¹ Throughout this article, I use the term “supplier” to encompass both financially independent downstream firms and vertically integrated divisions.

² Williamson, Oliver E., Markets and Hierarchies: A Study in the Economics of Internal Organization (New York, 1975Google Scholar); Williamson, , The Economic Institutions of Capitalism: Firms, Markets, Relational Contracting (New York, 1985Google Scholar); Thompson, James D., Organizations in Action (New York, 1967Google Scholar); quote from Economic Institutions of Capitalism, 76.

³ By “structure” I refer to the nature of relationships among firms (as defined in the next section) as well as traditional industrial organization variables such as concentration ratio, barriers to entry, and rate of technical change.

⁴ Hirschman, Albert O., Exit, Voice, and Loyalty: Responses to Decline of Firms, Organizations, and States (Cambridge, Mass., 1970Google Scholar).

⁵ I use the term “technical progress” very broadly; I mean it to encompass invention, innovation, and diffusion by both supplying and buying firms. I specifically wish to include organizational innovations such as just-in-time inventory techniques and quality assurance methods such as statistical process control.

⁶ See, for example, Williamson, Markets and Hierarchies, and the articles cited in Kaserman, David L., “Theories of Vertical Integration: Implications for Antitrust Policy,” Antitrust Bulletin 25 (Fall 1978): 105–28Google Scholar.

⁷ Flaherty, M. Therese, “Prices vs. Quantities and Vertical Integration,” Bell Journal of Economics 12 (Autumn 1981): 507–25CrossRefGoogle Scholar.

⁸ See Helper, Susan, “Supplier Relations and Technical Change: Theory and Application to the US Auto Industry” (Ph.D. diss., Harvard University, 1987Google Scholar), and Helper, , “An Exit-Voice Analysis of Supplier Relations,” in Morality, Rationality, and Efficiency: New Perspectives on Socio-economics, ed. Coughlin, Richard M. (New York, 1991Google Scholar), for more detail.

⁹ The terms “exit” and “voice” originated with Hirschman, Exit, Voice, and Loyalty. I have generalized his analysis to include cases where the resolution of problems requires not only more effort by the parties involved, but also irreversible investments in physical and organizational capital.

¹⁰ Stanford Business School Case, “Signetics (A),” 15, and “Signetics (B),” 8, Stanford, Calif., 1982.

¹¹ For an alternative conception of incentive systems in supplier relationships, see O'Hare, M., Leone, Robert, and Zegans, M., “Privatization of Prisons: A Managerial Perspective” in Private Prisons and the Public Interest, ed. McDonald, Douglas C. (New Brunswick, N.J., 1988Google Scholar).

¹² For formal analysis of this proposition, see Helper, “Supplier Relations and Technical Change,” chap. 6, and Helper, and Levine, David I., “Long-Term Supplier Relations and Product-Market Structure,” Journal of Law, Economics, and Organization 8 (Oct. 1992): 561–81Google Scholar.

¹³ See Helper, “An Exit-Voice Analysis,” and Helper, “Supplier Relations and Technical Change,” for details.

¹⁴ Of course, some technical changes (for example, the invention of the personal computer) require neither very much capital nor very much buyer-specific information exchange. However, these types of technical changes are less likely to be practical in a mature industry such as automobiles, in which huge investments in physical and organizational capital specific to a particular way of doing things have been built up over time. See Abernathy, William J., The Productivity Dilemma: Roadblock to Innovation in the Automobile Industry (Baltimore, Md., 1978Google Scholar); Abernathy, William J. and Wayne, Kenneth, “The Limits of the Learning Curve,” Harvard Business Review 52 (Sept.-Oct. 1974): 109–19Google Scholar; Abernathy, and Clark, Kim, “Mapping the Winds of Creative Destruction,” Research Policy 14 (Feb. 1985): 3–22CrossRefGoogle Scholar; and Clark, Kim, “Managing Technology in International Competition: The Case of Product Development in Response to Foreign Entry,” Harvard Business School Working Paper (Boston, Mass., 1985Google Scholar).

¹⁵ As this article will show, the exit system of supplier relations not only changed the locus of innovation from suppliers to customers; it also reduced the amount of innovation in the industry as a whole.

