首页    期刊浏览 2025年07月26日 星期六
登录注册

文章基本信息

  • 标题:Advertising rivalry in the U.S. automobile industry: a test of Bain's hypothesis.
  • 作者:Ramrattan, Lall B.
  • 期刊名称:American Economist
  • 印刷版ISSN:0569-4345
  • 出版年度:1994
  • 期号:September
  • 出版社:Omicron Delta Epsilon

Advertising rivalry in the U.S. automobile industry: a test of Bain's hypothesis.


Ramrattan, Lall B.


Introduction

Working within Chamberlin's(13) theory that predicts a firm's market behavior from its market structure, Joe Bain(6) classified the U. S. Automobile industry as an oligopoly with high product differentiation. He predicted "effective collusion on price" combined with "rivalry in sales promotion" for that industry(6), p. 311. For the 1957-1971 period, Boyle and Hogarty(12) found that U. S. auto firms colluded in their price behavior, explained by indexes of comfort and performance and a dummy variable for power steering and brakes. Ramrattan(40) expanded that study for the 1972-1987 period, and reached the same conclusion. By failing to falsify Bain's price collusion hypothesis, those studies have opened the door to advertising expenditure as a positive performance indicator. Bain actually listed physical design, product reputation, conspicuous-consumption motives, and dealership systems beside advertising,(6) p. 300, as bases for firms to differentiate their products. Studying that array of non-price weapons, including R & D expenditures, Ramrattan(39) has argued that advertising is the preferred rivalry weapon for the auto firms. A good advertising copy shelters a firm from the rain of competition for about a year before rivals start imitating. Other non-price weapons do not have that trigger-power, and in some cases are dependent on advertising for their bang. A validation of Bain's Advertising hypothesis would enhance our understanding of product differentiation within the Bain-Chamberlin's paradigm that predicts a firm's market performance from the industry's market structure.

In this paper, we set price collusion and advertising rivalry within the ring of international changes that may have impacted domestic firms advertising policies. The U. S. Automobile firms suffered great instability in the post OPEC cartel era. On October 20, 1973, OPEC increased oil prices by 70 percent, and instituted an oil embargo on the U. S. By the end of that year, oil prices nearly quadrupled from $3.01 to $11.65 a barrel. Again in December 1978, oil prices increased by 14.5 percent consequent to the Iranian revolution, cutting supplies by 5 million barrels a day. March, 1979 saw another 9 percent increase by the OPEC, with agreement for members to add surcharges. Domestic oil prices were decontrolled in January 1981. By then, the double digits inflation rate was foiled with double digits interest rate. Early attempts by the Carter administration to curb inflation by credit restraint had an unintended effect on auto consumption, which was exempted from the restraint. Meanwhile, General Motors' stock price moved cyclically in all the recessions following the OPEC cartel in 1974.

We find the U. S. Firms exploring newer and better ways, globally and domestically, to confront foreign competition. Small firms particularly experienced cashflow setbacks, making advertising less affordable and probably more dependent on sales. Lending credence to Chrysler's new ad to the effect that firms either lead, follow, or stay out of the competition, American Motors strategically merged with Chrysler Corporation in 1987, while foreign firms continued to make steady inroads into domestic firms' market share. This paper appraises Bain's advertising hypothesis in those domains of competitive activities. After adjusting some auxiliary assumptions to accommodate the global competitive environment, we argue for postponing a falsification of Bain's advertising hypothesis.

The Model

Game Theory represents a progressive research program in modeling price collusion and advertising rivalry. While Cournot(14) and Bertrand(9) allowed only quantity and price adjustments respectively in modeling oligopolistic behavior, John Nash(35) allowed adjustment through anything and everything, including advertising. We seek an advertising game with at least two firms, N, as players, with similar strategies, S, a payoff function, II, and an information set, I. The information set must include a knowledge of how the firms price their products, the level of a rival's previous advertising outlay, and the states of the firm own cashflow, sales, market share, and marginal cost. As leaders can use information not available to the followers,(41) and to the extent we have identified a leader-follower pattern in the empirical section, the game will be played with nonsymmetric information in extensive form,(48) p. 441. If the information the leaders use to set their budgets are in public domain, then the followers will have access to it. But the leaders can have variables in their decision set that the followers do not know about. Followers can only speculate on the use of local information such as the gradient of the leader's profit function, while the whole global set of information lays unexplored.

