Table 3

Panel data diagnostics

TestNull hypothesisTest statisticp-valueInference
Hausman test (FE vs. RE)RE is consistent and efficientχ2 = 547.330.000FE required; RE inconsistent
Wooldridge test for autocorrelationNo first-order serial correlationF = 72.260.000Autocorrelation present
Modified Wald test for heteroskedasticityHomoskedastic disturbances across panelsχ2 = 393,739.890.000Heteroskedasticity present
Pesaran test for cross-sectional dependenceNo contemporaneous correlation across panels77.480.000Cross-sectional dependence present

Note(s): Diagnostics justify firm and year fixed effects with Driscoll–Kraay standard errors (max lag = 2) as the primary estimator. Benchmark FE with firm-clustered SEs and RE–GLS are reported only for comparison in baseline results

Source(s): Created by authors

or Create an Account

Close subscription notice
Close access options