DOES CONDITIONING BIAS EXIST WITH ASIAN HEDGE FUNDS’ RETURNS?
Lan T.P. Nguyen, M. Y. Cheng, Sayed Hossain and Malick O. Sy
Multimedia University (MMU)
a, b, and c
Royal Melbourne Institute of Technology (RMIT)
d
Abstract
The availability of various databases raises an important question on the reliability of hedge fund data. Researchers using different databases may derive different conclusions on the performance of hedge funds, either overstating or understating the fund’s performance. The differences can be explained by a phenomenon called “data conditioning bias” as defined by Ackermann, McEnally, and Ravenscraft (1999). There are six forms of biases, namely survivor, termination, self-selection, liquidation, backfilling, and multi-period sampling biases. In this paper, we focus on three biases that might exist in a chosen sample of Asian hedge funds. These biases are survivorship bias, multi-period bias, and omission bias as defined in the paper. We choose a crisis-free period, i.e. January 2000 and June 2008 for the study. We select the largest possible sample size (456 funds) with at least 4-year-monthly data in order to avoid survivorship bias. Our results show that the studied sample has passed through the test with significant survivorship bias, but no significant multi-period as well as omission bias. Overall, we could conclude that the sample is fit to be used for further analysis in Asian hedge funds.
Keyword: conditioning bias, survivorship bias, multi-period bias, omission bias, Asian hedge funds
INTRODUCTION
Although hedge fund industry has existed for more than 50 years since its first introduction in 1940, information about hedge funds are not as widely available as compared to other types of traditional investment. This is mainly due to the specific characteristic of hedge funds, which are subjected to modest regulation by Security Commissions (SECs).
Basically, hedge fund managers are not allowed to advertise their investment publicly, thus they provide fund information to certain data vendors, on voluntary basis. Hedge fund managers normally release monthly funds information for two purposes: (1) to inform
1

existing investors on funds’ current performance and (2) to attract new ones to increase funds’ assets. Thus, hedge funds data are only available to the “qualifying public” (Cappocci
& Hubner, 2004). Few data vendors are supplying universal hedge funds data, they are
Paradigm LDC, TASS Management Limited (TASS), Hedge Fund Research Inc. (HFR),
Managed Account Reports Inc. (MAR), Asia Hedge and Eurekahedge. Most of the researches on hedge funds are done based on the data provided by the above databases. The last two data vendors are specializing in providing Asian hedge funds data. The availability of different databases creates an issue on the reliability of hedge fund data as researchers who use different hedge fund data may end up with different conclusions on the performance of hedge funds, either overstating or understating the performance of funds.
The differences in findings can be explained by the phenomenon called “data conditioning bias”, which often exists in hedge fund and mutual fund returns.
According to Ackermann, McEnally, and Ravenscraft (1999), data conditioning bias can be classified into six forms namely survivor, termination, self-selection, liquidation, backfilling, and multi-period sampling. It is stressed by Ball and Watts (1979) that a frequent problem in the time series analysis of firms’ financial data is the need to impose survivorship criteria in order to obtain sufficient observations for estimation purposes.
Survivor bias is defined by Ackermann, McEnally, and Rayenscraft (1999) as the bias of considering only the performance of funds that are alive and present in the database at the end of the sample period. Termination and self-selection biases are two subsets of survivor bias. As information on hedge funds are reported to the data vendors on voluntary basis, the supply of funds information may stop for two reasons. Firstly, if a fund performs poorly, the fund may close and exit from the database, this is called discontinuation.
Secondly, when a fund has raised enough capital and does not need to attract more investors, it will stop reporting its status to the data vendor and this is referred to as self-selection.
Excluding funds that performed poorly and died will cause an upward bias in funds’ average return. Excluding well-performing funds that do not want to increase their capital anymore may cause a downward bias in hedge funds’ average return. Funds that experience liquidation often take sometimes to liquidate their funds after declaring as dead funds. Thus their post-liquidation performance, which is often not good, might be reported in the database. If the performance of funds with and without the post-reporting returns appears to be significantly different, then liquidation bias exists.
2