Price collusion implies that a firm's demand function, !Mathematical Expression Omitted^, will posit a predetermined price vector, !Mathematical Expression Omitted^, set by firms agreeing to price the amenities of their product similarly, and a controlling advertising decision vector, !Mathematical Expression Omitted^, which maximizes a firm's profit or market share. The subscript on advertising indicates that it can vary for different firms. The shape of the demand curve will depend on how advertising competition proceeds. As Shubik stated,(48) p. 138, ". . . neither the kinked oligopoly demand nor the contingent-demand construction is of much use in studying oligopolistic competition unless supplemented by an appropriate ad hoc study of marketing realities." Because the auto industry is highly concentrated and differentiated, Bain,(5) p. 311 predicted that "advertising . . . will be matched by rivals as surely as price cuts, but less easily, less quickly, and less exactly, thus giving the individual seller a better chance to gain an edge through his own ingenuity." That argument gives a green light to rapid imitation and retaliation, but a possible amber or even red light to advertising wars because a firm, through its ingenuity, can have a superior advertising copy that would take rivals time to imitate.

A leader will have an elastic demand curve if it initiates an increase in advertising and rivals remain neutral. According to Schnabel,(43) p. 40, the rivals need not react with a proportionate increase in advertising to regain lost profits or market share as the market would have expanded. If rivals were to believe their advertising copies were superior, and decide to match the leader's increase or even up the stake, then, the leader's demand curve would be less elastic. In the worst case scenario, a firm may think its advertising copy is inferior, a probable cause for the years of negative cashflow for Chrysler and American Motors. An alternative to being optimistic or pessimistic about the effectiveness of their advertising copy is a joint profit maximization solution through tacit collusion, where the group's allocation of advertising is more than offset by gross revenues.

Profit usually means gross earnings less average or indirect cost. Firms maximize their profits using the intelligence in their information set. In the domestic scene, profit is driven mainly by advertising, where the leaders set their budgets, and the followers react. In the open economy, profit is sensitive to shocks such as the oil price hikes due to OPEC. Such a shock from the open economy can have a disabling effect on the driver. However, in a gaming environment, if such an effect diffuses proportionately across firms, then it should have only a scaling effect on profits, and the play of the game should not be affected. The following payoff function resembles Balch's,(7) except for the proportional effect of OPEC, !Zeta^, on the firms market share and therefore profits, and the nonsymmetric information set, I. The effect of !Zeta^ on profits will depend on how the market share function is specified. We would expect brand loyal customers to be less influenced by OPEC, setting a positive lower limit on market shares. We seek an equilibrium strategy vector, !Mathematical Expression Omitted^, for which at least two firms advertise at a positive level. We will discuss conditions for a Nash equilibrium below.

!Mathematical Expression Omitted^

Where:

!Mathematical Expression Omitted^ and !Mathematical Expression Omitted^ are defined as above.

A!C.sub.i^ is the !i.sup.th^ firm average production cost per unit.

M !S.sub.i^ !Mathematical Expression Omitted^ is the expected market share for the !i.sup.th^ firm.

!Mathematical Expression Omitted^ is the !i.sup.th^ firm expected profit.

!Zeta^ is the proportional effect of OPEC on firm market share.

!I.sub.i^ is the !i.sup.th^ firm information set.

The above predictions are basic to the Chamberlin's paradigm that pits one firm against the rest of the market. Bain,(5) p. 279 has suggested that "the forces

of antagonism and distrust among rivals are so strong that no collusive agreement among sellers is possible and that no regular pattern of reactions . . . becomes established." He looked to "a mathematical theory of strategy of games . . . where each player seeks to avoid the possibility of considerable loss yet gain as much as possible consistent with safety" for a possible answer. Several attempts have been made to model advertising rivalry in a gaming environment. Michael Balch's(7) noncooperative and Shubik's and Levitan's(46) joint noncooperative and cooperative games are notable instances.