¹⁶ For more on the ability of dominant firms to alter the cost curves they face, see Lazonick, William, Business Organization and the Myth of the Market Economy (New York, 1991Google Scholar).

¹⁷ Thanks to Kim Clark and Albert Hirschman for discussion of this possibility.

¹⁸ On U.S. managers' short time horizon, see, for example, Hayes, Robert and Abernathy, William, “Managing Our Way to Economic Decline,” Harvard Business Review 58 (July-Aug. 1980): 67–77Google Scholar. In contrast, supplier relations in the Japanese auto industry seem better characterized as a case of continuous voice, as Susan Helper, “Comparative Supplier Relations in the US and Japanese Auto Industries,” Business and Economic History, 2d ser. (1990): 153–62, shows.

¹⁹ Seltzer, Lawrence H., A Financial History of the American Automobile Industry (Boston, Mass., 1928), 19–20Google Scholar.

²⁰ “Silver Anniversary Issue,” Automotive Trade Journal (1924), 15; Sloan, Alfred P. Jr, Adventures of a White-Collar Man, ed. Sparkes, Boyden (New York, 1941), 9, 60Google Scholar.

²¹ Nevins, Allan and Hill, Frank, Ford: The Times, The Man, The Company (New York, 1954), 232Google Scholar; Nevins, and Hill, , Ford: Expansion and Challenge (New York, 1957), 22–23, 88ff., 173–74Google Scholar; Rosenberg, Nathan, Perspectives on Technology (New York, 1976CrossRefGoogle Scholar); Langlois, Richard and Robertson, Paul, “Innovation and Vertical Integration in the American Automobile Industry, 1900–1940,” Working Paper, Department of Economics and Management, University College, University of New South Wales, Canberra, Australia, 13Google Scholar.

²² Sloan, Adventures of a White-Collar Man, 37–39; see also Nevins and Hill, Ford: The Times, The Man, The Company, 212.

²³ Seltzer, A Financial History, 89–90, 100.

²⁴ Chandler, Alfred D. Jr, Strategy and Structure: Chapters in the History of the American Industrial Enterprise (Cambridge, Mass., 1962), 118Google Scholar; Sloan, Adventures of a White-Collar Man, 29; Thompson, G., “Intercompany Technical Standardization in the Early American Automobile Industry”, Journal of Economic History 14 (Winter 1954): 1–20CrossRefGoogle Scholar.

²⁵ Sloan, Adventures of a White-Collar Man, 92–93.

²⁶ Sloan reported that the owners of Dayton Engineering Laboratories (later GM's Delco Division) had similar feelings. Ibid., 98–99.

²⁷ Dodge v. Ford Motor Company, 1070 Mich Sup. Ct. Briefs and Records, cited in Seltzer, A Financial History, 101.

²⁸ Sloan, Adventures of a White-Collar Man.

²⁹ Nevins and Hill, Ford: The Times, The Man, The Company, 460.

³⁰ Klein, Benjamin, Crawford, Robert G., and Alchian, Armen A., “Vertical Integration, Appropriable Quasi-Rents, and the Competitive Contracting Process,” Journal of Law and Economics 21 (1978): 297CrossRefGoogle Scholar; Chandler, Alfred D. Jr, and Salsbury, Stephen, Pierre S. du Pont and the Making of the Modern Corporation (New York, 1971Google Scholar).

³¹ Langlois, Richard and Robertson, Paul, “Explaining Vertical Integration: Lessons from the American Automobile Industry,” Journal of Economic History 49 (June 1989): 361–75CrossRefGoogle Scholar.

³² Nevins and Hill, Ford: The Times, The Man, The Company, 461.

³³ P. Heldt, “Parts Makers' Role Gets Bigger as Automotive History Unfolds,” Automotive Industries, 6 May 1933, 546–48, 554, argues that in the 1920s, independent suppliers made significant improvements in parts including brakes, crankshafts, pistons, valves, springs, carburetors, radiators, clutches, steering gears, and axles.

³⁴ Langlois and Robertson, “Innovation and Vertical Integration,” 22–24; Langlois and Robertson, “Explaining Vertical Integration,” 374–75.

³⁵ Graham, Margaret B. W. and Pruitt, Bettye, R&D for Industry: A Century of Technical Innovation at ALCOA (New York, 1990), 148Google Scholar; quote from Nevins and Hill, Ford: Expansion and Challenge, 447.

