A Nonlinear Analysis of Fiscal Solvency in Turkey: MLSTAR Multi Logistic Smooth Transition Autoregressive Model

44301

In economics literature, fiscal discipline is not only important in light of fiscal policy, but, it also deserves a central role in the success of monetary policies. In this study, the methodology to test fiscal dominancy focuses on the nonlinear Multi Logistic Smooth Transition Autoregression model (MLSTAR), which is different than two regime STAR models by allowing modeling fiscal variables with more than two regimes; and also different than the mainstream linear stationarity testing approach of Hamilton and Flavin (1986), Hakkio and Rush (1991) and Trehan and Walsh (1988). For the modeling steps of MLSTAR model, we generalized the two regime STAR type nonlinearity tests proposed among many by Teräsvirta (1994) based on Luukkonnen et al. (1988) in order to model STAR type additive nonlinearity with more than two regimes. In the empirical section, linear stationarity is rejected for interest payments of domestic and foreign debt ratio by ADF, PP and KPSS unit root tests. Futher, nonlinear KSS unit root test suggests a nonlinear unit root in the series. For the sample and variables analyzed, among nonlinear models, MLSTAR model is accepted vis-a-vis the nonlinear LSTAR model in light of forecast accuracy in terms of MSE, MAE, RMSE error criteria and Diebold-Mariano equal forecast accuracy tests. Accordingly, in addition to accepting the threshold effects and asymetric adjustment between regimes in interest payments on debt series, which deserves special attention on the success of budget policies; the fiscal dominance and high costs on the public debt in Turkey put significant burden on the achievement of fiscal solvency especially before and during the economic crises periods.

41

Nonlinear time series , fiscal discipline , STAR models

Ege Academic Review (EAR)

58

Ege Academic Review (EAR)