Strategies can be pure or mixed. A player may not know which pure strategy an opponent will choose, but he may know the probability with which the opponent will mix or choose pure strategies. The pure advertising strategy set is an n-tuple, S = (!s.sub.1^, !s.sub.2^, . . . !s.sub.n^), which may represent different advertising budgets--high, low, medium, cooperative, noncooperative; advertising expenditures in different media, or advertising expenditures on different brands. If the game is played with symmetric strategies, and nonsymmetric information, the leaders, by being better informed should gain more than the followers, unless their advertising copies are inferior. However, because players are tagged leaders or followers, we cannot interchange their roles in the play as would be allowed by symmetry. It does not make sense to flip a coin to determine which firm will lead when we can identify the leader empirically.

We expect a class of solutions in games of pure strategy as in case of Okuguchi and Ferenc,(37) of mixed strategies as in the case of Nash,(35) and of nonsymmetric information in the case of Dubey and Shubik.(20) For a game in mixed strategy between General Motors and Ford, where the strategy sets !s.sub.1^ !is an element of^ !S.sub.1^ and !s.sub.2^ !is an element of^ !S.sub.2^ are convex and compact, General Motors's reaction may be !!Mu^.sub.1^(!s.sub.1^); Ford's reaction, !!Mu^.sub.2^(!s.sub.2^). If those reaction functions are quasi concave, then the battle space, !S.sub.1^ X !S.sub.2^, will be convex. The state s = (!s.sub.1^, !s.sub.2^) is an equilibrium pair if the mapping s !right arrow^ !Mu^(s) from S = !S.sub.1^ X !S.sub.2^ to S itself has a fixed point. The mapping need only be upper hemi-continuous to guarantee a solution !See(32) p. 106, and(15, 16)^.

In cases where the product of the strategy sets is oval instead of Cartesian, an equilibrium solution may not exist. Discontinuity caused by capacity limitation is known to obstruct equilibrium(47, 45). The merger of Chrysler and American Motors Corporation has minimized that problem, but by reducing the number of participants, it had an opposite effect as well. If we cut the convexity assumption, we can use the Shapley-Folkman theorem as a band aid. In technical jargon, the Hausdorff distance between the average of sets and the convex hull in the limit tends to zero, and enables an approximate equilibrium. !See Anderson(1, 2, 3)^. However, conditions for Nash equilibrium are necessary but not sufficient. When multiple equilibrium pairs exist, the payoffs need not be equivalent. In other words, the equilibrium pairs are not interchangeable,(32) p. 106. Players will have their preferred equilibrium and can either play another game,(32) p. 66, use a "payoff-dominance" technique,!29^ or use a "focal point" method,!42^ to find the best equilibrium pair. However, the equilibrium to that additional game could be unstable, and may yield lower profits to both players, creating motivation for tacit collusion.

The prisoner's dilemma game makes collusion possible in a noncooperative environment. This point was emphasized by Andreu Mas-Colell,!33^ p. 15, who wrote that the "concept of a blocking coalition can . . . be identified with that of a feasible set of contracts an entrepreneur can put together at his own gain and with the voluntary participation of everyone concerned." The players (defendants) are not allowed to communicate. Their dilemma is that they prefer the cooperative outcome, which does not allow them to take advantage of each other's play. At best, they can signal one another about their intention to collude by playing the collusive strategy for a while. But over the infinite horizon, "most intelligent people would not play accordingly", because a "player can obtain sequential information about his opponents strategy and if it appears not to be optimal, then he can attempt to exploit the deviation",!32^ p. 103. General Motors may have expended $6 million in the previous period. Ford follows with $6 million in the current period, and General Motors may rejoin with $6 million the next round. But if Ford did $10 million instead, then GM might rejoin with $10 million. In this noncooperative environment, General Motors could well end up with a cooperative strategy vector {$6, $6, $6}, where advertising budget is usually lower, or the play {$6, $6, $10}, with two plays of cooperative strategy followed by one play of noncooperative strategy. Friedman,!22^ prescribed starting with a higher than noncooperative payoff strategy, continuing that play for as long as rivals play accordingly or switching to a cooperative strategy if an agreement is possible. Telser!49^ advocated starting from a joint profit position. If one player cheats on the agreement, then the other player will have the opportunity to impose a penalty at a later period. The side of the noncooperative vs. cooperative coin that will prevail can be explored further via an empirical test. A reaction coefficient close to one between two players tells that the firms are responding almost fully to rivals previous period outlay.

