A simple and trustworthy asymptotic t test in difference-in-differences regressions

Abstract

We propose an asymptotically valid t test that uses Student's t distribution as the reference distribution in a di¤erence-in-di¤erences regression. For the asymptotic variance estimation, we adopt the clustering-by-time approach to accommodate cross-sectional dependence. This approach often assumes the clusters to be independent across time, but we allow them to be temporally dependent. The proposed t test is based on a special heteroscedasticity and autocorrelation robust (HAR) variance estimator. We target the type I and type II errors and develop a testing-oriented method to select the underlying smoothing parameter. By capturing the estimation uncertainty of the HAR variance estimator, the t test has more accurate size than the corresponding normal test and is just as powerful as the latter. Compared to the nonstandard test developed in the literature, the standard t test is just as accurate but much more convenient to use. Model-based and empirical-data-based Monte Carlo simulations show that the t test works quite well in …nite samples.

Effects of Smart City Policies on Green Total Factor Productivity: Evidence from a Quasi-Natural Experiment in China

“…The DID method has become increasingly popular in policy analysis [48]. It studies the differential effect of a policy that does not affect everybody at the same time and in the same way on a treated group versus a control group [49].…”

An assimilation of TRIZ in dissecting the statistical outcomes of tactile sensitivity, pinch force and endurance among elderly people

“…With the assumption of homogeneity in variance, the pooled estimate for the error term for the t-statistic can be used. The t-test was used as it is more accurate than a corresponding normal test and is just as powerful (Liu & Sun, 2019). The t-test can determine if a significant difference between the means of two groups existed.…”

Asymptotic F tests under possibly weak identification

“…The data for our experiments are generated as follows. Given the model parameters 0 and 0 ; we generate c t and d t according to the Gaussian VAR(1) model in (17) and then compute v t = v (c t ; d t ; 0 ; 0 ) ; which is the solution to the Euler equation in (16). With fc t ; d t ; v t g T t=1 ; we compute the real returns of the asset…”

“…In the econometrics literature, the orthonormal series LRV estimator has recently been used by, for example, Phillips (2005), Müller (2007), Sun (2011, 2013, 2014a,b,c), Liu and Sun (2019), and Lazarus, Lewis, Watson (2016, 2018).…”