Conditional assessment of the impact of a Hausman pretest on confidence intervals

TL;DR

This paper evaluates how a Hausman pretest affects confidence intervals for slopes in panel data, emphasizing the importance of conditional analysis and finite sample validity, ultimately advising against using such intervals post-pretest.

Contribution

It provides a conditional assessment of the Hausman pretest's impact on confidence intervals, highlighting limitations and offering finite sample insights.

Findings

Confidence intervals after a Hausman pretest are unreliable.

Conditional analysis reveals the pretest's adverse effects.

Finite sample results support avoiding post-pretest intervals.

Abstract

We assess the impact of a Hausman pretest, applied to panel data, on a confidence interval for the slope, conditional on the observed values of the time-varying covariate. This assessment has the advantages that it (a) relates to the values of this covariate at hand, (b) is valid irrespective of how this covariate is generated, (c) uses finite sample results and (d) results in an assessment that is determined by the values of this covariate and only 2 unknown parameters. Our conditional analysis shows that the confidence interval constructed after a Hausman pretest should not be used.