Researchers use difference-in-differences models to evaluate the causal effects of policy changes. As the empirical correlation across firms and time can be ambiguous, estimating consistent standard errors is difficult and statistical inferences may be biased. We apply an approximate permutation...