Ciclos de Negócios em uma perspectiva não linear: modelo auto-regressivo de “transição suave” para o Índice Geral de Produção Industrial Brasileiro e Bens de Capital

Abstract

The aim of this paper is to investigate the aspects and consequences of nonlinearities in the Brazilian General Production Index and Capital Goods Production Index, which can be adequately described by the Smooth Transition Autoregressive (STAR) models. This family of models embodies an asymmetric behavior that allows the business cycles to alternate endogenously between two distinct regimes. Moreover, a test is performed in order to distinguish between nonlinearities and additive outliers (AO). The results are interpreted by means of the characteristic polynomial roots, which provide important information about the dynamic proprieties, and by deterministic extrapolation of the estimated STAR model. The main finding is that the general production and capital goods production moves from deep recession into high growth very aggressively.