Political Periods, Institutional Quality, and Economic Growth in Afghanistan (1996–2024): Historical-Period Associations and Univariate Mean-Break Diagnostics

Abstract

This study evaluates whether historically assigned political periods coincide with statistical changes in Afghan macroeconomic series and whether the estimated period growth association survives reasonable econometric sensitivity checks. The author’s 1996–2024 annual archive (N = 29), two near-identical Stata files and an archived Stata output log are cross-checked. The five-period ordinary least squares (OLS) model reproduces the archived joint restriction test, F(4, 19) = 3.52, p = .0261, although all individual period coefficients are imprecise. Newly calculated heteroskedasticity-consistent Wald tests yield p = .1203 (HC1) and p = .3004 (HC3), a one lag HAC test yields p = .0293, illustrating covariance sensitivity. Removing 2021 gives p = .1186; moving the final-period boundary to 2022 gives p = .2268. Archived univariate mean-break screens locate tax-revenue breaks near 2001 and 2008, whereas the GDP growth candidate near 2019 is supported at 10% but not 5%. A separate constrained least squares enumeration reproduces the break date minimum residual sum of squares partitions but not the xtbreak test distributions. The results distinguish historical categories, observational conditional contrasts, and univariate changes in means; they do not identify causal political transition effects. Original source vintages and treatment of early zero coded observations remain unverified.