When a new fund is added, the database providers typically request for the full performance history of the fund, thus only funds that survived the backfilling period are included. If the performance of funds with and without backfilling period is significantly different, then backfilling bias exists. Researchers often choose to study samples of funds that exist for a certain period of time. Multi-period sampling bias exists when funds that failed to survive for the whole period and new funds that emerged during that period are excluded. Conditioning on survival over multiple years may impart an upward performance bias.
Acknowledging the possibility of providing misleading information to investors as well as academicians, a number of hedge funds providers have began to keep data on funds that stop reporting to them. Examples are HFR who started to do this in December 1992 and
MAR in December 1993. The database that we used in this study - Eurekahedge, has kept all dead funds’ details since its establishment. For some special cases, if funds are requested to be removed from the database due to some reasons, their historical data will not be reported in the database. However, there are a few funds that do so. Given the practical purpose of our study is to provide investors information on funds that survive till the ending period of our study and also after that, we select funds that were introduced between January 2000 and
June 2008 and have at least 4 years of full monthly data, which means that funds that are introduced after July 2004 are not in our sample.
In this study, we attempt to examine if there is any significant survivorship as well as multiperiod biases exist in the chosen sample of funds during the study period. We are also interested to find out if omission bias exists. We define omission bias as upward bias that could occur as dead funds exist from the database upon their managers’ request.
LITERATURES ON CONDITIONING BIAS IN HEDGE FUNDS DATA
Even though data biases have been studied widely in mutual fund literatures
(Grinblatt and Titman, 1989; Brown et al., 1992; Blake et al., 1993; Brown et al., 1995;
Malkiel, 1995; Elton et al., 1996; Carhart, 1997; Carpenter et al., 1999; Bu et al., 2007), the issue become more severe in the case of hedge funds due to the lack of regulation in the hedge fund industry.
Fung and Hsieh (1997) estimate the attrition rate in hedge funds by examining 139 funds selected from the Paradigm database with at least 3 year returns in 1994 to avoid new
3

funds whose incubation period less than 3 years. The authors find that the attrition rate of hedge fund is about 4.3% per year. The authors find that the low attrition of hedge funds has little impact on their result. For selection bias, the authors find evidence suggesting that this bias could be eliminated due the offsetting effect made between fund managers who are eager to market their funds in databases to attract more investors and managers, who have no intention to do so as their desired maximum fund capacities have reached. To avoid multi-period sampling bias, the authors require funds to be included in their sample of study to have at least 36 months of data. Thus, thus the results of their study benefit investors who desire to invest in a fund with a track record of at least 36 months.
Ackermann, McEnally, and Ravenscraft (1999) investigate the impact of six forms of related data-conditioning biases: survivor, termination, self-selection, liquidation, backfilling, and multi-period sampling. To study survivor bias is tested in a manner that expose a difference between the performance of surviving and disappearing funds. The authors include the performance of each disappearing fund during its entire available 1988–
1995 return history into the set of extant funds whose returns span over the same interval.
The set of extant funds include all funds that report at the end of their sample period even if they do not report for a full two years. Their inclusion allows them to separate survivor bias from the multi-period sampling bias. The authors then weight each fund’s return by the number of months of the disappearing fund’s return history with the purpose of avoiding the possibility that those funds with the shortest return histories, which are most likely to produce performance outliers, have the largest impact on their results. Their finding shows that disappearing funds tend to be below–average performers; however results are not statistically significant. The underperformance of disappearing funds is larger for the Sharpe ratio than the monthly total return, implying the high volatility of this group of funds. They also find that there is insignificant positive correlation between variance and disappearance.
In examining termination bias, the authors find that terminating funds underperform the comparable set of extant funds in five of six comparisons, and three out of those are statistically significant. When testing self-selection bias, the authors find that funds that stop reporting outperform the relevant set of extant funds in five of six cases although the difference is never significant. Their results support the fact that both superior and inferior funds voluntarily end reporting. As most of funds after stop reporting due to liquidation are still followed up by the data vendor, and thus their subsequent monthly performance are recorded. Majority of these liquidated funds experience substantial lost in value during the
4