³⁶ In the days of single sourcing, an important union goal in the case of a strike was to prevent the transfer of automaker-owned tooling to another plant, leading to “games of ‘hide-and-seek’ along Detroit side streets, with union members in pursuit of the transfer trucks.” Alexander, Kenneth, “Market Practices and Collective Bargaining in Automotive Parts,” Journal of Political Economy 69 (Feb. 1961): 17CrossRefGoogle Scholar. Strikes were a frequent occurrence in the early postwar period; for example, GM reported that in mid-May 1946, 142 of its suppliers were affected by strikes. Katz, Harold, The Decline of Competition in the Automobile Industry, 1920–1940 (New York, 1977), 263Google Scholar.

³⁷ Herzenberg, Stephen, “Towards a Cooperative Commonwealth: Labor and Restructuring in the U.S. and Canadian Auto Industries” (Ph.D. diss., Massachusetts Institute of Technology, 1991), 225Google Scholar.

³⁸ This section and the next are a condensed, updated version of Helper, “Supplier Relations and Technical Change,” chaps. 4 and 5. The information is drawn from an extensive reading of the trade press and of secondary sources, and from interviews with auto industry experts between 1984 and 1989. I had formal interviews with fifty-eight people, ranging in rank from purchasing agent to division president and including representatives of the three leading U.S. automakers, seven suppliers, and the United Auto Workers. (Requests for confidentiality preclude identifying these sources by name.) In all interviews, I started with the same list of open-ended questions (reproduced in Helper, “Supplier Relations and Technical Change,” chap. 5, appendix), and then asked additional questions regarding the interviewee's areas of expertise.

³⁹ This typology is a simplification, of course; there were important variations across buyers, across suppliers, and over time. However, my interviews did indicate that most supplier relationships seemed to cluster into the three groups I have defined, and that the rules of the game for each group were fairly stable during the period under study.

⁴⁰ This is not to say that there was little transfer of information between buyer and supplier. Before a supplier was awarded a contract, it would have to allow the automaker to inspect its facilities and provide general information on its production costs and financial status. Once the contract for a particular input was signed, the buyer provided additional product information in the form of engineering change orders.

The foregoing indicates that the simple exit mode of supplier relations, though offering the automaker a great deal of opportunity for exit compared to the other modes discussed in this article, provides less such opportunity than would a Walrasian market. In Léon Walras's conception, the only information that crosses firm boundaries is price and product characteristic information broadcast simultaneously from and to many buyers and suppliers. The reason for the difference is that even in the simple exit mode, automakers' investments are slightly specific to a particular group of suppliers. Because an automaker cannot instantaneously switch to another supplier if one falls through (and because the law does not provide for full compensation of the automaker should this situation arise), the automaker must gather information about suppliers to minimize this possibility.

⁴¹ This estimate is very crude. It is based on 1) the finding in Kamath, R. and Wilson, C., “Characteristics of the U.S. Automotive Supplier Industry,” Joint U.S.-Japan Auto Study, no. 10 (Ann Arbor, Mich., 1983Google Scholar) that in 1980, the top forty auto suppliers (.8 percent of the total) made about 31 percent of the total volume of $34 billion in sales of “parts, components, and raw materials” to U.S. automakers whose producers the authors were able to identify; and 2) the observation that in general, it was only the largest suppliers who had much interaction with the automakers. Therefore, the bottom 99.2 percent of suppliers were in the simple exit mode.

⁴² Supplier and automaker interviews, October 1984 and August 1985.

⁴³ Written, legally binding contracts lasting longer than one year were extremely rare, however. See Porter, Michael E., Cases in Competitive Strategy (New York, 1983Google Scholar), and University of Michigan Transportation Research Institute, “The International Automotive Challenge: Strategy Beyond Cost and Quality,” UMTRI Research Review 17 (Sept.-Dec. 1986): 1–48Google Scholar.

⁴⁴ Interview, October 1984.

⁴⁵ Both buyers and suppliers agreed that the automakers were more often guilty of this practice than were suppliers.

⁴⁶ Supplier interviews, October 1984. When asked to comment on this incident, a Ford executive freely admitted that in the past, “we cut the legs out from more than one supplier” (August 1985).

