Which is Responsible for China Open-ended Funds Performance: Chance or Security-picking Ability?

Abstract: The first investment trust in the world, the Foreign & Colonial Investment Trust, was started in 1868 by F&C in order to“give the investor of moderate means the same advantages as the large capitalists in diminishing the risk of spreading the investment over a number of stocks”. 63 years later M&G launched the first modern unit trust in the UK. Yet in China, not until the fund called Huaanchuangxin was issued on Sep.11, 2001, the history of open-ended fund of our own remained blank. However, this new kind of fund was developing by leaps and bounds, taking most places of trust units in China within only 6 years.Actually, coming along with the expansion of open-ended fund volume are not only inspired passion of investors but also increasing attention of professionals. In funds performance evaluation, scholars are always attempting to tell the most promising fund from the rest given different criteria and they have already found many interesting results. In this paper, I try to focus on a specific question that has not yet been answered: which is responsible for China open-ended funds performance, chance or security-picking ability?It is known to all that security-picking ability of funds manager is an essential factor of funds performance. The better this ability is, the more outstanding funds performance would be. But funds performance may derive from good luck other than security-picking ability. Chances are that in a bull market some managers without any genuine ability could create a portfolio that brings high yields by simply picking securities at random. This point is also indicated in the latest publication by Dr. Yizhen Feng, a researcher of TopView Finance Center. After analyzing a large number of samples that range from September 2007 to March 2008, he finds that: some large and celebrated funds corporations failed to beat the market and their actions on selling or buying securities did not have stronger bonds with the forthcoming market index than those of small funds whose performance were very amazing. Common sense tells us that large and celebrated funds corporations tend to hire more excellent managers with better security-picking ability since they could afford higher expenses and offer the managers a more attractive financial package than the small fund corporations do. So why performances supervised by excellent managers were worse than those by average ones? Maybe chance was the key.In order to answer which is responsible for China open-ended funds performance, chance or security-picking ability, my study proceeds as follows:Section 1 briefly describes the history of funds development at home and abroad and then shows the positive role performance evaluation plays in finance market.Section 2 reviews common measures such as Treynor ratio, Sharp ration, Jensenαand so on, and all kinds of static and dynamic models. Also, it predicts and analyzes two main trends in academic study.Since most contributions of foreign literature are contained in Section 2, I focus on two latest studies in Section 3. In these studies, different researchers apply the same technique, a cross-section bootstrap methodology, to funds in order to distinguish chance from security-picking ability. In the meanwhile, I introduce recent studies on China fund market and find many researches that use Meton-Heriksson model and/or Treynor-Mazuy model to evaluate managers’security-picking ability neglect the prerequisite of these models: yields of funds reflect managers’genuine ability. Actually, in many cases high yields come from good luck rather than security-picking ability. What is more, I find that sampling error due to relatively small sample size in local studies may account for most paradox conclusions on China fund industry.Section 4 introduces all the methods used in my study. After describing Fama-French three factor model and bootstrap technique, this paper lists 6 steps of combined application: Step 1:to use F-F model to calculate eachαof individual funds by regression with actual data. So we can obtain excess yield rateαi, t-statistics, residuals e?i tand some other estimated parameters. Step2-6: the bootstrap procedure is applied to construct a new distribution of t-statistics for each fund. At 90% confidence(:1) Ifα_i> 0 and t > (t|^), that is , the excess yield rateαiof this fund is greater than that of 90% of the excess yield rates constructed by 1000 bootstraps, it means performance of this fund derives from excellent security-picking ability of managers rather than good luck. Ifα_i> 0 and t < t(t|^), the funds performance is believed to come from good luck.(2) Ifα_i< 0 and t > t(t|^), that is , the excess yield rateαiof this fund is lower than that of 90% of the excess yield rates constructed by 1000 bootstraps, it means that performance of this fund is due to excellent security-picking ability of managers rather than good luck. Ifα_i< 0 and t < t(t|^), the funds performance is believed to come from bad luck.Without special note, the data range in my study is from February 27, 2004 to August 31, 2007. Section 3 describes how indexes, including SMB, HML, weekly return rates are obtained based on data from Wind database, Shengyin & Wanguo index series, Hushen 300 index of stock market and S&P/CITIC Composite Bond Index.Section 6 firstly applies Augmented Dickey Fuller test to all the data and find that they are all stable time series. This result guarantees the effectiveness of F-F model in performance evaluation. Then the outcome of JB test applied on series of residuals indicates that t-test, the common technique to measure statistical significance of Jensena , is ineffective for over 80% of China funds, calling for a better measure to evaluate funds performance. After comparing Jensen a’s t-statistics with t? -statistics, which is constructed by 1000 bootstraps, my study reveals that chance, namely, aggregate effects of uncontrollable factors in the market rather than genuine security-picking ability, contributes to most China funds performance whether better or worse. This finding along with insignificantαin Fama-French three factor model proves that majority of fund managers in China have no better ability of picking security than benchmark although statistically significantαfrom a relatively small size of funds indicate a handful of managers do have genuine picking security skills. Then my paper compares the performances of Growth Fund and Growth-Income fund, and the result is that the performance of Growth Fund is much more dependent on chance than that of Growth-Income fund. In the end, this section analyzes the persistence of China funds by cross-section and reveals the fact that the persistence of funds weekly return is statistically significant.Section 7 concludes the whole study and shows future prospects.The main contributions of this paper include not only application of new techniques but also some improvements to recent researches as following:First of all, the sample size in local investigations is too small to represent the whole funds market in China. As a result, sampling error may explain most paradox conclusions among recent studies. In my study, data contain the information of nearly all the funds and its wide time range guarantees effects of conclusions.Secondly, even though most scholars substitute weekly data for monthly ones, the residuals of performance model usually fail to pass the test of normality, causing inaccuracy in evaluation. My study applies bootstrap statistical technique to construct a new t-distribution. So no matter distribution of residuals is normality, we can make an accurate evaluation…
Key words: performance evaluation; security-picking ability; Fama-French model; bootstraps