Asymmetric Price Adjustment in the US Gasoline Industry: Evidence from Bayesian Threshold Dynamic Panel Data Models

This paper investigates the gasoline price adjustment to changes in the input cost price for a panel of 48 US states using a monthly data set covering the period 1994–2011. We build, for the first time, a non-linear threshold panel vector-error-correction model (PVECM) and propose efficient Markov chain Monte Carlo (MCMC) Bayesian techniques. Our findings indicate that states with high margin experience a slower adjustment and a more asymmetric response to input price cost shocks. Our results are robust to potential structural breaks in the threshold parameter, which is important as market conditions change over time and are very sensitive to production/consumption constraints. Lastly, we attribute fluctuations in the gasoline prices to input cost shocks, arguing that the peak responses occurring one month after the shock are short-lived.