Technology adoption in response to changes in market conditions: Evidence from the US petroleum refining industry

Abstract

This paper analyzes the determinants of the levels and rates of technology adoption for petroleum refineries that survived the period 1980–1989, during which the conditions of product demand and crude oil supply changed significantly. Regression models are specified to investigate the growth of technology-related capacity, the growth of technology complexity, and the rates of adoption estimated from a diffuse model of technology use. Both levels and rates of adoption are hypothesized to be affected by refinery size, regulatory status, elements of local markets, and geographical factors. Empirical results generally suggest that compared with the supply-side factors of crude oil sources and regulatory subsidies, refinery size and demand-side factors, such as total consumption, consumption growth and fluctuation, and changes of the consumption mix, are responsible for the most part of the determination of technology adoption for refineries surviving the 1980s.

Keywords:

• Technology adoption
• Petroleum refining industry

Acknowledgements

I would like to thank two anonymous referees for their detailed comments on the earlier manuscript. Their comments not only improved this article, but also provided many suggestions for the further studies.

Notes

¹See discussions of Measday and Martin (1986).

²Three major regulatory programs from 1973 to 1981 are the Suppliers–Purchasers Rule, the Buy–Sell Program, and the Crude Oil Entitlements Program. Dickens (1979) has detailed discussions on government regulations.

³Measday and Martin (1986) argue that on the demand side, the cause for the difference in responses to the two oil-price shocks was the long response time required to realize changes in the demand for and supply of energy resources. For example, substantial reductions in consumption by the electric power-generation sector did not begin until 1980.

⁴Heavy-range products include residual fuel oil, lubricants, asphalt, and road oil. Middle-range products include jet fuel, kerosene, and distillate fuel oil. Light-end products include motor gasoline, aviation gasoline, and liquefied petroleum gases.

⁵According to the SRB provision, small refineries with daily average crude oil runs to stills of 0–10 TB/D in that month receive a number of additional entitlements equal to the refinery's daily average crude oil runs to stills for that month multiplied by a fraction, the numerator of which is $0.96 and the denominator of which is the entitlement price for that month. For 10–30 TB/D refineries, the numerator of the fraction is changed to $9,600 plus $0.315 for each barrel by which such refinery's daily average crude oil runs to stills exceeds 10 TB/D for that month. For 30–50 TB/D refineries, the numerator is $15,900 minus $0.095 for each barrel exceeding 30 TB/D. For 50–100 TB/D refineries, the numerator is $14,000 minus $0.10 for each barrel exceeding 50 TB/D. For 100–175 TB/D refineries, the numerator is $9000 minus $0.12 for each barrel exceeding 100 TB/D. Obviously, SRB subsidies decrease when the refinery's daily average crude oil runs to stills are larger than 30 TB/D.

⁶ Petroleum Supply Annual, which is a new version of Petroleum Refineries in the United States and US Territories, was first published in 1982.

⁷Empirical results for the full sample of 164 refineries and for the sample of 130 refineries are very similar in the first model.

⁸In order to understand how large the potential estimation bias caused by the sample-selection procedure (refineries with zero and negative β's are omitted for the sample) is, in the regressions that are not reported in this paper, I run a standard sample-selection econometric model to check the robustness of empirical results. This model includes a probit selection equation, which has a discrete dependent variable equal to one if the refinery is selected as a sample. The inverse Mills ratio obtained from the probit equation is then added to the adoption rate regression as an additional explanatory variable. The insignificant estimated coefficient for the inverse Mills ratio suggests that the selection bias does not cause a serious problem in the estimation.

⁹The area of the average BEA in the lower-48 states is 16,376 square miles, equivalent to that of a circle with a radius of 72 miles or a square with side-length equal to 128 miles. Most trips by petroleum tank trucks are within a 50 mile radius of their origin. See Competition in the Oil Pipeline Industry published by Antitrust Division of US Department of Justice (1984).

¹⁰In order to examine whether pipeline owners, especially refiner owners of pipeline, do or do not discriminate against non-owner refineries in the use of their owned lines, Livingston (1979) groups 174 non-major refineries reporting production on 1 January 1977 into seven categories according to the ownership of the products pipelines, or lines, if any, which by their locations they might conceivably use or like to use. He finds that only 26 out of 174 non-major refiners, i.e., 37% of the non-major refining capacity, are located so that they might conceivably suffer discrimination in the use, or conceivably in the use, of a competitor owned product pipeline and are not in a position to use a non-refiner owned line. But even among these 26 refineries, the potential for discrimination is very limited.

¹¹Supposing that the consumption of each year is c _t , the average growth rate is computed as (((c ₈₁/c ₈₀)+(c ₈₂/c ₈₁)+···+(c ₈₉/c ₈₈))/9)−1.

¹²The 13 Bureau of Mines Petroleum Refining Districts are East Coast, Appalachian #1, Appalachian #2, Indiana–Illinois–Kentucky, Minnesota–Wisconsin–North and South Dakota, Oklahoma–Kansas–Missouri, Texas Inland, Texas Gulf Coast, Louisiana Gulf Coast, North Louisiana–Arkansas, New Mexico, Rocky Mountain, and West Coast. The detailed areas included in each district can be seen in various issues of Petroleum Supply Annual.

¹³Although the ratios vary each year, generally Appalachian #1, Oklahoma–Kansas–Missouri, Texas Inland, North Louisiana–Arkansas, New Mexico, Rocky Mountain, and West Coast have significantly lower foreign–domestic crude receipt ratio than other six districts. The detailed figures are available from the table of ‘Refinery Receipts of Crude Oil by PAD District’ in various issues of Petroleum Supply Annual.

¹⁴The relative relationship among all refining districts in the motor gasoline percentage was fairly stable in the 1980s. Using the data reported in Petroleum Supply Annual published by the Energy Information Administration of the US Department of Energy (1982–1989), the average value of motor gasoline percentage over the 1980s can be computed for each refining district. There are seven districts having the average percentage higher than the US average. They are East Coast, Appalachian #2, Indiana–Illinois–Kentucky, Minnesota–Wisconsin–North and South Dakota, Oklahoma–Kansas–Missouri, Texas Inland, and Rocky Mountain. Other six districts are lower than the US average.

¹⁵In fact, many other interacting relations are considered in unreported regressions. However, only the interactions between size and three change-of-consumption related variables have significant effects.

Additional information

Notes on contributors

Ming-yuan Chen

E-mail: [email protected].