Abstract

This study investigates the non-linear dynamics between public expenditure and economic growth across Brazilian states from 2001 to 2021. Using a dynamic panel threshold model based on the Seo and Shin (2016) estimator, the research addresses endogeneity and identifies regime-dependent marginal effects without imposing a predetermined functional form. The empirical results indicate a non-linear relationship consistent with the BARS curve framework. For total and current expenditures, the estimated thresholds are 15.30% and 12.96% of GDP, respectively. While spending remains productive in both regimes, its marginal effectiveness is significantly reduced after crossing these thresholds, suggesting that the typical Brazilian state operates in a zone of diminishing returns. In contrast, capital expenditure exhibits an effectiveness threshold at 1.31% of GDP, below which its impact on growth is statistically negligible. These findings remain robust across alternative specifications, including a quadratic GMM model. The absence of a strictly negative marginal effect in the upper regime might be associated with the institutional constraints from the Fiscal Responsibility Law (LRF) and the high degree of heterogeneity among states, which potentially dampen the emergence of a more severe detrimental impact in the aggregate analysis.

This work is licensed under a Creative Commons Attribution 4.0 License.