Robustez de regressões de crescimento frente à incerteza sobre a especificação do modelo: quão robustos são os regressores para o caso brasileiro?

Abstract

Although there is a vast literature that seeks to identify robust regressors able to influence economic growth at the international level, this issue still needs research that have satisfactory results for the Brazilian case. This article aims to contribute to the identification of robust variables that influence the Brazilian economic growth. The Resende and Figueiredo (2010) dataset is reviewed and we seek to circumvent the problem of uncertainty about the model specification. For this purpose, a Frequentist Model Averaging technique proposed in Hansen and Racine (2012) is used. This method is known by Jackknife Model Averaging and it is not as restrictive as the Extreme Bound Analysis and not as permissive as the approach of Sala-i-Martin (1997). The results suggest that, among the 22 investigated covariates, only the overall tax burden would have significant influence on growth. In general, our results remain indicating caution when working with growth regressions that use state data. Moreover, the work motivates further studies and draws attention to the importance of resuming this line of research.