Experimental Gasoline Markets Cary A. Deck University of Arkansas Bart J. Wilson George Mason University Evidence-Based Public Policy Conference Fall, 2005

Motivation Few industries evoke such strong sentiments by consumers, retailers, wholesalers, and policy makers as gasoline. Why? Consumer and business demand for gasoline is inelastic. Modern economies depend on a large volume of gasoline. Retail prices are posted nearly everywhere we drive.

Motivation The practice of zone pricing has been a particularly contentious topic in the public policy debate. Zone pricing is the industry term to describe the practice of refiners setting different wholesale prices for retail gasoline stations that operate in different geographic areas or zones. Chevron contends that they “price our wholesale gasoline to our dealers at prices that will allow them to be competitive in relation to their nearby competition.” Connecticut Attorney General Richard Blumenthal proposed legislation to ban zone pricing claiming that it “only benefits the oil industry, to the detriment of consumers.”

Motivation Another issue is divorcement, the legal restriction that refiners and retailers cannot be vertically integrated. Maryland was the first state to pass such legislation in 1974 with a handful of other states following suit. Bill Lockyer, California Attorney General, in a task force report states that “the key to enhancing competition at the retail level is to eliminate vertical integration by petroleum companies.” However, this runs counter to basic economic theory and evidence from field studies [Barron and Umbeck (1984) and Vita (2000)].

Motivation Yet another topic that has led to much public debate is a “rockets and feathers” phenomenon in retail prices. This is the perception that retail gasoline prices rise faster than they fall in response to cost shocks. Beyond gasoline, Peltzman (2000) finds the phenomenon in 2/3 of industries he tested. Using monthly national prices, he finds that “rockets and feathers” is uncorrelated with concentration. For gasoline, Borenstein et al. (1997) presents a collusive theory based upon trigger strategies. Other explanations posited include inventory costs, menu costs, and consumer search costs (e.g., Johnson 2002, Castanias and Johnson, 1993).

Industry Background World Market Price Oil Field Refiners Wholesalers Unbranded Rack PriceBranded Rack Price Wholesalers Unbranded Stations Company Operated Stations (Branded) Lessee Stations (Branded) Dealer Owned Stations (Branded) Transfer PriceDealer Tank Wagon Retail Customers

Environment and Institution Consumers Each buyer has a value v for one unit of gasoline. A fraction  i of buyers have a preference for brand b i, i.e., these buyers gain additional utility if they consume brand b i. Each buyer has an initial location on a “city grid” and incurs a quadratic travel cost to reach a station. All retail prices are public information. Consumers purchase one unit from the station offering the greatest net utility, assuming it is positive. Robot buyers operate in the market.

Environment and Institution Refiners Only refiner i can sell its branded gasoline b i at a cost per unit of c i. Refiner i sets wholesale per unit prices (DTW) for K units of gasoline. Retailers A retailer j is contractually obligated to carry a particular brand and only observes the DTW for that refiner. Each retailer has an exogenously determined location in the “city grid.” Retailer j sets the retail price p j for a unit of gasoline and its costs include DTW j and an operating cost of e j.

Treatments Zone Pricing (4 Refiners and 4 Lessee Dealers) Refiners set DTW prices for each retail location carrying its brand. Each retailer observes two location specific DTW prices but cannot shift inventory between locations. Uniform Pricing (4 Refiners and 4 Lessee Dealers) Refiners must set one price for both retail outlets carrying its brand. Company Operated (4 Retailers) Retailers and refiners merged so that DTW for brand i outlets is c i.

Experimental Design and Procedures

Store 6 P=?

Experimental Design and Procedures Twelve laboratory sessions, four in each treatment. Each session lasted no longer than 90 minutes. - a period lasted about 2 seconds. Subjects were undergraduate students. The average payoff was $18.25, including $5 for showing up on time. World oil prices First 600 periods c i constant. Last 600 periods c i followed a random walk with changes occurring every 34 to 60 seconds.

Results: Zone (Wholesale) Pricing Wholesale Prices Posted Retail Prices

Results: Uniform Wholesale Pricing Corner Retail PricesCenter Retail Prices Zone Treatment Uniform Treatment

Why is Zone Pricing not Harmful?

Who Benefits from Uniform Wholesale Prices? The stations.

Who Benefits from Uniform Pricing? Zone PricingUniform Pricing

Which Buyers are Harmed? 16.9% reduction in utility17.8% reduction in utility

Results: Company-Operated Corner Retail PricesCenter Retail Prices Zone Treatment Company-Operated Treatment

Which Buyers Benefit? 20.1% increase in utility24.4% increase in utility 50.6% increase in utility

Who Benefits from Vertical Integration? Zone PricingCompany-Owned

Price Dynamics: Center Stations Zone Pricing: Prices and Costs are Cointegrated Uniform Pricing: Prices and Costs are Not Cointegrated

Price Dynamics: Corner Stations Zone Pricing: Prices and Costs are Cointegrated Uniform Pricing: Prices and Costs are Not Cointegrated

Price Dynamics: Company-ops Corner Stations: Prices and Costs are Cointegrated Center Stations : Prices and Costs are Cointegrated

“Rockets and Feathers”: Experiment Level

Policy Conclusions Banning Zone Pricing When zone pricing is banned, consumers in the clustered area pay 11% higher prices than when zone pricing is permitted. Consumers in isolated areas pay the same prices with zone pricing as they do when it is prohibited. Banning zone pricing nearly triples average station owner profits, but has no effect on refiner profits. Divorcement Consumers in the clustered area and isolated areas respectively pay 13% and 17% lower prices with vertical integration than with divorcement.

Rockets and Feathers Conclusions Station prices in the clustered area adjust quickly with zone pricing, but still rise faster than they fall. Station prices in the isolated areas adjust more slowly than in the clustered area, but rise as fast as they fall. With company-owned stations, station prices adjust symmetrically to changes in station costs, but this response is much slower than with vertical separation. With uniform wholesale pricing, station prices and costs are not cointegrated.

The End.