     0.454              0.666
4                                     0.831              0.948
High [[beta].sup.EMBI+]               1.588              1.877
High [[beta].sup.EMBI+]                                  1.494#
  -Low [[beta].sup.EMBI+]
Return Diff.                                            (2.168)#
High [[beta].sup.EMBI+]                                  1.201#
  -Low [[beta].sup.EMBI+]
4-Factor Alpha Diff.                                    (2.949)#
High [[beta].sup.EMBI+]                                  1.246#
  -Low [[beta].sup.EMBI+]
9-Factor Alpha Diff.                                    (3.201)#

Quintiles                            Average          Next Month
                                [[beta].sup.EMFX]       Average
                                                        Returns

Low [[beta].sup.EMFX]                 -0.097             0.369
2                                     0.262              0.540
3                                     0.551              0.790
4                                     0.998              0.981
High [[beta].sup.EMFX]                1.872              1.832
High [[beta].sup.EMFX]                                   1.463#
  -Low [[beta].sup.EMFX]
Return Diff.                                            (2.159)#
High [[beta].sup.EMFX]                                   1.200#
  -Low [[beta].sup.EMFX]
4-Factor Alpha Diff.                                    (3.026)#
High [[beta].sup.EMFX]                                   1.253#
  -Low [[beta].sup.EMFX]
9-Factor Alpha Diff.                                    (3.240)#

Note: Numbers in bold denote statistical
significance indicated with #.

Table V. Bivariate Portfolios of Hedge Funds Sorted by
[[beta].sub.MSCI], [[beta].sub.EMBI+], and [[beta].sub.EMFX]
Sequentially

Quintile portfolios are formed every month from January 1999 to March
2012 by sorting hedge funds based on their [[beta].sub.MSCI],
[[beta].sub.EMBI+], and [[beta].sub.EMFX] separately. Then, within
each beta-sorted portfolio, hedge funds are further sorted into
sub-quintiles based on the other two emerging market betas. For
instance, "Quintile MSCI,1" is the portfolio of hedge funds with the
lowest [[beta].sub.MSCI] within each [[beta].sub.EMBI+] or
[[beta].sub.EMFX] quintile portfolio (depending upon which beta's
effect is controlled for) and "Quintile MSCI,5" is the portfolio of
hedge funds with the highest [[beta].sub.MSCI] within each
[[beta].sub.EMBI+] or [[beta].sub.EMFX] quintile portfolio (depending
upon which beta's effect is controlled for). The table reports the
next month average returns of hedge funds for each quintile. The last
three rows represent the differences between Quintile 5 and Quintile
1, the monthly returns; the alphas with respect to the Four-Factor
Model of Fama-French-Carhart; and the alphas with respect to the
combined  Nine-Factor Model of Fama-French-Carhart and Fung-Hsieh.
Average returns and alphas are defined in monthly percentage terms.
Newey-West (1987) t-statistics are reported in parentheses. Numbers in
bold denote statistical significance.

                             Panel A

[[beta].sub.MSCI]         Controlling              Controlling
Quintiles             for [[beta].sub.EMFX]   for [[beta].sub.EMBI+]

                              Next                     Next
                              Month                   Month
                             Average                 Average
                             Returns                 Returns

MSCI.l                        0.414                   0.397
MSCI,2                        0.569                   0.559
MSCI,3                        0.861                   0.822
MSCI,4                        1.003                   1.039
MSCI,5                        1.665                   1.695
MSCI,5-MSCI.l                1.250#                   1.299#
Return Diff.                (2.063)#                 (1.996)#
MSCI.5 -MSCI.l               1.050#                   1.058#
4-Factor                    (2.824)#                 (2.613)#
  Alpha Diff.
MSCI.5-MSCI,1                1.106#                   1.111
9-Factor                    (3.094)#                 (2.815)#
  Alpha Diff.