⁴⁷ The commitment they received from the automakers was an important reason why these firms were willing and able to make the investments necessary to remain technically capable. A hypothesis I will test in future research is that this commitment is what allowed these firms to become technically capable in the first place—that is, that the firms that were innovative independent suppliers in 1970 were remnants of the automakers' previous use of the voice strategy in the first decades of this century.

⁴⁸ Interview with Ford purchasing staff executive, August 1985; see also Clark, Kim B. and Fujimoto, Takahiro, Product Development Performance: Strategy, Organization, and Management in the World Auto Industry (Boston, Mass., 1991Google Scholar).

⁴⁹ Wright, J. Patrick, On a Clear Day You Can See General Motors (New York, 1979), 49Google Scholar.

⁵⁰ Keller, Maryann, Rude Awakening: The Rise, Fall, and Struggle for Recovery of General Motors (New York, 1989), 104Google Scholar.

⁵¹ Interview with GM manager at a former Fisher plant, 4 December 1990; interview with James Meehan, professor of economics, Colby College, April 1986. This example illustrates that the ability of a vertically integrated firm to “effect internal adaptations by fiat” (Williamson, Economic Institutions of Capitalism, 75) is not automatic; incentive systems must be carefully established and maintained. Similarly, the example illustrates that the costs of haggling and the danger of a “hold-up” may be greater with a powerful internal supplier than with an outside firm.

⁵² E. Raymond Corey, “General Motors (A): Organization of the Procurement Function,” Harvard Business School Case 9–576–251 (1977), 15; interview with manager at ex-Fisher plant, 4 December 1990. In 1971, the Chevrolet Vega was unable to meet emissions standards. GM's Rochester Products Division refused to help with the problem, so Chevrolet had to add an air pump to help exhaust gases burn more effectively. Holley Carburetor was willing to work with Chevrolet on the problem and designed a carburetor that eliminated the need for the pump and saved Chevy $3 million per year. GM management then awarded the contract to produce the Holley-designed carburetor to Rochester Products. Only after strenuous protests by DeLorean was Holley “allowed to keep a little piece of the business.” Wright, On a Clear Day, 78–79.

At Chrysler, “blind ambition” drove its Dayton plant to claim that it could make radiators, even though it had no relevant experience and the equipment it was proposing to buy was obsolete. Chrysler eventually rejected the plan, but only at the highest possible level; other disputes were not resolved so satisfactorily; interview with Chrysler engineer, August 1985.

⁵³ Knight, A., “Automotive Report,” Smith, Barney, Harris, Upham and Co., New York, 10 Feb. 1983Google Scholar.

⁵⁴ Chandler, Strategy and Structure; Chandler, Alfred D. Jr, Giant Enterprise: Ford, General Motors, and the Automobile Industry (New York, 1964Google Scholar).

⁵⁵ Crandall, Robert, “Vertical Integration in the United States Automobile Industry” (Ph.D. diss., Northwestern University, 1968Google Scholar).

⁵⁶ Knight, “Automotive Report.” In contrast, the automakers have not been nearly so successful at protecting their rents from their suppliers of labor. For most of the period since 1913, when Henry Ford's $5 day was double the prevailing manufacturing wage, the automakers have paid their workers wages above the manufacturing average. Between 1948 and 1980, the ratio of auto assembler wages to the average production worker's wages ranged from 1.17 to 1.55. See Katz, Harry, Shifting Gears: Changing Labor Relations in the U.S. Automobile Industry (Cambridge, Mass., 1985Google Scholar), Table 2.1.

⁵⁷ At Chrysler in the 1970s, the cost of expected midstream changes accounted for 20–30 percent of automotive program budgets. Wall Street Journal, 23 Feb. 1988, 24.

⁵⁸ Barriers to entry into some industries could be quite low. In the late 1960s, for example, “a simple [metal] finishing operation could be started in rented quarters for under $10,000.” Arnett, H. and Smith, D., The Metal Finishing Industry, Michigan Business Reports, n.s. no. 10 (Ann Arbor, Mich., 1977), 3Google Scholar.

⁵⁹ Interview, October 1984; written comments by a Ford executive on a 1986 draft of this manuscript.

⁶⁰ Porter, Cases in Competitive Strategy; interview with executives at Fleck Manufacturing, a small ($5 million sales) Canadian wiring-harness maker, May 1987.

