Forecasting Iranian Inflation Using Non- Linear Univariate and Time-Varying Parameter Models

Nowadays in economic policy making literatures, price stability is considered as the main purpose of monetary policy. Based on exact inflation forecasts, one can performpolicies that provide liquidity for production sector and control price volatilities. Occasional breaks and regime changes in macroeconomic time series are importantproblems in forecasting. When confronting such problems, applying non-linear and time-varying parameter models are more suitable. In this paper, the performance oftime-varying parameter model (TVAR) and common non-linear models (TAR and STAR) relative to benchmark Auto-Regressive model are evaluated. Using quarterlyCPI data during 1369:2 to 1390:4, out of sample forecast results show that TAR and TVAR models have better performance of one to four seasons compared tobenchmark model in all forecast horizons.

کلیدواژه ها:

Forecasting ، Inflation ، Non-linear Model ، Time-Varying Parameter Model