redemption period. At the authors’ request, HFR tracked a majority of terminating funds through redemption period to examine whether the information recorded in their database reflects the timing and the amount of money returned to investors. Their findings are the average loss in fund value beyond the information contained in the database is only 0.7 percent, and the average delay between the final return reported in the database and actual redemption is 18 days. In short, post-reporting returns appear to have a negligible impact on their results. Given the fact that it is hard to know when a fund is added to the database in order to determine the exact amount of backfilling, the authors use an indirect approach commonly employed in hedge fund research papers, that is to eliminate the first two years of reported data. The authors compare the results for each of their time periods with and without the first two years of each fund’s returns eliminated. They find that eliminating the first two years of each fund’s return data increases the raw return by 0.05 percent per year and decreases the Sharpe ratio by 0.003, implying that non-reporting or disappearing hedge funds contain more winners than is typical of other databases where reporting is more mandatory or universal. To test multi-period sampling bias in their sample, the authors compare funds with less than full performance history during 1994 – 1995 to funds with less full history during this period, that is disappear or begin during this interval. Their results show that fund with a less than complete 1994 – 1995 return history actually outperform funds with a complete history by 12 basis points per month, implying that there is no evidence conditioning on a fund existing for multiple years biases performance upward. In general, their findings are consistent with the mutual fund literature that hedge funds that cease to exist perform poorly. However, in comparison with mutual funds, conditioning biases for hedge funds in the study are found to be weaker.
Brown, Goetzmann, and Ibbotston (1999) exclude the all returns of funds in the year that they become defunct and construct a sample at a point in time and draw inferences from the prior history of returns. By imposing two survival conditionings in their sample, that are a fund is required to survive the entire 7-year history and a fund need to exist in the last period of the sample, which is 1995. Their conditioning on existence at the end of the period imparts a bias in raw returns of about 3% per year, which is the average over all funds in the index. Their analysis of a database that includes both defunct and surviving funds suggest that survival conditioning may have important effects on the ex post observed historical performance. They suggest that investors who use past track records should anticipate that
5

historical returns probably exceed ex ante expected future returns. Thus, they should expect future volatility if they depend on the past performance.
Liang (2000) examines survivorship bias in hedge fund returns by comparing two large databases HFR (1,162 funds) and TASS (1,627 funds). The author finds that the survivorship bias exceeds 2% per year. The biases are found to be different across styles.
The author reconcile the conflicting results about survivorship bias in previous studies by showing that the two major hedge fund databases contain different amounts of dissolved funds. Their finding shows that poor performance is the main reason for a fund’s disappearance. In addition, the author also finds significant differences in fund returns, inception date, net asset value, incentive fee, management fee, and investment styles for the
465 common funds covered by both employed databases.
Fung and Hsieh (2000) follow the method used by Malkiel (1995) analyze the survivorship bias of the TASS hedge fund database as from 1994 to 1998. Their samples contain 1,120 surviving and 602 defunct hedge funds. Their estimated survivorship bias is
3% per year, which is the difference between the average returns of 13.2% and 10.2% for surviving and observable portfolios respectively. The authors also study the survivorship bias for funds-of-hedge funds, which is estimated only 1.4% per year less than half that of individual hedge funds. For instant history bias, the authors update the Brown, Goetzmann and Park (1997) results for commodity funds. After deleting 27 monthly returns of an average incubation for the period 1989-1997, their adjusted observable portfolio has an average return of 11.9%, which is lower than their observable portfolio’s return of 15.5%.
Thus, the instant history bias for commodity funds is estimated about 3.6% per year. To estimate instant history bias for TASS database, the authors take into account an average incubation period of 12 monthly for each hedge fund, the instant history bias is estimated about 1.4% per year. The authors also prove that funds-of-hedge funds contain less instant history bias (0.7%). To measure selection bias in hedge funds, the authors suggest the use of equally-weighted average of the returns of funds-of-hedge funds in a database. Their result shows that there is little selection bias in large sample of funds-of-hedge funds. This is due to the fact that funds-of-hedge funds invest in a diversified portfolio of hedge fund styles and less prone to capacity constraints which make them more likely to disclose their track records in order to attract more capital. To investigate to the impact of the requirement on funds to have a minimum return history on their average returns, the authors create a
6