⁶¹ Teece, David, “Profiting from Technological Innovation: Implications for Integration, Collaboration, Licensing, and Public Policy”, Research Policy 15 (Dec. 1986): 285–305CrossRefGoogle Scholar.

⁶² Chandler, Giant Enterprise.

⁶³ White, Lawrence J., The Automobile Industry since 1945 (Cambridge, Mass., 1971CrossRefGoogle Scholar); Moritz, Michael and Seaman, Barrett, Going for Broke—The Chrysler Story (New York, 1981Google Scholar).

⁶⁴ Arnett, H. and Smith, D., The Tool and Die Industry, Michigan Business Reports, n.s. no. 1 (Ann Arbor, Mich., 1974), 28 and 39Google Scholar. Arnett and Smith report that the “vast majority” of shop owners they surveyed considered historical costs only in preparing bids.

⁶⁵ Arnett and Smith, The Metal Finishing Industry, 27.

⁶⁶ Porter, Cases in Competitive Strategy.

⁶⁷ Abernathy, Productivity Dilemma.

⁶⁸ Sloan, Alfred P. Jr, My Years with General Motors (Garden City, N.Y., 1963), 69Google Scholar, cited in Abernathy, Productivity Dilemma, 34. See also Leslie, Stuart W., Boss Kettering (New York, 1983), 123–47Google Scholar, and Chandler, Strategy and Structure, 154.

⁶⁹ Abernathy, Productivity Dilemma, 35. It may seem unlikely that an incident that occurred nearly seventy years ago would have much influence on corporate behavior today. But this example was cited as late as 1987 by GM engineers as a reason to avoid technological risks. (Conversation with Robert Kazanjian, University of Michigan, reporting on a class he had taught at the GM Tech Center, Feb. 1987.)

⁷⁰ Abernathy, William J., Clark, Kim B., and Kantrow, Alan M., Industrial Renaissance: Producing a Competitive Future for America (New York, 1983Google Scholar); Cole, Robert E. and Yakushiji, Tazio, eds., The American and Japanese Auto Industries in Transition (Ann Arbor, Mich., 1984Google Scholar).

⁷¹ Interview, August 1985.

⁷² Ibid.

⁷³ Interview with Chrysler purchasing executive, August 1985.

⁷⁴ Interview, May 1988.

⁷⁵ General Motors president F. James McDonald in University of Michigan Transportation Research Institute, “The International Automotive Challenge,” 12.

⁷⁶ Wall Street Journal, 31 July 1984, 1; Ford interviews, August 1985.

⁷⁷ Auto industry executives' use of this concept seemed very similar to that now in vogue among game theorists; see, for example, Kreps, David and Spence, A. Michael, “Modelling the Role of History in Industrial Organization,” Harvard University Discussion Paper 992 (July 1983Google Scholar). For example, I interviewed in October 1984 a supplier who had GM-owned tooling at his plant that would take nine months and cost $4–5 million to move to another supplier. When asked why he did not try to raise his price at contract renegotiation time, he said, “Because our reputation would be shot to hell. We might get away with it once, but nobody in the business would ever deal with us again; they'd say, ‘You're nothing but a bunch of dishonest crooks.’”

⁷⁸ Signetics (B), 5, 7–8.

⁷⁹ Ward's Auto World, June 1987 and Nov. 1988, 59; Mary Connelly, “Acustar stronger since Chrysler plan to sell it fell apart,” Automotive News, 6 March 1989, 20.

⁸⁰ Michigan Business, April 1987, 71; Michelle Krebs, “GM ready to merge 5 parts units into 2,” Automotive News, 6 March 1989, 3, 6.

⁸¹ Modern Michigan, Winter 1989, 13.

⁸² Robert M. Sinclair, vice-president, engineering, at Chrysler, in University of Michigan Transportation Research Institute, “the International Automotive Challenge,” 22. In the automotive industry as a whole, half of all engineering was done by engineering service firms, according to Ralph Miller, president, Modern Engineering Service Company (ibid., 31). Cash-rich GM bucked this trend with its purchases of Electronic Data Systems (for $2.5 billion in 1984) and Hughes Aircraft Company (for $5 billion in 1985), acquisitions that significantly increased its computer and electronics capabilities. (Fortune, 8 July 1985, 22).

