The impact of imports on price competition in the automobile industry.
Traditionally, U. S. automobile firms followed a stable policy in pricing cars. General Motors sets its prices based on a target return on its investments, and other firms imitate its price. According to Donaldson Brown (1924), General Motors estimated sales, cost of unit sold, and added a 15 to 20 percent rate of return to determine its price.
During the 1960's, the price differences between Ford Motor Company and General Motors were only $10 to $20; between Chrysler Corporation and Ford Motor Company, only $40 to $50 (Scherer, 1980, p. 181). In the subcompact category, General Motor's Chevette and Ford's Pinto prices differed by only $11 to $73. For the 1957-1971 period, Boyle and Hogarty (1975) found that the U.S. firms colluded in their price behavior in a hedonic way.
In the 1970's and 1980's, U. S. auto firms adopted several newer pricing policies in response to high oil prices, inflation, interest rates, and frequent recessions. According to R. L. Polk's data (1988, p. 32), the share of imports climbed from 14.53 percent in 1972, to 32.09 percent in 1987. The increase was steady between 1978-1982, and between 1982-1987. Domestic firms retaliated by making smaller cars, cutting wages, closing plants, improving efficiency, and adopting measures to reduce government regulations. Dealer's discount and customer's rebate were widespread. Chevrolet, Ford Motor Company and American Motors practiced a "Two-tier" pricing policy, charging lower prices for subcompacts in the western states where import competition was strongest. General Motors started "interim pricing" --a smaller than usual price increases at the beginning of the model year followed by more frequent increases later. Firms also tried basing price increases on "product improvement" only. Chrysler Corporation did not follow the leader's policies frequently. In one instance, it priced subcompacts in direct retaliation to prices of imports.
The purpose of this paper is to assess whether the increased foreign competition in the 70's and 80's has affected the price behavior of U. S. automobile firms. We used a hedonic price model developed by Boyle and Hogarty (1975) to explain price-collusive behavior among domestic firms for the period 1972-1987. For years in which the traditional price policy was not maintained, we isolated the cheater, and offered explanations for why the hypothesis may have failed. Finally, as it was argued that the U. S. and Japan may have formed an auto cartel via voluntary import quotas, we hypothesized that domestic ad foreign firms may have colluded in price policies for those years in which the hypothesis has failed.
The statistical model for the determination of price uniformity is expressed formally thus: [Mathematical Expression Omitted] where p = price, X = quality characteristics, U = an error term, i = ith model, and t = time period which runs from 1 to k.
In that test, the term "quality" refers to weight, height, length, width, engine size, number of doors, power steering and brakes, and other similar brand characteristics. The performance index represents horsepower divided by curb weight; the comfort index, head and leg room times width. Price refers to list or factory-delivered suggested retail price. It includes standard equipment, federal excise tax, and dealer fee, and excludes state and local taxes and transportation costs.
Boyle's and Hogarty's (1975) best specification was the log of price on thee log of comfort and performance, and a dummy variable for power amenities. The log variables are weighted by the share of U. S. car model year output. In cases where import brands are used, the close comparable share data is imported new-car sales in the U.S.
The null hypothesis ([H.sub.0]) is that estimated list price-attribute relationships are nondifferent among General Motors, Ford Corporation, and Chrysler Corporation. We excluded American Motors Corporation for lack of adequate data. If the sample data by firms for each year cannot falsify the hypothesis, then we inferred that the firms' pricing behavior expressed in the price-attribute relationship was collusive rather than competitive.
The test statistic is F= [[(A--B)/(T--1) (K + 1)]/[B/(N--T(K + 1)]], where A = residual sum of squares from the combined sample, B = the residual sum of squares for each firm's sample summed over firms, T = number of subsamples, K = number of independent variables, and N = sample size. Although Boyle and Hogarty  suggested a five percent significant level, we considered a wider band as well.
We have drawn the sample to allow comparison with Boyle's and Hogarty's results. Before dropping American Motors Corporation from the study, we verified that its data yielded singular matrix solutions in most years. Some of the limitations on the sample size are listed below.
--Models lacking characteristic data are omitted.
--Convertibles and station wagons are not used.
--If the difference in a model is due only to engine specification, then all the models are not used.
--If the model is available either as (1) a 2-door sedan, (2) a 4-door sedan, or (3) a hard-top, only the 2-door sedan is used.
Another consequence of these limitations is that they did not yield enough observations on foreign firm to test the hypothesis that U. S. firms have colluded with foreign firms for those years we conducted the test. According to Sande Milton (1986), for a t-value of 2, an [R.sup.2] of 0.9, three independent variables (k), and a minimum addition to [r.sup.2] of. 01 when a new variable is added, we need 44 observations, i.e., n = k + 1 + [[t.sup.2](1-[R.sup.2])/[r.sup.2]] = 3 + 1 + (4*.1)/.01 = 44. This represents nearly half the size of all import brands in 1986. For pragmatic reasons therefore, we have created one hypothetical import firm, comprising of all import models.
