Improving the finite sample performance of tests for a shift in mean

•We propose bias-corrected tests for a shift in mean.•We correct the bias of the long-run variance estimator.•Bias correction is achieved by taking a structural break into account.•The proposed tests have good size and high power.

It is widely known that structural break tests based on the long-run variance estimator, which is estimated under the alternative, suffer from serious size distortion when the errors are serially correlated. In this paper, we propose bias-corrected tests for a shift in mean by correcting the bias of the long-run variance estimator up to O(1/T)O(1/T). Simulation results show that the proposed tests have good size and high power.