                             Panel B

[[beta].sub.EMFX]         Controlling              Controlling
Quintiles             for [[beta].sub.MSCI]   for [[beta].sub.EMBI+]

                              Next                     Next
                              Month                   Month
                             Average                 Average
                             Returns                 Returns
EMFX,1,
EMFX.2                        0.750                   0.601
EMFX,3                        0.785                   0.713
EMFX,4                        0.789                   0.720
EMFX,5                        0.972                   0.958
EMFX.5-EMFX,1                 1.216                   1.520
Return Diff.                 0.466#                   0.919#
EMFX,5-EMFX,1               (2.027)#                 (2.063)#
4-Factor                     0.376#                   0.728#
Alpha Diff.                 (2.087)#                 (2.547)#
EMFX,5-EMFX, 1
9-Factor                     0.329#                   0.708#
Alpha Diff.                 (1.987)#                 (2.427)#

                             Panel C

[[beta].sub.EMBI+]      Controlling for            Controlling
Quinti|es              [[beta].sub.EMBI+]     for [[beta].sub.EMFX]

                              Next                     Next
                              Month                   Month
                             Average                 Average
                             Returns                 Returns
EMBI+,1
EMBI+,2                       0.847                   0.536
EMBI+,3                       0.775                   0.732
EMBI+,4                       0.817                   0.825
EMBI+,5                       0.933                   1.038
EMBI+,5-EMBI+,1               1.140                   1.380
Return Diff.                  0.293                   0.843#
EMBI+,5-EMBI+,1              (0.912)                 (2.022)#
4-Factor                      0.193                   0.753#
Alpha Diff.                  (0.782)                 (2.383)#
EMBI+,5-EMBI+,1
9-Factor                      0.190                   0.775#
Alpha Diff.                  (0.884)                 (2.417)#

Note: Numbers in bold denote statistical significance indicated
with #.

Table VI. Univariate Quintile Portfolios of Hedge Fund Alphas
Generated from Alternative Emerging-Market Specific Factor Models

Quintile portfolios are formed every month from January 1999 to March
2012 by sorting hedge funds based on their [[beta].sup.MSCI],
[[beta].sup.EMBI+], and [[beta].sup.EMFX] separately. Quintile 1 is
the portfolio of hedge funds with the lowest [[beta].sup.MSCI],
[[beta].sup.EMBI+], and [[beta].sup.EMFX] and Quintile 5 is the
portfolio of hedge funds with the highest [[beta].sup.MSCI],
[[beta].sup.EMBI+], and [[beta].sup.EMFX]. The table reports the
average [[beta].sup.MSCI], average [[beta].sup.EMBI+], average
[[beta].sup.EMFX], and the average next month returns for each
quintile. The last three rows represent the differences between
Quintile 5 and Quintile 1, the alphas with respect to the
single-factor MSC1 emerging market equity index model; the Alphas with
respect to the Four-Factor emerging markets modified model of
Fama-French-Carhart; and the Alphas with respect to the combined
Nine-Factor emerging markets modified model of Fama-French-Carhart and
Fung-Hsieh. Average returns and alphas are defined in monthly
percentage terms. Newey-West (1987) t-statistics are reported in
parentheses. Numbers in bold denote statistical significance.A76

Quintiles                             Avg.            Next
                               [[beta].sup.MSCI]     Month
                                                      Avg.
                                                     Returns

Low [[beta].sup.MSCI]                -0.030           0.319
2                                    0.142            0.549
3                                    0.323            0.748
4                                    0.588            1.101
High [[beta].sup.MSCI]               1.042            1.794
High [[beta].sup.MSCI]--Low                          0.553#
  [[beta].sup.MSCI]
Single-factor Alpha Dif.                            (1.993)#
High [[beta].sup.MSCI]--Low                          0.579#
  [[beta].sup.MSCI]
Modified 4-Factor Alpha Dif.                        (1.961)#
High [[beta].sup.MSCI]--Low                          0.616#
  [[beta].sup.MSCI]
Modified 9-Factor Alpha Dif.                        (2.090)#

Quintiles                            Avg.             Next
                               [[beta].sup.EMBI+]    Month
                                                      Avg.
                                                     Returns

