Econometric modeling of the relationship among macroeconomic variables of Thailand: Smooth transition autoregressive regression model

There is evidence of non-linear in macroeconomic variables of Thailand. To understand the behavior of this economic indicators, this paper used the smooth transition autoregressive model (STAR) that was developed by the contribution of Ter?svirta and Anderson (1992). This paper used quarterly data for the period of 1997:3 to 2012:1. The linearity test shows that almost all variables have non-linear behavior with logistic STAR (LSTAR) except in the investment growth rate. The results show that GDP growth rate have suddenly moved from negative to positive on a quarterly period and most of its observation remained in the upper regime. Moreover, inflation rate, unemployment rate, and interest rate are slow in adjustment to reducing their values, and most of them that was based on observation belonged in the lower regime. For investment growth rate, this paper used linear AR(2) model to estimate the parameter and prediction. The results showed that there is tremendous fluctuate movement in the percentile of -30% to 30%.