Specification and Data

A. Hypothesis 1: Cash Flow Model

We will explore a firm's advertising response, !A.sub.i^, to its rivals's outlay in some previous period, !A.sub.j,t - s^, given its cashflow situation, !P.sub.t - s^. Grabowski's and Mueller's work(24) suggests that a log-linear Cobb-Douglas type specification, with a one period lag, s = 1, is most suitable for rivalry problems. Because that specification explained the data well for the period 1960-1978, Ramrattan,!39^ we make it the starting point for the model in this paper.

Over time, the net income plus depreciation measure of cashflow has been nonstationary, and even negative in many years for small firms, as well as for General Motors in 1990. Therefore, fitting the log-transformed model of Grabowski and Mueller!24^ is impossible. As the negative values were more systematic than random, dropping the observations, or substituting the mean of the series, or replacing the observation with an instrument may have radical influence on the results!38^ (Ch. 9). Other variables such as market share and sales did not out-perform cashflow independently or in combination via principal component analyses or in weighted average form.

B. Hypothesis 2: OPEC Cartel Hypothesis

We want to pick up the influence of foreign firms on domestic firms advertising rivalry policies particularly since the OPEC cartel period in the last quarter of 1973 when foreign competition became most visible. For that purpose, we introduced an OPEC dummy variable, D, with a value of one for the post OPEC cartel periods, and zero otherwise. Essentially, the dummy variable performs a Chow test,!27, 28^ between the two periods, whose significance would demarcate the sub-periods. A significant dummy coefficient would also corroborate the impact of foreign firms expansion in the U.S., which was accelerated after the OPEC cartel was formed. We anticipate the coefficient in the latter period to be less stable. In the latter period, the domestic firms have followed expansive policies abroad. Success in that effort should filter back as positive influences on their domestic promotional efforts.!21, 44, 10^ Finally, the dummy variable is a catch all for the Multinational Corporation (MNC) acting as a conduit for foreign capital movements, adding life to its product-life-cycle, sheltering its threatened export position, taking advantage of extractive ventures, initiating interpenetration of investments deals, and taking care of its 'special advantage' that requires 'own control'.!10^ We propose the following specification where variables, except the dummy, are in log-arithmetic form:

!Mathematical Expression Omitted^

Where A = Advertising Expenditure.

P = Cashflow (NI + Depreciation).

i;j = !i.sup.th^ (leader) and !j.sup.th^ (follower) firm.

!Alpha^, !Beta^, !Gamma^, !Lambda^ = Estimates.

t = Time.

D = Dummy (one for 1974-90, zero otherwise).

s = period of lag (One Year).

C. Hypothesis 3: Follow-the-Leader Hypothesis

"What is good for the country is good for General Motors, and what's good for General Motors is good for the country."!51^ The above specification facilitates a test of the much argued General Motors leadership hypothesis for the automobile industry. To falsify that hypothesis requires a test for each permutation of rivalry relationships; i.e., with n = 4 domestic firms, and r = 2 firms in a rivalry relationship, 4]/(4-2)] = 12 relationships must be estimated. American Motors is included for periods which data is available in order to avoid a survival bias in the model. For this hypothesis, we seek a reaction pattern where firms consistently follow General Motors, the leader, in setting their advertising budget.

It would be more convenient to accept one firm, say General Motors, as the leader and to test each firm responses to General Motors outlay. Besides assuming what is to be proven, that procedure will leave the problem of determining how General Motors sets its advertising budget. Only general rule-of-thumb answers are possible for this problem.!30, 34, 8^ Further, simplistic solutions may explain how General Motors sets its advertising budget, but not how Ford, Chrysler, and American Motors set their budget. To reconcile this problem, we will specialize experiment with and without the leadership hypothesis. That will require 24 regressions -- 12 for the leadership case and 12 for the non-leadership case, including the reverse cases as well.