Low [[beta].sup.EMBI+]               -0.112           0.383
2                                    0.195            0.639
3                                    0.454            0.666
4                                    0.831            0.948
High [[beta].sup.EMBI+]              1.588            1.877
High [[beta].sup.SEMBI+]--Low                        0.646#
  [[beta].sup.EMBI+]
Single-factor Alpha Dif.                            (2.189)#
High [[beta].sup.EMBI+]--Low                         0.641#
  [[beta].sup.EMBI+]
Modified 4-Factor Alpha Dif.                        (2.078)#
High [[beta].sup.EMBI+]--Low                         0.678#
  [[beta].sup.EMBI+]
Modified 9-Factor Alpha Dif.                        (2.312)#

Quintiles                            Avg.             Next
                               [[beta].sup.EMBI+]    Month
                                                      Avg.
                                                     Returns

Low [[beta].sup.EMFX]                -0.097           0.369
2                                    0.262            0.540
3                                    0.551            0.790
4                                    0.998            0.981
High [[beta].sup.EMFX]               1.872            1.832
High [[beta].sup.EMFX]--Low                          0.595#
  [[beta].sup.EMFX]
Single-factor Alpha Dif.                            (2.463)#
High [[beta].sup.EMFX]--Low                          0.608#
  [[beta].sup.EMFX]
Modified 4-Factor Alpha Dif.                        (2.409)#
High [[beta].sup.EMFX]--Low                          0.643#
  [[beta].sup.EMFX]
Modified 9-Factor Alpha Dif.                        (2.613)#

Note: Numbers in bold denote statistical significance indicated
with #.

Table VII. Multivariate Fama-MacBeth Cross-Sectional
Regressions of One-Month-Ahead Fund Excess Returns on the
Emerging Market Betas after Controlling for the Effect of
Liquidity Beta

This table reports the average intercept and slope
coefficients from the Fama-MacBeth (1973) cross-sectional
regressions of one-month-ahead hedge fund excess returns on
the multivariate emerging market betas after controlling for
the effect of the liquidity beta. In the first stage,
liquidity (L1Q), MSCI, EMB1+, and the EMFX factor betas
([[beta].sup.LIQ], [[beta].sup.MSCI], [[beta].sup.EMBI],
[[beta].sup.EMFX] are estimated for each fund from
the time-series regressions of hedge fund excess returns on
the LIQ, MSCI, EMBI+, and EMFX factors using a 36-month
rolling window period. In the second stage, the
cross-section of one-month-ahead funds' excess returns are
regressed on the funds' aforementioned factor betas (in
different groupings) each month during the period January
1999 to December 2010. Newey-West (1987) t-statistics are
given in parentheses to determine the statistical
significance of the average intercept and slope
coefficients. Numbers in bold denote statistical
significance of the average slope coefficients.

Intercept   [[beta].sup.LIQ]    [[beta].sup.MSCI]    [[beta].sup.EMBI+]

0.171            0.040#               1.533#
(1.455)         (2.479)#             (2.014)#
0.212            0.036#                                    1.146#
(1.409)         (2.118)#                                  (2.538)#
0.224            0.049#
(1.790)         (2.621)#
0.191            0.031#               1.501#               0.627
(1.907)         (2.049)#             (1.976)#             (1.757)
0.155            0.042#               1.437#
(1.454)         (2.419)#             (1.964)#
0.203            0.040#                                    0.895#
(1.698)         (2.286)#                                  (2.162)#
0.181            0.031#               1.495#               0.608
(1.683)         (1.876)#             (2.043)#             (1.703)
-0.225           0.030#               1.407#               0.694#
(-1.196)        (1.895)#             (2.138)#             (2.191)#

Intercept   [[beta].sup.EMFX]    Lagged     Size        Age
                                 Return