The data come primarily from the "Automotive News Market Data Book Issue" for each year. In some instances, we have used "Ward's Automotive Year Book", especially for model share data. The explanations we offered for firms' behavior are drawn from generally available sources such as the Business Periodical Index, the New York Times Index, Advertising Age, and Auto Basics.
Results and Discussions
As Tables 1 and 2 show, the variables have explained a large percentage of variation in list price. Except for 1984, 1985 and 1987, the level of the coefficients appears fairly stable. The dummy variable was highly significant in all years except in 1986, and 1987. The intercept term was the weakest. It performed poorly in 1972, 1978, 1979, 1983, 1985, and 1987. Apparently, it correlated highly with the dummy variable. However, fitting a constant with a near perfect power dummy variable has not been a problem for this model up to 1987. No singular matrix solution resulted, and an attempt to force the model through the origin resulted in negative [R.sup.2]. The performance index was weak in 1978, 1983, 1984, and 1985, and the comfort index was weak only in 1980. [Tabular Data 1 and 2 Omitted]
The negative coefficients for the performance index in 1984, 1985 and 1987 do not falsify the hypothesis. The year 1984 marked a comeback of performance for domestic firms. Firms stepped up nonprice competition against the increased share of imports. In that year, General Motors positioned the first American mid-engine sports car, and the wedge-shape Pontiac Fiero. Chrysler pitted the Daytona as a new performance car, and Ford mobilized new turbo charged and diesel engines for its Mustang, Cougar, Capri, Topaz and Lynx. Each firm offered an engine in excess of 200 horsepower. At the same time. American cars were becoming lighter with more high-stress and high-strength plastic styling. In 1987, 80 percent of the cars were fitted with four- and six-cylinder. The full effect of performance in those years were only partly captured in the performance index. The residual that picked up the other part appeared to have dominated the horsepower to weight ratio we have used. The net result was negative coefficients in those years. As no superfluous variables appeared in the specification, we followed the advise of Rao and Miller (1971, pp. 38-46) in keeping the equations.
Comparing the test statistics with the critical points of the F-distribution, the results of Table 2 indicate that we should have no doubt in accepting the hypothesis in 1973, 1975, 1982, 1985, 1987. At the ten percent significance level, we should accept it in eleven of the sixteen years. However, we should reject it in 1978, 1980, 1981, 1983, 1986.
Why did the hypothesis fail in those five years? In 1978, the "interim pricing" policies described above were practiced. In 1980, firms increased prices every quarter in response to inflation rates. "Sometimes a manufacturer will toss in an extra hike, as Chrysler did in March on Omni and Horizon." (Automotive News, 1980, p. 62) In 1981, Ford Motor Company and Chrysler Corporation hesitated to follow leader's price policy. "GM hiked prices a substantial $351 effective April 13, bringing its 1981-model-year average to $914 and pushing the average price of the average-equipped GM car above ten thousand dollars. It become $10,200 on that date.
Then came a couple of surprises. Ford, which almost always follows GM in such matters, said it would not hike prices before the end of April. And Chrysler said it would make no increase |until absolutely necessary'." (Automotive News., 1981, p. 50)
In 1983, the U. S. auto makers were looking for newer direction in pricing that year's models. It was third year in which the industry was in recession. General Motors' average price increase was somewhat flat, 1.9 percent. Still, the other firms were reluctant to follow. Ford Motor Company average price increase was a modest 0.4 percent. Chrysler went the other way, cutting the base sticker price on its convertibles and other models.
In 1986, General Motors reduced its incentive programs as the industry had its most profitable second year in 1985. Its average price increase was 2.9 percent or about $350 per unit. Chrysler Corporation again refused to follow, reducing prices on its Omni Plymouth Horizon to about $710 below current base prices of subcompacts.
Table 3 presents additional tests for the years the hypothesis has failed. We attempted to find different grouping of firms that set prices uniformly. The results show that General Motors and Ford Motor Company have colluded in 1986. The leadership role had changed in that year with Ford Motor Company topping General Motors in earning, reversing a historic trend since 1924. Chrysler Corporation behavior in going alone appears to corroborate Schwartzman's words that "Small firm . . . will be more inclined to utilize price cuts in an effort to increase sales" [1970, p. 107]. The subgrouping of firm also indicate that Chrysler Corporation had abandoned the price-cartel in 1986. [Tabular Data 3 Omitted]
The voluntary import quotas of the 1980's has raised the issue to whether the domestic firms accommodated, extinguished or neutralized foreign firms' price policy. Dolan and Lindsey (1988, p. 960), have argued that an international automobile cartel was formed between Japan and the U. S. consequent to the price benefits firms have enjoyed. We have tested the price hypothesis with domestic vs. imported models to identify any international price-cartel. Data availability limited the test to the broad hypothesis of U. S. and U. S. firms vs. all Imports for 1983, and 1986. Table 4 displays the results.
