| Summary: | 碩士 === 亞洲大學 === 國際企業學系碩士班 === 94 === The purpose of this research is first focused on comparing the performance of mutual funds’ investment approaches, the dollar-cost averaging and lump-sum approaches. Moreover, this study attempts to test whether the trend time is effective or not, where the trend time is often used by dollar-cost averaging approach and suggests that the investment should be done when the times surpass 1. The sample is 229 equity funds. The research periods are from January in 2000 to April in 2006.
This study collects the net asset value of funds on the first trading day every month. The data is acquired from the mutual models in the Taiwan Economic Journal Data Bank, TEJ. I set seven different investment horizons, which is included three months, six months, one year ,…, and five years.
This study sets up different investment horizons and computes the single returns and compounded returns on fund investment for each month. For dollar-cost averaging, the returns of the free cash on time deposits are incorporated. The t- and non-parametric tests for paired observations are employed to compare the return series between two investment approaches. The performance indices included not only the original returns but also the risk-adjusted returns. No matter I use the t- and non-parametric tests for paired observations to test returns between lump sum and dollar-cost averaging, the finding is that lump sum probably beats dollar-cost averaging when the investment horizon is short. As the investment horizons extend, the dollar-cost averaging obviously beats lump sum. Additionally, to determine whether the trend time is effective, this study also computes the probabilities of positive returns when the trend times, calculated using past 6, 12 and 24 months’ net asset values of funds, are above 1 and are not larger than 1, respectively.
The conclusion indicates that the dollar-cost averaging works well in the long run, as we know, it is better than the lump sum. The trend time applied to the dollar-cost averaging is not effective because we can also gain considerable positive returns, even not according to the rule when the trend time is larger than 1.
|
|---|