0.171
(1.455)
0.212
(1.409)
0.224             0.881#
(1.790)          (2.151)#
0.191
(1.907)
0.155             0.653#
(1.454)          (2.146)#
0.203             0.774#
(1.698)          (2.233)#
0.181             0.564#
(1.683)          (2.131)#
-0.225            0.510#         0.071#     -0.178    -0.003#
(-1.196)         (2.248)#        (3.758)#  -(0.875)   (-2.396)#

Intercept   Management   Incentive   Redemption     Minimum
               Fee          Fee        Period      Investment

0.171
(1.455)
0.212
(1.409)
0.224
(1.790)
0.191
(1.907)
0.155
(1.454)
0.203
(1.698)
0.181
(1.683)
-0.225        0.233#      -0.001        0.003        0.086#
(-1.196)     (3.508)#    (-0.261)      (1.476)      (2.639)#

Intercept   Dummy      Dummy
            Lockup    Leverage

0.171
(1.455)
0.212
(1.409)
0.224
(1.790)
0.191
(1.907)
0.155
(1.454)
0.203
(1.698)
0.181
(1.683)
-0.225       0.218     0.022
(-1.196)    (1.342)   (0.279)

Note: Numbers in bold denote statistical
significance of the average slope coefficients indicated with #.

Table VIII. Univariate Fama-MacBeth Cross-Section
Regressions of t + n Month Ahead Hedge Fund Excess Returns
on Emerging Market Betas

This table reports the average slope coefficients from the
Fama/MacBeth (1973) cross/sectional regressions of
multi-month-ahead hedge fund excess returns on the MSCI, EMBI+,
and EMFX betas. In the first stage, monthly MSCI, EMBI+, and
EMFX betas are estimated for each fund from the univariate
time/series regressions of hedge fund excess returns on the
MSCI, EMBI+, and EMFX factors over a 36/month rolling window
period. In the second stage, the cross-section of t + n
month ahead funds' excess returns are regressed on the
funds' betas each month during the period January 1999 to
March 2012. Newey/West (1987) t-statistics are reported in
parentheses to determine the statistical significance of the
average slope coefficients. Numbers in bold denote
statistical significance of the average slope coefficients.

t + n             [[beta].sup.MSCI]    [[beta].sup.EMBI]+
Predictability

t + 2             1.390 (2.126)#        1.002 (2.546)#
t + 3             1.398 (2.176)#        1.022 (2.610)#
t + 4             1.371 (2.038)#        1.017 (2.602)#
t + 5             1.364 (1.916)#        1.009 (2.466)#
t + 6             1.338 (1.931)#        0.994 (2.477)#
t + 7             1.287 (1.823)#        0.959 (2.308)#
t + 8             1.284 (1.874)#        0.939 (2.353)#
t + 9             1.248 (1.858)#        0.919 (2.275)#
t + 10            1.265 (1.974)#        0.890 (2.249)#
t + 11            1.255 (1.933)#        0.901 (2.179)#
t + 12            1.281 (1.936)#        0.915 (2.141)#

t + n             [[beta].sup.EMFX]
Predictability

t + 2             0.771 (2.229)#
t + 3             0.775 (2.341)#
t + 4             0.761 (2.250)#
t + 5             0.761 (2.176)#
t + 6             0.693 (2.111)#
t + 7             0.681 (2.090)#
t + 8             0.672 (2.054)#
t + 9             0.612 (1.873)#
t + 10            0.607 (1.899)#
t + 11            0.600 (1.834)#
t + 12            0.580 (1.739)#

Note: Numbers in bold denote statistical significance of
the average slope coefficients indicated with #.

