| dc.description.abstract | Fund performance is an important indicator for evaluating the abilities of fund managers. When testing fund performance, conducting multiple hypothesis tests simultaneously increases the probability of committing Type I errors, which means incorrectly concluding that the fund performance has significant excess returns. To address this issue, we employ methods that control the False Discovery Rate (FDR). Furthermore, we adopt the method proposed by Giglio et al. (2021) to address the challenges posed by omitted factors and missing data in asset pricing models. We use principal components, extracted through principal component analysis, as latent factors, combined with CAPM, Fama-French three-factor model, and Carhart four-factor model to form multifactor models. These models are used to evaluate the performance of domestic equity and balanced mutual funds in Taiwan from 2014 to 2022.We divide the funds into those with a duration of more than three years and those with more than five years for analysis. Empirical results show that, for funds with a duration of more than three years, the multifactor model combining the Fama-French three-factor model with the latent factors extracted through principal component analysis outperforms all other models. This model identifies two funds with significantly excess returns under an asymptotic distribution. For funds with a duration of more than five years, the empirical results demonstrate that the CAPM combined with the latent factors extracted through principal component analysis performs the best. Under an asymptotic distribution, this model can identify three funds with significantly excess returns. In our empirical results, out of a total of 32 models, 2 funds were found to have statistically significant positive excess returns in 28 of the models. We also considered different numbers of latent factors to confirm the robustness of our testing results. | en_US |