Diesel prices in Brazil: A dynamic fractional integration analysis

Abstract

We investigate the behaviour of retail diesel prices in Brazil using fractional integration, with weekly data from January 2010 to April 2020. In this period, we have 3 episodes of relevant economic implications in the country under analysis: i) impeachment of the Brazilian President;ii) the lorry-drivers' strike;iii) the rise of the global Covid-19 epidemic. We use a sliding windows approach to analyze price dynamics over time. The results suggest that, at the beginning of the sample, prices were non-stationary and non-mean reverting. Over the time, diesel prices become non-stationary with mean-reversion in Midwest, South and Southeast regions, while in the North and Northeast we cannot reject non-stationarity and non-mean reversion (d > 1). Results are relevant for market agents and policy-makers, as it can be inferred whether exogenous shocks are temporary, despite taking some time to dissipate completely.