A0607
Title: Simple and trustworthy asymptotic $t$ tests in difference-in-differences regressions
Authors: Cheng Liu - Wuhan University (China) [presenting]
Yixiao Sun - UC San Diego (United States)
Abstract: Two asymptotically valid $t$ tests in a difference-in-differences (DD) regression are proposed when the number of time periods is large while the number of individuals can be small or large. Each of the two $t$ tests is based on a special heteroscedasticity and autocorrelation robust (HAR) variance estimator that is tailored towards the inference problem in the DD setting. The difference between the two $t$ tests is that one is based on the sandwich variance estimator of a general form while the other is based on the sandwich variance estimator of a special form. By capturing the estimation uncertainty of the HAR variance estimators, both $t$ tests have more accurate size than the corresponding normal tests. They are also as powerful as the latter tests. Compared to the nonstandard tests that are designed to reduce the size distortion of the normal tests, the proposed $t$ tests are as accurate but are much more convenient to use, as critical values are from the standard $t$ table.
