Modeling Gasoline Consumption Behaviors in Iran Based on Long Memory and Regime Change

In this study, for the first time, we model gasoline consumption behavior in Iran using the long-term memory model of the autoregressive fractionally integrated moving average and non-linear Markov-Switching regime change model. Initially, the long-term memory feature of the ARFIMA model is investigated using the data from 1927 to 2017. The results indicate that the time series studied has a long-term memory. Therefore, after this step and determining the autoregressive lag (AR) and moving average (MA) values, the demand for gasoline in the Iranian economy is estimated using ARFIMA model (1.0.28.2). We also estimate gasoline consumption in Iran using Markov-Switching model, with the MSH model based on the lowest Akaike with 3 regimes and 2 lags. Finally, for modeling gasoline consumption behaviors, the Markov-switching model based is superior to the ARFIMA model.  Our findings indicate that if we do not use the most appropriate model for estimating future demand for gasoline, policy making in this area will not be optimal. Keywords: Long-term memory, Regime change pattern, Modeling gasoline consumption. JEL Classification: C53, Q47, C58 ,Q41