D. Estimation

The sample data for the 1960-1990 period and other sub-periods would require other modifications to stay within Bain's advertising paradigm. For analytical purpose, we will call the above specification the "affordable" rivalry hypothesis, implying that a firm's advertising is constrained by its cash flow. It opposes other hypotheses such as "frontal attack" on say market share or sales. In retrospect, using foreign firms market share in place of sales or cashflow did not perform well on preliminary trials. In fact, the traditional sales and cashflow variables did not perform well in their original form, because of multicolinearity among the independent variables. To correct for that, we regressed the firms advertising variable in the independent variable set on the firm cash flow. The residual, which is net of interaction is calculated and used instead of the independent advertising variable in a new regression. This procedure is not without its price. If the original model is true, then in specification analysis, the orthogonalization process will likely bias the non-advertising coefficients, Giliberto,!23^ and Neuberger.!36^ The payoff is that both cash flow and sales have improved in their performance. We have accepted some theoretical bias but gained some efficiency in the estimates.

E. Source and Coverage of Advertising Data

As Bain!6^ observed, only "transaction type" data are available on advertising. Advertising Age provides the only consistent data set for the time period 1960-1990. The Leading National Advertisers (LNA) publication has advertising data only from 1973, and lags two to three years in publication. Media Decision publishes data for even fewer years, and has changed its format for reporting brand data from 1986 onwards. Other data were taken from generally available sources -- Compustat tapes, Datext, Infotrac, Moody's industrial Manual, and firms 10K and Annual reports.

While Advertising Age provides a long enough timeseries, it does not cover a consistent set of media from 1960-1990. In 1960, it covered seven media-newspapers, general magazines, farm magazines, business publications, network television, spot television, and outdoor. In 1964, spot and network radio media were added, and in 1974, the business publications medium was dropped. We were able to use only four media-newspapers, general magazines, spot television, and network television, for four firms--General Motors Corporation, Ford Motor Company, Chrysler Corporation, and the American Motors Corporation.

TABULAR DATA OMITTED

F. Test and Result

Table 1 presents the results for the four-media total advertising outlays for the GM leadership hypothesis. Equations 1 and 1!prime^ show advertising affordability measured by cashflow; equations 1a and 1a!prime^, affordability measured by sales. Prime numbered equations are results corrected for multicolinearity. The Cochran-Orcutt first order serial correlation correction procedure was performed on all the equations. Table 2 presents other reaction patterns besides the General Motors leadership.

The advertising coefficients in Tables 1 and 2 are highly significant. Twenty out of twenty four, 83 percent, are significant -- 15 at the 99 TABULAR DATA OMITTED percent confident level, and 5 are significant at the 95 percent level. Only four equations are insignificant at the 95 percent level, and they involve AM vs. CH, the small firms that were most affected in the OPEC cartel era. Correcting for multi-colinearity, the advertising coefficients average .68 under the cashflow hypothesis, and .60 under the sales hypothesis. Being significantly less than one, those averages do not allow any implication for perfect competition alluded to by Grabowski and Mueller,!24^ and Schnabel.!43^ However, when account is taken for expansion in the market as Schnabel argued, the overall results do indicate significant competition as Bain's advertising hypothesis predicted.

The foil between cash flow and sales does not weaken the conclusion of advertising rivalry. Cashflow is up one on the number of significant coefficient over sales in the GM leadership hypothesis, and down one in the non-leadership hypothesis. It swinged from poor to much improved performance when the multicolinearity procedure is performed, is highly significant for the GM leadership hypothesis, and not significant for AM vs. CH estimates in the other reaction patterns in Table 2. Sales does not do well as a surrogate for cashflow in the FD vs. GM rivalry, but it elevates the standing of the smaller firms in the results of Table 1. As an alternate to cashflow, sales does not bring out the OPEC cartel influence at all. In the one case where the dummy coefficient with sales shows up significant, viz., in equation 1a!prime^, the sales coefficient is insignificant. On the other hand, its significance in Table 2 with multicolinearity corrections outshines that of cashflow, which is even insignificant for AM vs. CH. Also, the average sales reaction is 0.948 vs. 0.47 for advertising for significant sales and advertising coefficients, respectively. A firms advertising outlay is twice as responsive to its previous level of sales than previous level of cashflow.

