Evaluating the size of the bootstrap method for fund performance evaluation

•Gauge the size of bootstrap methods with and without cross-sectional dependence.•Without cross-sectional dependence, the size of the fund-by-fund bootstrap is good.•With cross-sectional dependence, the fund-by-fund bootstrap is biased.•We propose a new panel bootstrap model with unobservable interactive effects.•Caution to use the bootstrap to distinguish skills from luck in the future.

We investigate the validity and reliability of the bootstrap approach in fund performance evaluation by gauging the size. Monte Carlo simulations suggest that cross-sectional dependence may alter the size of this test and we propose a new panel bootstrap approach.