portfolio that include only funds with at least 36 months of return history. The results show little evidence of multi-period sampling bias.
Fung and Hsieh (2002) suggest the solution to mitigate data biases is to construct hedge fund benchmarks that can measure the performance of hedge funds through observing the investment experience of hedge fund investors themselves that is fund of funds (FOFs).
Due to the natural process of diversification of fund of funds, measurement errors are minimized. Their empirical results suggest that FOFs do not suffer from spurious biases arising from unrealistic asset allocation schemes and selection bias is avoided as all individual hedge funds have some FOF investors that might report their performance to different hedge fund database.
Edwards and Caglayan (2001) employ Managed Account Reports (MAR/Hedge) data of 1665 hedge funds by August 1998 to study the performance and managerial skills of hedge funds. In attempting to estimate the survivorship bias in their sample, the authors include the return histories for 496 non-surviving hedge funds in their sample, and compute excess returns or alphas for a sample of only surviving funds and a sample of both surviving and dead funds. They find that a survivorship bias of 1.85%, which is comparable to the previous studies, Liang (2000) and Fung and Hsieh (2000). The authors also estimate instant history bias in their sample and find that the average return for hedge funds during their first year of existence is about 1.17% points higher than their average returns in subsequent years. Therefore, the authors exclude the first 12 months of returns for all funds in their sample to avoid an instant history bias. To avoid multi-period sample bias, the authors impose 24-month-minimum requirement for funds that are included in their sample. The authors do not follow Fung and Hsieh (2000)’s method which selects only funds with at least 36 monthly return data to avoid the excess exclusion of non-surviving funds that would cause a survivorship bias problem.
Capocci and Hübner (2004) examine survivorship bias defined and formulated by both Ackermann et al. (1999) and Liang (2000) for the whole sample period and for two sub-periods of 1984-1993 and 1994-2000. Their results for the whole period have a monthly survivorship bias of 0.36% (4.45% per year) calculated by using formula in Ackermann et al. (1999), and a bias of 0.07% per month (0.9% per annum) calculated by using formula in
Liang (2000). Results for two sub-periods show low and higher bias for the period before and after 1994 respectively. They find that the reason for low bias for the period before 1994
7

is due to the non-availability of data on dissolved funds before 1994 for Managed Account
Reports (MAR) and before 1993 for Hedge Fund Research (HFR). This implies that data provided by the databases for this period were not reliable. The authors’ results have higher value for the whole period and for the second sub-period, and similar result for the first subperiod as compared to those obtained in Ackermann et al. (1999). The difference in their result is mostly due to the different time period analyzed. Their results for period 1994-2000 based on Liang (2000)’s definition, are lower than the 0.3% monthly bias found in Fung and
Hsieh (2000), the 3% bias found in Liang (2001), and the industry bias of 3% stressed by
Amin and Kat (2001). In examining instant return history bias (backfilling bias), the authors employ methods used by Park (1995), Brown et al. (1999), and Fung and Hsieh (2000) with two steps. In the first step, the authors estimate average monthly return using an observable portfolio that invests in all funds from their database each month. In the second step, the authors estimate the average monthly return using adjusted observable portfolio that invests in all these funds after deleting the first 12, 24, 36, and 60 months of returns. Their estimated bias for the whole sample is approximately 0.9 per year, lower that what is found by Fung and Hsieh (2000). However, their results for sub-period of 1/1994-6/2000 are similar to what is obtained in Fung and Hsieh (2000). The difference in their results is due to the different time period studied. Their results imply that the longer the estimation period, the bigger the bias. Due to the greater instant history bias found for period before 1994, and period after
1994 when 24 months or more are deleted, the authors separate and focus only on performance of hedge funds for the second sub-period of 1994-2000, in which less bias exists. The uneven bias exist in two sub-periods does not allow the authors to produce sound results for the whole sample period of 1984-2000.
Baquero, Horst, and Verbeek (2005) take into account look-ahead bias known as multi-period sampling bias over a period of 1994-2000 while studying the performance persistence in raw returns and style-adjusted returns of 1,797 hedge funds. The authors use a multiplication of the performance measure such as the average return over the ranking period, with a weight factor, which is the ratio of an unconditional non-liquidation probability in the numerator and a conditional non-liquidation probability in the denominator. Results show that for the one-quarter horizon, the corrected results indicate a clear pattern of positive persistence in raw fund returns. For the annual horizon, the pattern is also consistent with positive persistence, but statistically insignificant.
8

