Regional Gasoline Price Dynamics

Abstract

A large literature has argued that gasoline prices respond more rapidly to increases in oil prices than to decreases in oil prices. Moreover, some of this literature has found heterogeneous asymmetry in gas price responses across cities. Here, we reconsider the causes of heterogeneous asymmetric pass-through. Consistent with the previous literature, we find heterogeneity in the magnitudes of asymmetric pass-through across cities. We also find a large number of cities that exhibit no asymmetries. We then examine whether heterogeneous asymmetry results from city-level differences in (i) the demand for gasoline, (ii) the supply of gasoline (proxied by the distance from Cushing, Oklahoma), or the cities' fiscal environments (proxied by the level of taxes). We examine whether these characteristics affect either the magnitudes of the asymmetries or the presence of asymmetries. While we find that city-level characteristics cannot (robustly) explain variation in the magnitudes of the asymmetries, they do seem to affect the probability that a city experiences asymmetric pass-through.

Introduction

Oil shocks have been important factors in most postwar recessions (Hamilton, 2011). One of the primary channels through which oil price shocks are felt by consumers is through their effect on gasoline prices. Similar to other prices and exchange rates, pass-through indicates the degree to which oil price fluctuations affect gasoline prices. The extant literature has identified a number of asymmetries in the pass-through of oil price shocks to gasoline prices, with particular focus on the hypothesis that gasoline prices respond more rapidly to increases in oil prices and less rapidly to decreases in oil prices.

The previous literature finds mixed evidence of asymmetry in the gasoline market, with variety in the type of asymmetry, the level of geographic granularity (e.g., national versus regional), and the time sample and frequency of data used. Much of the extant literature focuses on the asymmetric pass-through of input costs to retail gasoline prices at the U.S. national level. Most studies find asymmetry at the national level (e.g., Karrenbrock, 1991, using monthly data on wholesale to retail pass-through; Borenstein, Cameron and Gilbert [BCG], 1997, using semi-monthly crude oil to retail pass-through; Balke, Brown, and Yucel, 1998, using various weekly intermediate good prices; and Chen, Finney, and Lai, 2005, using spot and future oil prices). A few papers, however, find no evidence of asymmetry (Bachmeier and Griffin, 2003, using daily crude oil to wholesale gasoline prices; and Douglas, 2010, using weekly input costs to retail gasoline prices).

At the subnational level, evidence of asymmetric pass-through is also mixed. For example, Deltas (2008) finds evidence of asymmetric pass-through from wholesale gasoline prices to retail gasoline prices at the state level. States with higher average retail-wholesale margins have higher degrees of asymmetry, suggesting that sticky prices and asymmetric price responses in the gasoline market may be in part due to retail market power. Ye et al. (2005) also find asymmetries in the pass-through of wholesale gasoline prices to retail prices at the Petroleum Administration for Defense District level. Adilov and Samavati (2009) find heterogeneous asymmetries between weekly oil and gasoline prices: Gasoline prices in California, Texas, and Washington react faster to oil price increases than decreases, while gasoline prices in Massachusetts, Minnesota, and Ohio react faster to oil price decreases. Chesnes (2016) documents the largest degrees of asymmetry in Louisville, Minneapolis, Cleveland, and Detroit—Midwestern cities—and the smallest degrees in San Francisco and the West Virginia suburbs of Washington, DC.

Research examining the factors underlying asymmetric pass-through largely focuses on market power (Verlinda, 2008; Tappata, 2009; Radchenko, 2005; and Hong and Lee, 2020, for Korea) or consumer search costs (Davis, 2007; Lewis, 2011; and Yang and Ye, 2008). Remer (2015) determines that premium gasoline prices fall more slowly than regular fuel prices but rise at the same speed, supporting the theory that asymmetry occurs as a consequence of firms extracting informational rents from customers with positive search costs. Verlinda (2008) finds that factors associated with gasoline demand have little to no effect on the degree of asymmetry.

We evaluate asymmetric pass-through from oil prices to pre-tax gasoline prices at the metropolitan stastical area (MSA) level and consider the causes of heterogeneous pass-through across MSAs. Variation in the price of gasoline across locations has been well documented, but these differences have often been attributed to differences in state and local taxes. However, we find evidence of some remaining pre-tax heterogeneity. We test various hypotheses about the cause of regional variations in the pass-through of oil price shocks to gasoline prices. In particular, we consider geographic and demand issues. We estimate a hierarchical model in which the degree of pass-through is determined by city-level characteristics that reflect the demand for gasoline (e.g., the average commute time per worker) and the elasticity of the supply of gasoline (e.g., the distance from Cushing, Oklahoma).