Volatility of mutual fund return, GARCH modeling and value at risk

This paper examines conditional volatility through GARCH/EGARCH modeling using data on daily returns of nine mutual funds in Asia-Pacific emerging markets, and then compares the forecast performance of downside risk on four types of VaRs including conventional VaR, CF VaR, GARCH-type VaR. Empirical results show that return rates of most mutual funds in Asia-Pacific emerging markets have significant ARCH effects. However, some GARCH models of mutual funds like Korea, and Malaysia do not confirm stationarity. Following standard procedure, the study compares the forecast performance of conditional variance for GARCH model and EGARCH model and then the result shows EGARCH is superior to GARCH. The study also implemented several forecast evaluation criteria to compare and examine on which kind of VaR has the best prediction ability. Results confirm that GARCH-type VaRs have demonstrated the best performance in capturing downside risk estimate except for the Korea and Thailand funds. The study suggests that the risk-averse investors should hold mutual funds mixed with fixed income security and inter-markets portfolio so that the downside risk could be lower like the case of little Dragon funds. For risk neutral or risk favor investors, they can hold as any portfolio as they prefer, but they should make use of VaR estimates in this study to avoid potential downside risk before the bull market emerges. According to this paper, here are some findings and opinions: 1. The GARCH-type VaR are useful only for analyzing Philippine, Malaysian and Little Dragons mutual funds. Interestingly, the Normal VaR can be useful in the rest of three funds in risk management. However, in terms of performance, GARCH type Var showed the best performance relative to the other Var techniques. 2. From the results shown in figure 5.1, it seems that the GARCH-type VaR can not be completely applied to risk management of mutual funds for emerging markets in Asia Pacific. What is being suggested here is that a combination of GARCH-type VAR and normal VaR can be more useful on VaR estimation and risk management. With this idea, fund managers can engineer and control VaR and avoid unexpected losses when they are making investment decisions