Robust Standard Errors for Panel Regressions with Cross-Sectional Dependence
Daniel Hoechle. University of Basel. The Stata Journal (2007) 7, Number 3, pp. 281–312
Wednesday 14 June 2023, by Carlos San Juan
This paper may be useful for a student dealing with panel data using Stata as describes the xtscc program syntax to estimate Driscoll and Kraay (1998) standard errors
1 Introduction
In social sciences, and particularly in economics, analyzing large-scale microeconometric panel datasets has become common. Compared with purely cross-sectional data, panels are attractive since they often contain far more information than single cross-sections and thus allow for an increased precision in estimation. Unfortunately, however, actual information of microeconometric panels is often overstated since microeconometric data are likely to exhibit all sorts of cross-sectional and temporal dependencies.
In the words of Cameron and Trivedi (2005, 702), “NT correlated observations have less information than NT independent observations.” Therefore, erroneously ignoring possible correlation of regression disturbances over time and between subjects can lead to biased statistical inference.
To ensure validity of the statistical results, most recent studies that include a regression on panel data therefore adjust the standard errors of the coefficient estimates for possible dependence in the residuals.
However, according to Petersen (2007), many recently published articles in leading finance journals still fail to adjust the standard errors appropriately. Furthermore, although most empirical studies now provide standard error estimates that are heteroskedasticity- and autocorrelation consistent, cross-sectional or “spatial” dependence is still largely ignored.
http://fmwww.bc.edu/repec/bocode/x/xtscc_paper
doi.org/10.1177/1536867X0700700301
Also, downloadble below
Attached documents
-
I present a new Stata program, xtscc, that estimates pooled ordinary least-squares/weighted least-squares regression and fixed-effects (within) regression models with Driscoll and Kraay (Review of Economics and Statistics 80: 549–560) standard errors. By running Monte Carlo simulations, I compare the finite-sample properties of the cross-sectional dependence–consistent Driscoll– Kraay estimator with the properties of other, more commonly used covariance matrix estimators that do not account for cross-sectional dependence. The results indicate that Driscoll–Kraay standard errors are well calibrated when cross-sectional dependence is present. However, erroneously ignoring cross-sectional correlation in the estimation of panel models can lead to severely biased statistical results. I illustrate the xtscc program by considering an application from empirical finance. Thereby, I also propose a Hausman-type test for fixed effects that is robust to general forms of cross-sectional and temporal dependence.
RSS 2.0