Steri, Giorgino, and Viviani (2009) examine some potential biases in their sample while analysing the performance and persistence of Italian hedge funds. Following Capocci and Hübner (2004)’s method, the authors find that a survivorship bias of (-0.64%) exists in their sample, implying that the non-surviving funds have a higher average return than that of other funds. The authors believe that this is due to casualty as the number of funds that stopped reporting is small in their sample. To estimate the instant history bias, the authors use Capocci and Hübner (2004)’s method and find that there is (-0.21%) per year of history bias in their sample. The negative bias result might be due to either more favourable recent market conditions than past market condition or casualty effect. As selection bias is not possible to estimate due to the availability of information of why a manager stop reporting his performance, the authors assume this bias to be equal to zero or not relevant in their study.
RESEARCH METHODOLOGY
We now evaluate data reliability for Asian hedge funds through investigation on types of conditioning bias that might exist in our sample collected from Eurekahedge database, which in turns might have an impact on the performance of Asian hedge funds.
For this purpose, we examine the following 4 different samples between January 2000 and
June 2008. The total number of funds present in Eurekahedge by June 2008 is 1428, including surviving (1032) and dissolved (396) funds.
FIGURE 1: SAMPLES FOR CONDITIONING BIASES
In this chart, we present the breakdown for 4 different constructed samples used to test conditioning biases. Sample A contains all surviving funds, closed funds, and omitted funds.
Sample B contains all surviving funds and closed funds. Sample C contains all surviving funds. Sample D contains only surviving funds with at least 4 year monthly return data.
9

Eurekahedge normally keeps all history information of dead funds in the database unless fund managers of those dead funds request to the data vendor to remove them totally from the database. We identified 75 funds that were omitted from the database for this reason between January 2000 and June 2008. We then compare the performance of samples with (sample A) and without (sample B) these omitted funds. The omission of closed funds may cause upward bias in overall performance of Asian hedge funds during the study period. This kind of bias has not been mentioned in other literature before, which may be due to the non-reporting on these cases by the data vendors. Thus, we give a new name for this kind of bias as “omission bias”.
We form four different samples (A, B, C, and D) in order to test conditioning bias that might exist in the Eureka database as well as in our sample of Asian hedge funds.
Sample A represents all AHFs in the industry, Sample B represents AHFs that are present in the Eurekahedge database, Sample C represents AHFs that survive throughout the sample period, and Sample D represents AHFs that have had at least 4 years of performance as of
15 th
June 2008. Descriptive statistics for the four constructed samples (A, B, C, and D) described in Table 1 below.
10

TABLE 1: DESCRIPTIVE STATISTICS OF DIFFERENT SAMPLE OF ASIAN
HEDGE FUNDS
In this table, we report average monthly return, standard deviation, Sharpe ratio for different constructed samples of Asian hedge funds.
Samples
Sample B
Sample A
Sample C
Sample D
Average
Monthly
Return
0.83
0.83
0.91
0.89
Standard
Deviation
Sharpe
Ratio
4.36
4.33
4.42
4.37
-0.79
-0.79
-0.76
-0.78
To test the difference between two population means, we form the following null and alternative hypotheses for a two-tailed test.
Ho: µ
1
- µ
2
= 0
Ha: µ
1
- µ
2
µ≠ 0
Where: µ
1 and µ
2
denote the means for populations 1 and 2, respectively.
Our test statistics is as follows (Ott & Mendenhall, 1990)
are two samples’ means. S p
is the common population standard deviation and is calculated as n
1 and n
2
are the two sample sizes. s
1
and s
2
are the two samples’ standard deviations. The degree of freedom for the above t-test is
11

If t-statistics does not fall in the rejection region, we conclude that there is insufficient evidence to indicate a difference in the means of two samples. We exclude outliers from the samples before carrying our testing.
FINDINGS
Our t-test result in Table 2 confirms no evidence on the difference between two samples’ average returns. Results suggest that there is no significant evidence of omission bias exists in Asian hedge funds’ performance between January 2000 and June 2008.
To qualify our data for further tests, we attempt to test whether our sample C and D are totally free from two conditioning biases: survivorship bias and multi-period bias, respectively. To tests the survivorship bias in our sample C between January 2000 and June
2008, we measure the difference between the performance of sample that contains only surviving funds (sample C) and the sample that contains both disappearing and surviving funds (sample B). Disappearing funds are funds that disappear or close by the end of our sample period, 30 June 2008, but exist within the sample period. We do not intend to separate these funds into discontinued funds and self-stop-reporting funds as the data of these two sets of funds are not available after their closing date, and thus we would expect unavoidable self-selection bias as a result in Eurekahedge which would definitely cause downward bias in the performance of Asian hedge funds. Surviving funds are funds that survive by June 2008 and have return series data within the sample period. Following Liang
(2000), we carry out t-test for two samples reveals that average monthly return for sample of surviving funds (Sample C) is slightly higher (0.02%) compared to that of full sample data
(Sample B) for the study period. Our results suggest that including dead funds in a sample will cause downward bias in performance of Asian hedge funds. Clearly, poor performance is the main reason for funds to cease their operation. This somewhat confirms the reason why funds exit from databases as discussed by Ackermann, McEnally, and Ravenscraft
(1999). However, the result is never statistically significant.
We repeat the same procedure to test survivorship bias between samples of surviving funds (Sample C) and full sample including omitted funds (Sample A). We find similar results, that is there is an upward bias in returns for sample of surviving funds, but not significant at 1% level ( Table 2 )
12

