تجزیه و تحلیل عملکرد های صندوق های سهام بین المللی استرالیا

This paper analyses the performance of a sample of Australian international equity trusts over the period 1990–1999 using monthly data. Selectivity and timing performance is examined for a surviving sample and two alternative non-surviving samples. Additionally, alternative time frames are examined, as are various subcategories of fund classifications. Overall the results are consistent with much of the literature in that funds are unable to time the market and that there is an inverse relationship between market timing and selectivity measures. Adjustment for survivorship bias has a minimal impact. Furthermore, the results are relatively consistent between differing fund classifications; however, the time frame for the study does have an impact on our findings.

Generally, the literature on managed funds suggests their inability to outperform the market (Jensen, 1968, Chang and Lewellen, 1984, Eun et al., 1991 and Brown and Goetzman, 1995). However, at an individual level there is some evidence of superior timing and selectivity ability (Kon, 1983, Lehman and Modest, 1987, Lee and Rahman, 1990, Shukla and Trzcinka, 1994 and Daniel et al., 1997). There is also evidence of a negative relationship between timing and selectivity abilities (Coggin et al., 1993 and Hallahan and Faff, 1999). Additionally, results from a number of studies have indicated that funds have perverse market timing abilities (Kon, 1983, Chang and Lewellen, 1984 and Jagannathan and Korajczyk, 1986 (J&K), Sinclair, 1990 and Malkiel, 1995). The existence of perverse timing abilities results in a questioning of the model specification since the existence of perverse timing is irrational. J&K addressed misspecification in the Treynor and Mazuy (1966) (T&M) model. The current study extends the literature on the performance of Australian managed funds by providing an analysis of Australian international trusts.1 The sample incorporates all trusts that invested in international equities including general equity trusts, Insurance Bond Equity Trusts, superannuation equity trusts and wholesale equities. The paper focuses on the analysis of a sample representing surviving trusts; however, two non-surviving samples are also analysed and the impact of survivorship bias on results is considered. The study implements the Jensen (1968), T&M and Henriksson and Merton (1981) (H&M) models to assess selectivity and market timing abilities of unit trusts. Additionally, the specification of the models in line with J&K is examined. It is recognised that the choice of benchmark may affect the results and this limitation is noted. However, for the purposes of the current study, in order to allow us to focus on the assessment of the timing and selectivity ability of internationally invested funds, only one type of benchmark is employed throughout, namely, an international index benchmark. Our study incorporates a number of different types of funds including general, Asian, Insurance Bonds and Superannuation Funds. Kao et al., (1998) addressed varying fund classifications by applying the H&M model with different indices depending on the type of international equity funds. We apply a number of models and compare results for different categories of trusts using the same performance benchmark. This analysis allows for a greater understanding of the results as they apply to funds of different portfolio structures. Our study covers a 10-year sample period. In order to assess the robustness of the analysis it was considered appropriate to divide the sample into two time periods of 5 years each. This analysis allows a consideration of the applicability of results to different time frames. A number of studies have also considered the impact of survivorship bias. Elton et al. (1996) used a ‘follow the money approach’. Here the non-surviving fund was analysed in order to determine what happened to the fund, for example, was it taken over by a new or pre-existing fund? An average alpha for the full time frame of the study may then be calculated as an average of the alpha for the non-surviving fund and the new/takeover fund. Blake and Morey (1999) followed an approach suitable where a proportion of the funds simply cease to exist. They found an existing similar fund and used that return series to complete the returns of the non-surviving fund. Malkiel (1995), Fung and Hsieh (1997) and Liang (2000) assessed the effect of survivorship bias by comparing performance results of a full sample of all funds with a sample of surviving funds. Their findings indicated that survivorship bias does have an impact on results. We use an adaptation of these methods and consider two alternative non-surviving samples. The structure of the paper is as follows. In Section 2 the data and sample selection are outlined, Section 3 contains the methodology and Section 4 the results. A conclusion is provided in Section 5.

In this paper we investigate the timing and selectivity performance of a sample of Australian IET over the period 1990–1999. A surviving sample and two alternative non-surviving samples are examined to assess the potential impact of survivorship bias. Overall the performance results are consistent with much of the literature in that funds are unable to time the market and there is an inverse relationship between market timing and selectivity measures. An analysis of sub-periods within the sample indicates that the models are not all robust over differing time periods. However, the general conclusion on timing and selectivity abilities does not change between periods. Partitioning the sample into various subcategories does not provide any clear insight into possible reasons for the performance results. There is some variation in results when using a sample adjusted for non-survivors; although, it depends on how the adjustment was made. An approach that assumes investors move their funds to similar trusts, in the event of a trust ceasing to exist, has little effect. However, the inclusion of all funds, including those ceasing or starting sometime in the sample period, with no adjustment for missing data, does produce some different results. Even so, these differences are minimal and it is possible that this has been induced by incomplete data and/or the short time series. One interpretation of these results may be that survivorship bias is not important in the assessment of the performance of our sample international trusts as a group.