Going with the cashflow hypothesis, the influence of the OPEC cartel has changed advertising behavior for the domestic firms. The OPEC dummy variable tests for coefficient stability between the pre and post OPEC cartel periods. According to Gujarati,!27, 28^ it is a surrogate for the Chow F-test on the coefficients between the two periods, and it does not matter which of the binary digits is used in the first or second periods. All the dummy variable coefficients in the non-leadership hypothesis, and 2 of 3 in the leadership hypothesis are significant. Multiplicative forms of the dummy variable with the independent variables--advertising, cashflow, and sales, can isolate the difference between the subperiods. Here we investigate differences of the additive form only, picking up only proportional effects on all the domestic firms.

Overall, the model with advertising, cashflow, and sales indicate that firms do not follow General Motors alone in setting their advertising policies. In the case of AM vs. CH under the non-leadership hypothesis, we see that where the cashflow coefficient seems to falter, the sales coefficient seeks to mend. Again, because the dummy variable coefficients are significant for both Ford and Chrysler as imitators of General Motors within the cashflow hypothesis with multicolinearity adjustment, and because three out of four dummy coefficients are significant for the cashflow hypothesis for the non-leader case, we may point the finger on the OPEC cartel for causing this disruption of domestic ad policies.

Multicolinearity correction has its price. Except for AM vs. GM, variations in both the advertising and the dummy coefficients remained flat as expected. The constant term increased by large amounts, is highly significant for the cashflow model with multicolinearity corrections, and is insignificant for the sales model with multicolinearity correction when AM is involved, viz., in three out of twelve cases. One benefit stands out from the procedure: Cashflow makes a dramatic turn around in coefficient magnitude and significance. In effect, we have traded some bias for efficiency in the error estimate of the cash flow coefficients. Better yet, if we factor in that the dimension of the constant term is ambiguous anyway, the bias will carry less importance!19^ p. 226. Overall, therefore, the multicolinearity correction has improved the model.

The !R.sup.2^ and D.W. statistics validate the model ability to predict reactions. In general, the adjusted !R.sup.2^ are high, mostly in the 90s, and do not show much deviation from the unadjusted one, indicating no problem with sample size. Only the AM vs. GM Durbin-Watson estimate falls in the indeterminate range. All the other Durbin-Watson statistics are significant at the 95 percent confident interval, a result of the first order autocorrelation procedure employed.

Tables 1R and 2R show the converse reactions to Tables 1 and 2. The reverse reactions are obviously deficient. Even with multicolinearity corrections, the overall performance did not improve. Nevertheless, these two tables are necessary adjuncts to the first two for the overall finding. The weak results they portray narrows the research towards the G.M. leadership and non-leadership hypotheses of Tables one and two as the proper domain for ad rivalry investigations.

G. Conclusion

The results do not falsify Bain's advertising hypothesis with price collusion, modeled around the cashflow affordability hypothesis when multicolinearity is eliminated. We observed significant advertising coefficients -- 20 out of 24, and an equal number of significant cashflow as sales coefficients -- 5 out of 6. It would be hair splitting to argue that one cashflow coefficient checks out only at the 95 percent confidence level, while all the significant sales coefficients check out at the 99 percent level.

The significant OPEC dummy coefficients -- two of four in the General Motors's leadership model and three of four in the non-leadership model, indicating a significant difference between advertising rivalry before and after the initial OPEC period, do not falsify Bain's advertising conjecture. Although OPEC has impacted the firms cashflow, and in some years rendered it negative, substituting sales for cashflow in those years did not change the reaction patterns. While OPEC may have TABULAR DATA OMITTED elevated Ford to a leadership position in advertising rivalry, the advertising coefficients remained significant in large numbers as indicated, and their levels indicate

healthy rivalry, indicating that U.S. firms are robust about their advertising rivalry. OPEC influence must have been proportional on the U.S. firms. Some of it may even be substituted away.

The finding in this paper can be disaggregated to the media level. Following Ramrattan,!39^ we expect to find television, spot, network, and cable media, because of their visual impact, to be strong staging grounds for rivalry. However, disaggregating the model to media and brand levels entails data sparcity problems. Cable television data do not go back far in time and brand data even without the media dimension has many zero cells. Also, the brand data are inconsistent over time.