As a requirement for our performance tests for Asian hedge funds, funds that are included in our sample of study need to have at least 4 years of full monthly return data ending by 30 June 2008. This requirement ensures the high possibility of surviving for funds after the sample period, June 2008 as the average life of a disappearing fund is about 3.4 years for Asian hedge funds. Therefore, we would suspect that sample D may suffer from multi-period sampling bias that would overstate the performance of Asian hedge funds.
From the average returns of Table 2 , we run t-test to find out if there is significant difference that exists for the above-mentioned performance measures between two samples: sample of all surviving funds (sample C) and sample of surviving funds with at least 4 year monthly return data (sample D). Results from t-test show that sample with surviving funds with less than 4 years monthly return data for the period from July 2004 to June 2008 outperforms sample with surviving funds that have full 48 monthly data for the period by (-0.02%) per month, and this bias is not significant at 5 percent level. Overall, our results suggest that there is no evidence that multi-period bias cause performance upward in our sample that contains funds with at least 4 years of monthly data.
We then test whether any bias between samples with surviving funds with at least 4 year monthly return data (Sample D) and full sample excluding omitted funds (Sample B) by the data vendor. Results in Table 2 show the average return of sample with surviving funds with at least 4 years of data is slightly higher that of the sample with full funds excluding omitted funds, however it is not significant at 5% level. Thus, we cannot reject the hypothesis “no difference in returns between two samples”.
We also carry out test to confirm if there is no survivorship bias in our sample of surviving funds with at least 4 year monthly data (Sample D) and sample of all funds including omitted funds by the data vendor (Sample A). Our p-value for t-statistics for this test shown in Table 2 does not allow us to reject the null hypothesis at 1% level.
Overall, our results through all tests without outliers confirm there is no significant survivorship and multi-period sampling bias in sample of all surviving funds as well as sample of surviving funds with at least 4 year monthly data.
13

TABLE 2: BIAS FOR THE PERIOD OF JANUARY 2000 - JUNE 2008 FOR
SAMPLE WITHOUT OUTLIERS
In the following tables, we report bias in monthly returns between the following paired samples: (1) sample of all surviving funds and sample that contains both surviving and disappearing funds, (2) sample of surviving funds with at least 4 year monthly return data and sample that contains both surviving and disappearing funds, (3) sample of all surviving funds and sample of all funds including omitted funds by the data vendor, (4) sample of surviving funds with at least 4 year monthly return data and sample of all funds including omitted funds by the data vendor, (5) sample of all surviving funds and sample of surviving funds with at least 4 year monthly return data, and (6) sample of all funds excluding omitted funds and sample of all funds including omitted funds by the data vendor.
CONCLUSION
Samples
Sample C & Sample B
Sample D & Sample B
Sample C & Sample A
Sample D & Sample A
Sample D & Sample C
Sample A & Sample B
Difference t-statistics p-value
.02842 -.114 .909
.00950
.02069
.00178
-0.02
-.00772
.038
.083
.007
-1.69
-.986
.970
.934
.994
0.09
.326
In summary, sample of surviving funds have upward bias compared to the full sample of Asian hedge funds between January 2000 and June 2008, which is expected. This is due to the fact that disappearing funds seems to have poor performance and thus exist from the database. This finding is consistent with Ackermann, McEnally, and Ravenscraft
(1999)’s finding for HFR and MAR. However, the result is never significant. We also examine the multi-period sampling bias in our sample that contains only surviving funds with at least 4 year monthly return data by 30 June 2008. Our findings show no evidence that this bias exist in our sample of study. We have discovered the fact that 75 hedge funds that have stopped reporting to the database and requested to be removed from Eurekahedge between January 2000 and June 2008. In attempt to test if the omission of these funds’ data would cause any upward bias in Asian hedge funds sample between the study periods, we
14

