Recent developments in inference: practicalities for applied economics

The past two decades have seen phenomenal growth in applied microeconomic research. While much of the research has focused on generating unbiased estimators, there has also been substantial advances in statistical inference. We synthesize these recent advancements in calculating standard errors and test statistics for hypothesis testing in order to provide guidance for applied economists. In this chapter, we make three points. First, the applied economist needs to clearly articulate the challenges to inference that are present in the data and the source of those challenge. Second, modern computing power and statistical software means that applied economists have no excuse for not correctly calculating their standard errors. Third, it should become standard practice to rely on asymptotic refinements to the distribution of an estimator or test statistic, presenting appropriately calculated bootstrapped critical values for hypothesis testing in the main results, and not hypothesis tests that rely on first-order asymptotic theory.