Examining the Asymmetric Behavior across the Phases of Capacity Utilization Rates in Turkey

This is the first study that explicitly models and characterizes the asymmetric state dependent dynamics of the Turkish capacity utilization rates of the manufacturing industry. The analysis employs hidden Markov models to the mean and variance that are robust to potential structural breaks. The model parameters are estimated using EM algorithm together with the nonlinear filter to provide the maximum likelihood estimates without imposing any a priori restrictions on model parameters and infer the states through statistical estimation. The results reveal nonlinearity, determine the number of regimes and identify the nonlinear heteroscedasticity across different phases. The paper provides the smoothed probabilities of low, moderate and high capacity utilization regimes along with the fitted values for the Turkish capacity utilization rates. A three state specification with regime-dependent mean and variance dynamics is identified. The study also employs the estimated transition probabilities and determines the durations and persistence of staying in each particular regime for the capacity utilization rate fluctuations of the manufacturing industry in Turkey.