Table IX. Dynamics of Hedge Funds' Emerging Market Betas by Three Broad
Hedge Fund Strategies

Panel A reports the cross-sectional average of individual funds'
time-series standard deviation of emerging market betas, while Panel B
reports the cross-sectional average of individual funds' max minus min
emerging market beta differences for each of the three broad hedge
fund investment strategies separately. For  comparison purposes, the
cross-sectional averages of these two statistics across all hedge
funds (irrespective of the hedge fund strategies) are also reported in
the last row of each panel. As can be noticed by reading Panels A and
B from top to bottom, Directional Category, which includes the
Emerging Market and Global Macro hedge fund investment styles have
significantly higher standard deviations and max--min differences of
emerging market betas as compared to Nondirectional Category, which
includes the Equity Market Neutral, Fixed Income Arbitrage, and
Convertible Arbitrage hedge fund investment styles. Also, the
directional strategies' standard deviations and max--min differences
of emerging market betas are considerably larger when compared to the
all hedge fund group, while the nondirectional strategies' standard
deviations and max--min differences of emerging market betas are
noticeably smaller as compared to the all hedge fund group. Finally,
Semi-Directional Category, which includes the Long-Short Equity Hedge,
Event Driven, and Multi Strategy hedge fund investment styles, have
standard deviations and max--min differences of emerging market betas
that are very similar to the all hedge fund group.

                             [[beta].sup.MSCI]   [[beta].sup.EMBI+]

Panel A. Standard Deviation of Emerging Market Betas

Nondirectional strategies          0.06                 0.15
Semi-directional strategies        0.08                 0.18
Directional strategies             0.11                 0.23
All hedge funds                    0.09                 0.19

Panel B. Maximum Minus Minimum Emerging Market Beta Differences

Nondirectional strategies          0.20                 0.54
Semi-directional strategies        0.28                 0.69
Directional strategies             0.39                 0.88
All hedge funds                    0.31                 0.73

                             [[beta].sup.EMFX]

Panel A. Standard Deviation of Emerging Market Betas

Nondirectional strategies          0.14
Semi-directional strategies        0.18
Directional strategies             0.25
All hedge funds                    0.20

Panel B. Maximum Minus Minimum Emerging Market Beta Differences

Nondirectional strategies          0.51
Semi-directional strategies        0.66
Directional strategies             0.94
All hedge funds                    0.73

Table X. Portfolios of Emerging Market Betas for Three Broad Hedge
Fund Strategies

For each of the three broad hedge fund investment style categories
(Nondirectional, Semi-Directional, and Directional), univariate
quintile portfolios are formed every month from January 1999 to March
2012 by sorting hedge funds based on their emerging market betas.
Quintile 1 (5) is the portfolio of hedge funds with the lowest
(highest) emerging market betas in each hedge fund category. The table
reports the differences in next month returns and nine-factor alphas
between Quintiles 5 and 1. Newey-West (1987) t-statistics are given in
parentheses. Numbers in bold denote statistical significance.

                                                           Q5-Q1
                                                     Return Difference

[[beta].sup.MSCI] portfolios     Nondirectional        0.285 (0.571)
                                 Semi-directional       0.502(1.275)
                                 Directional           1.475(1.997)#
                                 All hedge funds       0.553 (1.050)

[[beta].sup.EMBI+] Portfolios    Nondirectional        0.212 (0.502)
                                 Semi-directional       0.372(1.014)
                                 Directional           1.494(2.168)#
                                 All hedge funds       0.432 (0.856)

[[beta].sup.EMFX] Portfolios     Nondirectional        0.225 (0.537)
                                 Semi-directional       0.447(1.353)
                                 Directional           1.463 (2.159)#
                                 All hedge funds        0.513(1.103)

                                                         Q5-Q1
                                                     9-Factor Alpha
                                                       Difference

[[beta].sup.MSCI] portfolios     Nondirectional       0.049 (0.367)
                                 Semi-directional     0.461(1.618)
                                 Directional         1.270 (2.937)#
                                 All hedge funds      0.374(1.609)

[[beta].sup.EMBI+] Portfolios    Nondirectional       0.006 (0.048)
                                 Semi-directional     0.282 (1.167)
                                 Directional          1.246(3.201)#
                                 All hedge funds      0.236(1.335)

[[beta].sup.EMFX] Portfolios     Nondirectional       0.075 (0.468)
                                 Semi-directional     0.383 (1.560)
                                 Directional         1.253 (3.240)#
                                 All hedge funds      0.365 (1.570)

Note: Numbers in bold denote statistical significance indicated with #.

