Abstract
The efficient market hypothesis states that past performance should be no predictor of future performance. I test this using free from survivorship bias, a sample of all New Zealand unit trusts in existence from 1987-1998. Almost no evidence of performance persistence is found over this period, suggesting that the efficient market hypothesis does hold for my sample. In the 1997-1998 period, marginal evidence of persistence was found, but none for other periods. I have focused on the top performing unit trusts to establish if it is possible to pick a winning trust to invest in based on past performance. Different categories of New Zealand unit trusts are incorporated in this study, and no category appears to have significantly better or worse returns than the others, and most unit trusts perform better than the 'market', on average.