⁸³ Wall Street Journal, 23 Feb. 1988, 24.

⁸⁴ Interview with a business manager at Chrysler's Component Business Operation (since renamed Acustar), August 1985.

⁸⁵ Gillett, Frank, “The Integrating Supplier” (MA Thesis, Massachusetts Institute of Technology, 1992Google Scholar).

⁸⁶ Interview with Daniel L. Luria, senior researcher, Industrial Technology Institute, Ann Arbor, Mich., February 1987.

⁸⁷ Helper, Susan, “Supplier Relations and Investment in Automation: Results of Survey Research,” Center for Regional Economic Issues, Case Western Reserve UniversityGoogle Scholar; Helper, , “Have Things Really Changed between Automakers and Their Suppliers?” Sloan Management Review 32 (Summer 1991): 15–28Google Scholar.

⁸⁸ Chemical Week, 11 Nov. 1987, 10; Wall Street Journal, 23 Feb. 1988, 22. Similar figures could be cited for General Motors and Chrysler.

⁸⁹ Knight, “Automotive Report,” 17.

⁹⁰ Chrysler interviews, August 1985.

⁹¹ As Langlois and Robertson, “Explaining Vertical Integration,” point out, product innovation often has these features, whereas (many types of) process innovation are characterized by the need to work out details centrally.

⁹² If corporate managers' decisions are not likely to be much better than those of their suppliers, then high administrative coordination only slows down decision making and diverts managerial attention. For this reason, a GM engineer said in 1987, “If I had the choice, I would work with outside suppliers all the time. They're more efficient, because they just go away and work on it, and bring it back when it's done. Then our testing department tests it, and we use it. If we do it internally, there's all this hassle over things like which group is going to do the project, whose budget it comes out of, and what department tests it.” (Possibilities for testing included the producing department, the using department, and the testing department.) See Williamson, Economic Institutions of Capitalism, 161, for further discussion of the difficulties of vertical integration “with [market] incentives held constant.”

⁹³ For a game-theoretic analysis of how a change in the buyer's threat point due to a change in final-product market structure can lead to a change in supplier relations strategies from exit to voice, see Helper and Levine, “Long-Term Supplier Relations.”

⁹⁴ General Motors, “Report on Conference Held at Jackson Community College, Jackson, Michigan” (1984), photocopy.

⁹⁵ Dana Milbank, “Air-Bag Woes Knock Wind Out of TRW,” Wall Street Journal, 12 March 1991, A4.

⁹⁶ Supplier interview, November 1984.

⁹⁷ See Helper, “Supplier Relations and Technical Change,” chap. 5, for more detailed discussion.

⁹⁸ Chrysler interviews, August 1985.

⁹⁹ See Roy D. Shapiro, “Toward Effective Vendor Management: International Comparisons,” Harvard Business School Working Paper (1985).

¹⁰⁰ By early 1989, there were 232 Japanese-owned or joint-venture plants in the United States in operation or in final planning stages. Of these, 85 had contracts with at least one of Ford, GM, or Chrysler. Joseph M. Callahan, “232 Japanese Supplier Plants Gunning for Your Business,” Automotive Industries, Feb. 1989, 89–112.

¹⁰¹ Andrea, David, Hervey, R., and Luria, Dan, “The Capacity Explosion: Implications for Michigan Suppliers,” AIM Newsletter 1 (March 1986): 6–8Google Scholar; interview with Daniel D. Luria, senior researcher, Industrial Technology Institute, Ann Arbor, Mich., August 1987.

¹⁰² Shaiken, Harley and Herzenberg, Stephen, Automation and Global Production: Automobile Engine Production in Mexico, the United States, and Canada (La Jolla, Calif., 1987Google Scholar), show that cost is significantly less, and quality and machine uptime levels approach U.S. levels at a U.S. automaker's engine plant in Mexico compared with one in the United States making exactly the same (fully de-bugged) product. On the difficulties of coordinating production between the United States and Mexico, see Roy D. Shapiro, “Rio Bravo Electricos,” Harvard Business School Case (1982).

¹⁰³ Akin to what Piore, Michael and Sabel, Charles, The Second Industrial Divide: Possibilities for Prosperity (New York, 1984Google Scholar), would call an “industrial district.”