Table XI. Market-timing Tests of Three Broad Hedge Fund Strategies

Panel A reports the [[beta].sub.2] slope coefficients from pooled
panel regressions using Henriksson and Merton's (1981) market-timing
model for three broad hedge fund categories. Individual hedge fund
excess returns are regressed on the market premium, as well as on the
factor implying market-timing ability using pooled panel regressions
for the sample period January 1999 to March 2012. The market-timing
ability of hedge funds is tested with Henriksson and Merton's (1981)
following model: [R.sub.i](t) = [alpha] + [[beta].sub.1] X(t) +
[[beta].sub.2] Y(t) + [epsilon](t), where [R.sub.i](t) is the excess
return on fund i, X(t) is the excess return on the market portfolio,
Y(t) = max{0, -X(t)} is the factor implying market-timing ability, and
[epsilon](t) is the residual. [[beta].sub.2] is the measure for
market-timing ability. A positive and significant value of
[[beta].sub.2] implies superior market-timing ability. For the
t-statistics reported in parentheses, clustered robust standard errors
are estimated to account for two dimensions of cluster correlation
(fund and year). This approach allows for correlations among different
funds in the same year, as well as correlations among different years
in the same fund. Numbers in bold denote statistical significance. In
addition to pooled panel regressions, Henriksson and Merton's (1981)
market-timing model is also run with time-series OLS regressions for
each individual hedge fund separately, estimating one [[beta].sub.2]
for each fund in the sample. Panel B reports the percentages of funds
with positive and significant [[beta].sub.2]'s from individual OLS
regressions for three broad hedge fund strategies separately

                                      Nondirectional

Panel A. Pooled Panel Regressions

[[beta].sub.2] from using MSCI as         -0.051
the market portfolio

                                         (-1.445)
[[beta].sub.2] from using EMBI+ as        -0.201
the market portfolio

                                         (-1.439)
[[beta].sub.2] from using EMFX as         -0.039
the market portfolio

                                         (-0.329)

Panel B. Percentages of Funds with Positive and Significant
[[beta].sub.2]

% of Funds with positive                   5.9
significant [[beta].sub.2]'s using
MSCI as the market portfolio

% of Funds with positive                   5.4
significant [[beta].sub.2]'s using
EMBI+ as the market portfolio

% of Funds with positive                   5.2
significant [[beta].sub.2]'s using
EMFX as the market portfolio

                                      Semi-Directional   Directional

Panel A. Pooled Panel Regressions

[[beta].sub.2] from using MSCI as          -0.138           0.178
the market portfolio

                                          (-3.362)         (2.398)
[[beta].sub.2] from using EMBI+ as         -0.444           0.143
the market portfolio

                                          (-2.682)         (1.845)
[[beta].sub.2] from using EMFX as          -0.200           0.307
the market portfolio

                                          (-1.425)         (1.974)

Panel B. Percentages of Funds with Positive and Significant
[[beta].sub.2]

% of Funds with positive                    4.0             21.8
significant [[beta].sub.2]'s using
MSCI as the market portfolio

% of Funds with positive                    3.3             12.2
significant [[beta].sub.2]'s using
EMBI+ as the market portfolio

% of Funds with positive                    4.3             11.0
significant [[beta].sub.2]'s using
EMFX as the market portfolio
COPYRIGHT 2014 Financial Management Association
No portion of this article can be reproduced without the express written permission from the copyright holder.
Copyright 2014 Gale, Cengage Learning. All rights reserved.

Article Details
Printer friendly Cite/link Email Feedback
Author:Caglayan, Mustafa Onur; Ulutas, Sevan
Publication:Financial Management
Geographic Code:1USA
Date:Mar 22, 2014
Words:18560
Previous Article:US and domestic market gains and Asian investors' overconfident trading behavior.
Next Article:Geography, informal information flows and mutual fund portfolios.
Topics:

Terms of use | Privacy policy | Copyright © 2022 Farlex, Inc. | Feedback | For webmasters |