In another vein, the analysis could be expanded to higher dimensions. We have modeled the data within a Seemingly Unrelated Regression (SUR) environment. Because the four firms belong to the same industry, one expects some inter-correlation in their error terms due to the presence of latent industry variables.!11^ With the four firms, the SUR resulted in mostly singular matrix solutions. Where estimation was possible, the statistics were poor, indicating low inter-correlation among firm's error terms.

Besides SUR, preliminary estimates were TABULAR DATA OMITTED done on market share, and with system of equations. The results were statistically weak, and therefore not developed further. Using market share for cashflow or sales, or pitting domestic firms's advertising against foreign firms's market share in the U.S. did not yield significant results. In a similar vein, the system of equations model did not improve the rivalry hypothesis. Mainly because feedbacks are filtered only through the leadership and non-leadership cases and not through their counterparts of Tables 1R and 2R, the system model did not perform well.

References

1. Anderson, R. M., An Elementary Core Equivalence Theorem, Econometrica, 46, pp. 1483-1487, 1978.

2. Anderson, R. M., A Market Value Approach to Approximate Equilibria, Econometrica, 50, pp. 127-136, 1982.

3. Anderson, R. M., Notions of Core Convergence, in W. Hildenbrand and A. Mas-Collel (editors), Contributions to Mathematical Economics, Holland: North-Holland, 1986, pp. 25-46.

4. Bain, Joe S., Industrial Organization, New York: John Wiley and Sons, Inc., 1959, Third printing 1967.

5. Bain, Joe S., Price Theory, New York: John Wiley and Sons, Inc., 1952, Third Printing, 1967.

6. Bain, Joe S., Barriers to New Competition, M.A.: Harvard University Press, 1956.

7. Balch, Michael, Oligopolies, Advertising, and Non-Cooperative Games, in Kuhn, H. W., and Szego, G. P., Differential Games, and Related Topics, N.Y.: North-Holland Publishing Company, 1971.

8. Baumol, William J., Business Behavior, Value and Growth, N.Y.: Harcourt, Brace and World, Inc., 1967.

9. Bertrand, Joseph as discussed by Edgeworth, Francis Y., The Pure Theory of Monopoly, Papers Relating to Political Economy, I, London: Macmillan and Co., Ltd., 1925, pp. 111-42.

10. Bhagwati, Jagdish, ed., Essays in International Economic Theory, Mass.: MIT Press, 1983, Vol. II.

11. Box, George E. P., Use and Abuse of Regression in Tiao, George C., The Collected Works of George E. P. Box, Belmont: Wadsworth Advance Books and Software, 1984.

12. 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.

13. Chamberlin, Edward H., The Theory of Monopolistic Competition, Mass.: Harvard University Press, 1965.

14. Cournot, A. A., Researches into the Mathematical Principles of the Theory of Wealth, trans. by Nathaniel T. Bacon, New York: Macmillan Co., 1929.

15. Debreu, Gerard, Economic Theory in the Mathematical Mode, The American Economic Review, June 1984, p. 269.

16. Debreu, Gerard, Four Aspects of the Mathematical Theory of Economic Equilibrium, in Mathematical Economics, New York: Cambridge University Press, 1983.

17. Debreu, Gerard, and Scarf, H., A Limit Theorem on the Core of and Economy, International Economic Review, September 1963, 4, pp. 235-46.

18. Debreu, Gerard, and Scarf, H., The Limit of the Core of an Economy, in McGuire, C. B., and Radner, Roy, eds., Decision and Organization: A Volume in Honor of Jacob Marschak, Amsterdam, 1972.

19. Dhrymes, Phoebus J., Econometrics, New York: Springer-Verlag, 1994.

20. Dubey, P., and Shubik, M., Trade and Prices in a Closed Economy and Exogenous Uncertainty and Different Levels of Information, Econometrica, 45, pp. 1657-1680.

21. Eiteman, et. Al., Multinational Business Finance, New York: Addison-Wesley Publishing Company, 1992, 6th Edition.

22. Friedman, J. W., Oligopoly and the Theory of Games, Amsterdam: North Holland, 1977.

23. Giliberto, Michael, Interest Rate Sensitivity in the Common Stocks of Financial Intermediaries: A Methodological Note, Journal of Financial and Quantitative Analysis, Vol. 20, No. 1, March 1985, pp. 123-126.

