Afghanistan; political periods; economic growth; structural breaks; heteroskedasticity-robust inference; influential observations; data provenance.

https://doi.org/10.5281/zenodo.23859163

Abstract

This article directly examines whether three governance dimensions show distinguishable tax–growth interactions in Afghanistan’s annual 1996–2024 archive (N = 29). Archived single-component OLS regressions are independently reproduced: tax × Government Effectiveness = 4.6158 (p = .221), tax × Rule of Law = 9.7837 (p = .371) and tax × Control of Corruption = −0.6010 (p = .940). A new matched, jointly estimated specification standardizes all three components to their observed one-standard-deviation scales while preserving the same years, FDI, trade and population controls. A conventional test for equal standardized interactions yields F(2,18) = 0.232 (p = .795); an HC3 test yields p = .764. A joint test that all three interaction coefficients equal zero also fails to reject (p = .784 conventional; p = .750 HC3). High regressor dependence and very small residual degrees of freedom limit interpretability. The dynamic ARDL(1,0,0,1,0) from 1997–2020 (N = 24) is contextual, not a component moderation test. The new direct contrasts do not support ranking governance components or claiming fiscal-growth policy effects; identification limits, WGI-vintage uncertainty, and overlap with previously published research are acknowledged.