Composite versus Disaggregated Governance and Sustainable Development: A Cross-Country Panel Analysis of Multicollinearity and the Apparent Sign Reversal of Regulatory Quality (2016–2023)

Abstract

Cross-country research on governance and sustainable development frequently draws on the Worldwide Governance Indicators (WGI) and increasingly seeks to identify the distinct contribution of regulatory quality. This study examines whether such disaggregation is statistically advisable using a country-year panel of 160 countries observed during 2016–2023 (N = 1,161). Two measurement strategies are compared on the same dataset: a composite governance index that averages four WGI dimensions (Strategy A) and a specification that enters the four dimensions individually (Strategy B). The four WGI dimensions are correlated between 0.71 and 0.94, and the variance inflation factors for Government Effectiveness, Rule of Law, and Regulatory Quality exceed conventional thresholds in Strategy B. Under Strategy A, the composite governance index is positively and significantly associated with Sustainable Development Goals (SDG) performance in pooled and random-effects specifications, consistent with prior literature. Under Strategy B, however, the pooled-OLS coefficient on Regulatory Quality reverses sign and is negatively associated with SDG performance, while Government Effectiveness retains a large positive coefficient. Quantile regressions indicate that the negative coefficient on Regulatory Quality is not robust across the SDG distribution and loses statistical significance at most quantiles, and the negative coefficients do not survive estimation with the Hausman-preferred fixed-effects model. The study interprets these patterns as a measurement artefact rather than as substantive evidence that regulatory quality reduces SDG performance, and recommends that cross-country analyses of governance and sustainability either rely on a composite governance index or formally diagnose multicollinearity before assigning substantive meaning to the coefficients of individual WGI dimensions. The contribution is offered as a methodological caution for associational, non-causal governance and sustainability measurement research, where dimension-specific claims about regulatory quality are increasingly common.