Home | Business News | Browse by Publication | I | International Advances in Economic Research

The consumer's micro-micro gasoline buying decision.

Article Excerpt
Abstract A "micro-micro" consumer problem of gasoline purchases is examined using daily price data. Comparing the optimizing consumer with one who buys gasoline at random, the paper finds optimizers save about 4% of their annual gasoline bill. The paper also provides some evidence about the costs of non-optimal gasoline buying strategies.

Keywords Gasoline * Consumer * Value of information

JEL Classification Q40

**********

This paper investigates the "micro-micro" consumer problem of gasoline purchases during a given planning period--that is, how much an idealized consumer (i.e., an "optimizer" or cost minimizer) who buys gasoline from one of four stations in Peoria, Illinois between November 18, 2003 and October 25, 2005 would pay for gasoline as compared to various other ad hoc gasoline buying strategies. Buying gasoline requires that consumers track gasoline usage and plan the next refueling before running out of gas. On the other hand, a finite tank size limits stocking up to take advantage of low prices. Refueling is time consuming so consumers will not refuel any more often than necessary. Finally, gas prices are prominently posted, visibly changing from day to day. A consumer may buy gas today and then be chagrined to see tomorrow that the price has fallen.

The "idealized" consumer assumed herein is allowed information that real consumers do not have--the price of gasoline tomorrow and the day after. Thus, the paper could also be seen as a test of the value of that information under various circumstances. The paper also measures the amount, on average, an idealized consumer can save by refueling more than once per planning period. An individual who has a good sense of the value of his or her time could thus use the experiments of this paper to provide some evidence about whether taking the time and trouble to buy gas as cheaply as possible is in fact optimal for him or her. In general, it is found that the optimizer saves at most about 8% on gasoline expenditures; often savings are about half that, around 4%, and under a considerable number of circumstances the savings are in the 2-3% range.

[FIGURE 1 OMITTED]

Although the larger problem of consumer demand for gasoline is well studied (gasoline demand being inelastic) Nicol (2003), Cheung and Thomson (2004), Ramanathan and Subramanian (2003), and Espey (1998), it seems that the micro-micro problem using daily prices in this paper has not been considered in the literature.

Data and Parameters

The gasoline prices were obtained from the (lowest grade) postings outside four stations in Peoria Illinois each day that an author was able to view them for the study period of November 18, 2003 to October 25, 2005. The stations, along a 3-mile length of a single major thoroughfare, included major brands, local brands, and stations with and without repair facilities. The maximum and minimum price per planning period is shown for the 52 12-day (non-overlapping) planning periods for which full data exists. The average range per planning period is about $0.20, and the range (of ranges) is $0.090 to $0.46 (Fig. 1).

The "micro-micro" problem requires consumer data on daily fuel use, tank size, and the length of the planning period. Our parameters are given in Table 1 and explained next.

Vehicle classifications are derived from cars.com ("large cars" being midsize and full size) whose selections include domestic and foreign, low priced and high priced cars. The parameter values are averages for the 2004 model year (the middle year of the gas price data). Daily fuel usage is calculated by

daily fuel use = [(miles per year)/(miles per gallon)]/(365 days per year) = gallons/day

where miles per year is from Federal Highway Administration data and average miles per gallon is from cars.com. Average tank size is from cars.com as well. The planning periods of 12 and 10 days are the longest feasible given the preceding parameters and the assumption that...