Monday, June 13, 2016

Understanding Gasoline Price Dispersion



One sentence summary: Refinery-specific costs, which have been ignored in the literature due to using local data sets, contribute up to 33% to the gasoline price dispersion within the U.S..



Abstract
This paper models and estimates the gasoline price dispersion across time and space by using a unique data set at the gas-station level within the U.S.. Nationwide effects (measured by time fixed effects or crude oil prices) explain up to about 51% of the gasoline price dispersion across stations. Refinery-specific costs, which have been ignored in the literature due to using local data sets within the U.S., contribute up to another 33% to the price dispersion. While state taxes explain about 12% of the price dispersion, spatial factors such as local agglomeration externalities, land prices, distribution costs of gasoline explain up to about 4%. The contribution of brand-specific factors is relatively minor.


Non-technical Summary
Retail prices of gasoline are considerably different across gas stations within the U.S. where consumer expenditure share of gasoline is about 5%. Since such price differences may be reflecting the frictions in the economy, understanding the reasons behind them is the key to an optimal policy that would improve the level and distribution of individual welfare.

For example, consider a typical day in 2010-2011 when the retail-level gasoline price difference between any two gas stations within the U.S. was as high as $2.25 (followed by $2.19) per gallon of regular gas. The gasoline price differences of both $2.25 and $2.19 were between Washington D.C. and Michigan on October 23rd, 2010 and December 16th, 2010, respectively. If you think that this price dispersion was due to differences in state-taxes per gallon, which ranged between 46.6 cents (for California) and 8 cents (for Alaska) in 2010 and between 49.6 cents (for Connecticut) and 8 cents (in Alaska) in 2011, you are only partially right, because, for a typical day of the very same sample period, the price difference between any two gas stations within any given state/district of the U.S. was as high as $1.57 (for Washington D.C. on October 15th, 2010) followed by $0.99 (for Iowa for October 10th, 2010).

The gasoline price dispersion was not due to outliers, either; because, according to Figure 2, for a typical day, the median price difference between any two gas stations within any given state/district of the U.S. was as high as $0.60 (for Hawaii) followed by $0.44 (for Washington D.C.) and by $0.41 (for Wyoming):


Therefore, gasoline price differences not only exist across states but also within states; accordingly, a detailed spatial analysis (rather than a state-level analysis) is required to understand the details behind gasoline price dispersion.

We use a daily gasoline price data set covering the period between September 10th, 2010 and January 31st, 2011 that involves brand information of gas stations as well as their location information at the exact address level. Combining this data set with the exact (address-level) location information of oil refineries, state-level taxes, crude oil prices, land prices (at the zip-code level), and local agglomeration externalities of spatial gasoline demand (namely, the distribution of nighttime lights data across space), we decompose the effects of crude oil prices, refinery costs, distribution costs, brand-specific costs, state taxes, land prices, and local agglomeration externalities (i.e., spatial demand and number of competitors) through a spatial analysis that models and estimates the transportation needs of individuals and distribution of gasoline from refineries to gas stations. While the crude oil prices (or any other time-varying effect that is common across gas stations) are considered to capture time-varying effects that are common across all gas stations within the U.S., other cross-sectional variables are considered to capture the effects due to spatial factors.

Estimation results show that local agglomeration externalities of spatial gasoline demand (measured by the standard deviation of nighttime lights) and the number of local competitors have negative and significant effects on gasoline prices as expected by the model. In terms of their magnitudes, one percent of an increase in local externalities can lower gasoline prices as much as 1.92%, while one percent of an increase in the number of competitors can lower gasoline prices as much as 0.19%. Distribution costs of gasoline from the nearest refiner to the gas station have positive and significant effects, also as expected; one percent of an increase in distribution costs leads to an increase up to about 0.41% in gasoline prices. The effects of land prices are also positive and significant (as expected) where one percent of an increase in land prices leads to an increase up to about 0.79% in gasoline prices. Also considering the unit coefficients in front of state and federal taxes (through restricted least squares), the R-squared values are all about 0.90, which is promising.

The corresponding variance decomposition of gasoline prices across time and space suggests that the highest contribution is by time fixed effects (or crude oil prices) capturing up to 51% (or 39%) of the price dispersion (i.e., nationwide effects are almost half of the overall effects), followed by refinery-specific costs capturing up to 33% of the price dispersion, state taxes capturing up to 12% of price dispersion, and spatial factors (such as local agglomeration externalities, land prices, distribution costs of gasoline) capturing up to 4% of price dispersion. The contribution of brand-specific costs is relatively minor.


The main contribution to the existing literature has been to show that refinery-specific costs explain a big part of the price dispersion across gas stations, which has been mostly ignored in the literature due to using local data sets covering either a couple of cities or states within the U.S..