find no evidence that this “omission bias” exist in our collected sample. Therefore, our biasfree sample is qualified to be used for further study.
REFERENCES
Ackermann, C., McEnally, R., & Revenscraft, D. (1999). The Performance of Hedge Funds:
Risk, Return, and Incentives. The Journal of Finance, Vol. LIV, No. 3, pp. 833-874.
Ball, R., & Watts, R. (1979). Some Additional Evidence on Survival Biases. The Journal of
Finance , Vo. XXXIV, No. 1, pp. 197-206.
Baquero, G., Horst, J. t., & Verbeek, M. (2005). Survival, Look-Ahead Bias, and Persistence in Hedge Fund Performance. Journal of Financial and Quantitative Analysis , Vol. 40, No.
3, pp. 493-517.
Blake, C., Elton, E., & Gruber, M. (1993). The Performance of Bond Mutual Funds. Journal of Business, Vol. 66, , 371 – 403. .
Brown, S. J., & Goetzmann, W. N. (1995). Performance Persistence. Journal of Finance,
Vol. 50, No. 2 , 679-698.
Brown, S. J., Goetzmann, W. N., & Ibbotson, R. G. (1999). Offshore Hedge Funds: Survival and Performance 1989-1995. Journal of Business, Vol. 72, , 91-118 .
Brown, S. J., Goetzmann, W. N., Ibbotson, R. G., & Ross, S. A. (1992 ). Survivorship Bias in Performance Studies. The Review of Financial Studies, Vol. 5, No. 4 , 553-580. .
Bu, Q., & Lacey, N. (2007). Exposing Survivorship Bias in Mutual Fund Data. Journal of
Business and Capocci, D. P. (2007). The Sustainability in Hedge Fund Performance: new insights. Working paper, HEC-ULG Management School , pp.1-21.
Cappocci, D., & Hubner, G. (2004). Analysis of hedge fund performance. Journal of
Empirical Finance , 55-89.
Carhart, M. M. (1997). On Persistence in Mutual Fund Performance. Journal of Finance,
Vol. 52, Iss. 1 , pp. 57-82.
Carpenter, J. N., & Lynch, A. W. (1999). Survivorship Bias and Attribtion Effects in
Measures of Performance Persistence. Journal of Financial Economics, Vol. 54 , pp. 337-
374.
15

Edwards, F. R., & Caglayan, M. O. (2001). Hedge Fund Performance and Manager Skill.
The Journal of Future Markets , Vol. 21, No. 11, pp. 1003 .
Elton Edwin J., M. J. (1996). Survivorship bias and Mutual Fund Performance. The Review of Financial Studies, Vol. 9, No. 4 , pp. 1097-1120.
Fung, W., & Hsieh, D. A. (1997). Empirical Characteristics of Dynamic Trading Strategies:
The Case of Hedge Funds. The Review of Financial Studies, Vol. 10, No. 2, , pp. 275 - 302.
Fung, W., & Hsieh, D. A. (2002). Hedge-Fund Benchmarks: Information Content and
Biases, . Financial Analysts Journal, Vol. 58, No.1
, pp. 22-34.
Fung, W., & Hsieh, D. A. (2000). Performance Characteristics of Hedge funds and
Commodity Funds: Natural vs. Spurious Biases. Journal of Financial and Quantitative
Analysis, Vol. 35, No.3
, pp. 291-307.
Grinblatt, M. a. (1989). Mutual Fund Performance: An Analysis of Quarterly Portfolio
Holdings. Journal of Business, Vol. 62, No. 3 , pp. 393-416.
Liang, B. (2000). Hedge Funds: The Living and the Dead. Journal of Financial and
Quantitative Analysis , Vol. 35, No. 3, pp. 309-326.
Lyman. (1999). Understanding Statistics.
Boston, USA: PWS-KENT Publishing Company.
Malkiel, B. ( 1995). Returns from investing in equity mutual funds 1971 to 1991. Journal of
Finance, Vol.50
, pp. 549-572.
Ott, L., & Mendenhall, W. (1990). Understanding Statistics. 309-311.
Sheridan, J. (2007). SPSS 14.0 for Window - Analysis Without Anguish.
Australia: John
Wiley and Sons Australia Ltd.
Steri, R., Giorgino, M., & Viviani, D. (2009). The Italian Hedge Fund Indusry: An
Empirical Analysis of Performance and Persistence. Journal of Multinational Financial
Management , Vol. 19, pp. 75-91.
16