Files in this item



application/pdfMoonsik_Kim.pdf (2MB)
(no description provided)PDF


Title:Essays on market structure and price in the Korean retail gasoline market
Author(s):Kim, Moonsik
Director of Research:Hong, Seung-Hyun
Doctoral Committee Chair(s):Hong, Seung-Hyun
Doctoral Committee Member(s):Cho, In-Koo; Deltas, George; Chung, Eun Yi
Department / Program:Economics
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):Market Structure
Korean Retail Gasoline Market
Instrumental Variable
Population Density
Density of Diesel Cars
Structural Model
Merger Simulation
Berry, Levinsohn & Pakes (BLP) model
Brand Change
Ownership Change
Contractual type between refiners and gas stations
Vertical relationship
Abstract:This study empirically examines the relationship between market structure and prices in the Korean retail gasoline market. Korea has experienced big changes in the number of gas stations. For example, nationally the number of gas stations increased 24.9% from 10,406 in 2001 to 13,003 in 2010, but in Seoul the number of stations dropped 24.9% from 816 in 2001 to 613 in 2013. In the literature, the empirical research about the relationship between the market structure and prices in the retail gasoline market has been done in two ways. First, studies with regression models have mainly focused on finding market structure as the determinant of retail prices. Second, a few studies with structural models have assessed the impacts of mergers on prices. However, previous studies had some limitations: (1) most regression analyses did not evaluate the long-term impact because they used cross-sectional data or short-term panel data; more importantly, they did not successfully correct an endogeneity problem because they used a controversial instrumental variable; (2) because quantity data at the station level were rarely accessible, it was not easy to utilize a structural model. I try to overcome these limitations of the previous studies in two ways. In the first chapter, I run a regression with long-term panel data and a new instrumental variable which has never been used in the literature. In the second chapter, I estimate a structural model without sales data from gas stations to evaluate the impact of changes to the market structure on prices and welfare. In Chapter 1, I employ monthly data between January 2003 and December 2011 from seven big cities in Korea to estimate the effects of station density on average retail prices and average sales per station. Instead of population density, which has been commonly used in previous studies, I use the density of diesel cars as the new instrumental variable. The density of diesel cars is a superior instrument to population density in two respects. First, the correlation between station density and the new instrument is obviously expected to be higher than the correlation between station density and population density, because demand for stations depends on the number of cars rather than the number of people. Second, population density may well represent demand for gasoline and therefore have a direct impact on the retail price. Meanwhile, the density of diesel cars satisfies instrument exogeneity because it is not a determinant of gasoline price. Estimation results yield the following findings. First, regarding a price equation, an OLS estimate of station density is negative but statistically insignificant. However, an IV estimate of station density is negative and statistically significant. The results demonstrate that the OLS estimate underestimates the impact of station density on prices. The IV estimation results show that a 10% increase in the number of stations per square kilometer is associated with a 0.68-0.95% decrease in retail prices. Second, regarding a sales equation, the OLS and IV coefficient estimates of station density are negative, statistically significant and very similar in magnitude. A 10% increase in station density is associated with a 4.2-5.9% decrease in station sales. These findings also imply that the number of stations is an endogenous variable with respect to price, but not with respect to sales. This is the same as the result of Sen and Townley (2010), who examined the retail gasoline industry in Canada. In Chapter 2, with June 2009 data from 270 gas stations located above the Han River in Seoul, Korea, I use a structural model and conduct counterfactual experiments to estimate the effects of market structure on prices and welfare. My research is different from previous studies. I estimate the model without quantity data from the gas stations by employing the idea of Thomadsen (2005), while previous studies used sales data by directly following the work of Berry et al. (1995). In general, it is difficult to obtain quantity data from gas stations because they keep them secret. Also, data richness allows me to introduce different contractual forms between refiners and stations in the supply model, while previous studies assumed because of data limitations that there was only one type of vertical relationship. The counterfactual experiments yield the following results. First, although the change to company-owned GS (or HD) stations increases their prices, the prices of company-owned SK stations decrease, decreasing average price. Consumer welfare also decreases because the base utility of GS (HD) stations is smaller than that of SK stations. Second, the change to non-company-owned SK stations decreases the average price because the decreasing effects of ownership changes on prices outweighs the increasing effects of change in vertical contracts on marginal costs, which increases consumer welfare. Third, the change to non-company-owned S-Oil stations greatly lowers the average price mainly because the refiner’s wholesale price is the lowest among all refiners. Finally, the exit of stations causes an increase in average price. The above analysis implies that drop in the number of gas stations leads to very different effects on price and welfare, depending on how brand and contractual form change. Therefore, policies to affect market structure should be developed and implemented with caution because they may have unanticipated effects.
Issue Date:2015-01-21
Rights Information:Copyright 2014 Moonsik Kim
Date Available in IDEALS:2015-01-21
Date Deposited:2014-12

This item appears in the following Collection(s)

Item Statistics