The results indicate a high probability that the domestic and foreign firms do not price the characteristic of their product similarly for the two years in which the price-collusion hypothesis failed for the domestic industry. Given the experimental nature of the domestic firms price policies, and the propensity of Chrysler Corporation, the smallest firm in the cartel, to cheat, the domestic firms behavior in 1983 and 1986 appeared as neither openly extinguishing nor accommodating, but mostly neutral. Only the smallest firm exhibited extinguishing tendencies. General Motors and Ford Motor Company preferred the oligopoly with differentiated product, choosing nonprice weapons such as R&D, advertising, styling, dealership arrangements, over price competition to rival foreign competition.
Our findings allow the inference that domestic firms in the automobile industry followed a uniform price policy when foreign competition was weak. Strong foreign competition has caused Chrysler Corporation to break away from traditional pricing policies. While import competition was not significantly disruptive to domestic price policies in eleven of the sixteen years between 1972 to 1987, its force was most visible in 1978, 1980, 1981, 1983, and 1986. The domestic industry showed its thick skin in those disruptive years when import quotas were in place. Firms had the chance to step up nonprice competition, maintaining their differentiated oligopoly status quo, while avoiding cooperating with foreign firms in setting price policies. [Tabular Data 4 Omitted]
Boyle, Stanley E., and Hogarty, Thomas F., Pricing Behavior in the American Automobile Industry, 1957-71, The Journal of Industrial Economics, Vol. XXIV (December, 1975). Bradley, Albert, in the Eighty-fourth Cong., First See., Senate Subcommittee on Antitrust and Monopoly, Senate Committee on the Judiciary, Hearings on General Motors, 1955, Pt. 7. P. 3593. Brown, Donaldson, Pricing Policy in Relation to Financial Control, Management and Administration, (February, 1924). Caves, Richard, American Industry: Structure, Conduct, Performance [N. J.: Prentice-Hall, Inc., (Second Edition), 1964]. Chamberlin, Edward H., The Theory of Monopolistic Competition [Mass.: Harvard University Press, 1965]. Dhrymes, Phoebus J., Price and Quality Changes in Consumer Capital Goods: An Empirical Study, in Griliches, Z. (ed.), Price Indexes and Quality Change [Mass.: Harvard University Press, 1971]. Dolan, Edwin G. and Lindsey, David E., Economics, 5th Edition [N.Y.: The Dryden Press, 1988]. Edwards, Charles E., The Dynamics of the United States Automobile Industry [Columbia: University of South Carolina Press, 1965]. Epstein, Ralph C., The Automobile Industry [New York: A. W. Shaw Company, 1928]. Ferber, R., and Verdoorn, P. J., Research Methods in Economics and Business [N.Y.: The Macmillan Company, 1962] Havrilesky, T., and Barth, R., Non-Price Competition in the Cigarette Industry, The Antitrust Bulletin, XIV (Fall, 1969), pp. 607-628. Hogarty, Thomas, Price-Quality Relations for Automobiles: A New Approach, Applied Economics, Vol. 7, 1975, pp. 41-51. Kennedy, E. D., The Automobile Industry, The Coming of Age of Capitalism's Favorite Child [New York: Reynal and Hitchcock, 1941]. Kwoka, Jr., John E., Market Power and Market Change in the U. S. Automobile Industry, The Journal of Industrial Economics, Vol. XXXII (June, 1984). Lanzillotti, Robert F., The Automobile Industry in Adams, Walter, (ed.) The Structure of American Industry [New York: The Macmillan Company, (Third Edition), 1961]. Maxcy, G., and Silberston, A., The Motor Industry [London: Allen and Unwin, 1959] Milton, Sande, A Sample Size Formula for Multiple Regression Studies, Public Opinion Quarterly, Vol. 50, Spring 1986. Moore, Donald A., The Automobile Industry In Adams, Walter, (ed.), The Structure of American Industry [New York: The Macmillan Company, (Revised Edition), 1954]. Polk, R. L., in Automotive News, 1988, p. 32. Scherer, F. M., Industrial Market Structure and Economic Performance [Chicago: Rand McNally College Publishing Company, 1970 (revised 1980)]. Roa, Potluri and Miller, Roger L., Applied Econometrics [Belmont: Wadsworth Publishing Company, Inc., 1971] Schwartzman, David, Oligopoly in the Farm Machinery Industry [Canada: Royal Commission on Farm Machinery, Study No. 12, 1970] Sherman, Roger, Oligopoly: An Empirical Approach [Mass.: Lexington Books, 1972] Telser, Lester G., Competition, Collusion, and Game Theory [Chicago: Aldine and Atherton, 1972]. White, Lawrence J., The Automobile Industry Since 1945 [Mass.: Harvard University Press, 1971]. White, Lawrence J., The Automobile Industry in Adams, Walter (ed.), The Structure of American Industry [New York: Macmillan Publishing Co., Inc., 1977]. Lall B. Ramrattan Economist at the U. S. Department of HUD, San Francisco Regional Office, and Professorial Lecturer of Economics at Golden Gate University, and Adjunct Professor of Economics at JFK University.