24. Grabowski, Henry G., and Mueller, Dennis C., Imitative Advertising in The Cigarette Industry, The Antitrust Bulletin, Vol. XVI, Summer 1971, pp. 257-292.

25. Grabowski, H., and Mueller, D. C., Non-Price Competition in the Cigarette Industry", The Antitrust Bulletin, XIV, 1969, pp. 607-628.

26. Grabowski, H., and Mueller, D. C., Non-Price Competition in the Cigarette Industry: A Comment, The Antitrust Bulletin, XIV, 1970, pp. 679-686.

27. Gujarati, D., The American Statisticians, February, 1970, pp. 50-52.

28. Gujarati, D., Basic Econometrics, New York: McGraw Hill Book Company, 1988, pp. 446-448.

29. Harsanyi, John C., Rational Behavior and Bargaining Equilibrium in Games and Social Situations, New York: Cambridge University Press, 1977, p. 101.

30. Jastram, Roy W., Advertising Outlays under Oligopoly, The Review of Economics and Statistics, Vol. XXXI, May, 1949, pp. 106-109.

31. Johansen, Leif, A Calculus Approach to the Theory of the Core of an Exchange Economy, The American Economic Review, Vol. 68, Number 5, December 1978, pp. 813-820.

32. Luce, R. Duncan, and Raiffa, Howard, Games and Decisions, N.Y.: McGraw-Hill Book Co., 1952.

33. Mas-Colell, Andreu, Perfect Competition and the Core, Review of Economic Studies, 1982, XLIX, pp. 15-30.

34. Modigliani, Franco, New Developments on the Oligopoly Front, Journal of Political Economy, Vol. LXVI, June 1958.

35. Nash, John, Equilibrium Points in N-Person Games, Proceedings of the National Academy of Sciences of the U. S. A., 1950, 36, pp. 48-49.

36. Neuberger, Jonathan, A., Risk and Return in Banking: Evidence from Bank Stock Return, Economic Review, Federal Reserve Bank of San Francisco, Fall 1991, No. 4, pp. 18-30.

37. Okuguchi, Koji and Szidarovszky, Ferenc. The Theory of Oligopoly with Multi-Product Firms, New York: Springer-Verlag, 1990.

38. Pindyck, Robert S., and Rubinfeld, Daniel L., Economic Models and Economic Forecasts, New York: McGraw Hill, Inc., Third Edition, 1991.

39. Ramrattan, Lall B., Rivalry in the U. S. Automobile Industry: An Advertising Approach, 1986, Unpublished Ph.D. Dissertation, The New School for Social Research.

40. Ramrattan, Lall B., The Impact of Imports on Price Competition in the Automobile Industry, The American Economist, Fall 1991.

41. Sacco, P. L., Adaptive Response and Cournotian Behavior. The Case of Linear Demand with Two Firms, Economic Notes, Vol. 20, No. 3, 1991.

42. Schelling, T., The Strategy of Conflict, Mass.: Harvard University Press, 1960.

43. Schnabel, Morton, A Model of Cigarette Advertising, The Antitrust Bulletin, Fall-Spring, 1973.

44. Shapiro, Alan C., Multinational Financial Management, Boston: Allyn and Bacon, 1991, 4th Edition.

45. Sherman, Rogers, Oligopoly: An Empirical Approach, Mass.: Lexington Books, 1972.

46. Shubik, Martin J., and Levitan, Richard, Market Structure and Behavior, M. A.: Harvard University Press, 1980.

47. Shubik, Martin J., Strategy and Market Structure, N. Y.: John Wiley and Sons, Inc., 1959.

48. Shubik, Martin, J., A Game Theoretic Approach to Political Economy, Mass.: The MIT Press, 1984.

49. Telser, Lester G., Competition, Collusion, and Game Theory, Chicago: Aldine Atherton, Inc., 1972.

50. Telser, Lester G., Economic Theory and the Core, Chicago: University of Chicago Press, 1978.

51. Wilson, Charles E., Senate Armed Forces Committee, 1952.
联系我们|关于我们|网站声明
国家哲学社会科学文献中心版权所有