Making cars and making money in the interwar automobile industry: economies of scale and scope and the manufacturing behind the marketing.
58 In this period, compensation was determined by factory management and not at the corporate level (personal communication to author from Mustafa Mohateram, General Motors Corporation Economics). Statistics permitting comparison of Chevrolet employee earnings to those of Ford employees at this stage are not available: GM did not make public such statistics even at the divisional level, let alone numbers for employees In the popular mind, the American automobile industry discovered mass production during the half-decade preceding the First World War and then spent the years between the two wars discovering marketing. Alfred Sloan's introduction of annual model changes, in this account, was the central and indeed the only important innovation in the business of making and selling automobiles during the interwar period.
This impression of the industry is reinforced by the historical literature. The most recent general works--by John Rae and James Flink--offer a picture of the interwar years no different for present purposes than that conveyed in E.D. Kennedy's 1941 study, in the relevant pages of Allan Nevins and Frank Hill's 1957 volume, or even in Alfred D. Chandler, Jr.'s Giant Enterprise, published over a quarter-century ago but still widely considered the authoritative documentary account.(1) In this view, Ford had grown to its dominant position through mass production; General Motors (GM) grew large through financial combination. GM became important through the development of a decentralized decision-making structure supported by systematically gathered data, and through two innovations in marketing--the development of a systematic product line policy ("a car for every purse and purpose") and a program of annual model changes. If the rise of Ford was the triumph of machine-tool builders and production supervisors, the rise of GM and the birth of the automobile oligopoly marked the triumph of the "white-collar man."(2) There are no references in Chandler's editorial text, and only two fragments of sentences in the excerpted documents, bearing on any underlying changes in the way cars were made.(3) Rae is even sketchier, and, though Flink observes that the practice of annual model changes, once established, had implications for optimal tool and die life, and so for design, he does not elaborate.(4)
The older literature does not entirely ignore changing production technology, but the focus is consistently on implications for the annual model change as a competitive strategy.(5) More recent works in the spirit of the history of technology do treat evolving production methods and equipment, but their emphasis is on the role of machines and of task definition in the provision of diversity rather than on the organization of production in itself.(6)
I argue in this article that such accounts can be quite misleading. The interwar period in fact saw a radical innovation in the organization of production within automobile firms, which played a fundamental role in the dramatic and enduring changes in competitive structure that became entrenched well before the Second World War. The innovations in marketing practice, like the changes in shop-floor technology that sustained them, were substantial and certainly left traces in public documents and in the public imagination. But as a successful commercial strategy, all these changes were predicated on--and indeed were so effective only because they could exploit--innovations inside the factories. GM's power as a mass producer, the example it set for Walter Chrysler in setting his own corporate strategy, the shake-out of firms that could not keep up with the annual model changes--all turned on the means by which the scale-based cost advantages of the Model T were initially subverted. Although the marketing strategy stands out in the popular mind, only the manufacturing behind it made the marketing shift possible.(7)
The first part of the article demonstrates that the idea of engineering coordination was on the minds of Pierre du Pont and Alfred Sloan even as GM's marketing strategy was first articulated. The next section lays out and analyzes certain details of the history of automobile production methods from the earliest days of Ford through Ford's initial perfection of mass production. The following section sketches the history of industrial organization in automobiles through GM's moment of innovation in the mid-l920s--that is, the competitive setting. The heart of the article then develops the details of how mass production came to General Motors and draws out the way it fitted into the marketing strategy. I then argue--from GM's behavior in the labor market, in the absence of direct primary evidence--that the production strategy remained in force, and effective, throughout the part of the interwar period not yet accounted for.
The Original Idea
In the late spring of 1921, not long after taking control of General Motors, Pierre du Pont instigated a study that set the strategic course of the corporation. The resulting proposals contained the famous idea of the "car for every purse and purpose" product line: "Chevrolet for the hoi polloi, Oakland ... for the poor but proud, Oldsmobile for the comfortable but discrete |sic~, Buick for the striving, |and~ Cadillac for the rich," as a writer for Fortune later put it.(8) But the segmented product line was not the only significant concept advanced. Sloan later wrote of the report:
The fundamental conception of the advantage to be secured in this business, we said, was expressed by co-operation and co-ordination of our various policies and divisions. It was natural to expect that co-ordinated operation of our plants should result in greater efficiency than was the case when the divisions were working at cross purposes, and the same could be said for engineering and other functions.... Under a plan of co-operation, the teamwork could thus attain increased volume and reduced cost.(9)
Keeping the formerly independent divisions from selling products that competed for customers (maximizing profits by optimizing the degree of competition, and hence total revenue) was clearly part of his vision, but the passage also suggests maximizing, profits by designing parts and production plans to minimize total cost. The 1923 Annual Report spoke somewhat more plainly about this: "Such a policy makes possible co-ordination, not otherwise practical, in engineering, manufacturing,... |as well as~ in distribution."(10)
Historians seem to have treated the engineering coordination suggested in the proposals as rhetoric or vague piety rather than as an injunction to the engineers and the production staff. The argument of this article is that production plans to minimize costs were a central part of the strategy, that in due course they were in fact implemented, and that they provided the practical foundation for making the marketing idea so profitable. The economies in question were substantial.
The Early Mass Production of Automobiles
The notion of manufacturing many superficially different products with a multitude of common parts now seems familiar. In the guise of the "flexible" mass production of automobiles, its invention is sometimes credited to the Japanese. Was it in fact a GM innovation? What were GM's fundamental innovations with respect to Ford's practice? An argument about this must begin with some background on Ford's methods and their diffusion.
It is well known that the production and assembly of automobiles in the early years of the American industry proceeded through essentially artisanal methods.(11) Before 1913 there were fragmentary prefigurations of a change, although the change itself was perceived by contemporaries as abrupt.(12) The watershed occurred in the Ford Motor Company's Highland Park, Michigan, plant between March and December 1913.(13)
The thrust of the Ford innovations was to routinize radically all but a handfull of the jobs at the plant. Work-force skills that were still crucial for producing autos elsewhere became progressively irrelevant to making cars at Ford, because Ford engineers built those skills into forges, foundries, and (ultimately) single-purpose machine tools that made completely interchangeable parts. Since this development virtually eliminated skilled fitting tasks whose timing was difficult to forecast, the pace of work throughout the factory became amenable to centralized managerial control to an unprecedented degree. The flow of materials through the factory was reorganized so that tasks came to the workers rather than vice versa, further expanding the possibilities for centralized control. Thus even the jobs not directly on the assembly lines came to be driven by the line's rhythms. Accomplishing all this required elaborate investments in physical capital, and the book value of the machines and tools on the Ford shop floor increased more than sevenfold between December 1909 and December 1913.(14)
Henry Ford first became a public figure at this stage, partly because of the armies of vehicles emerging from his factory as this production regime fell into place, and partly because he abruptly doubled the sums he was paying his work force at a time when the regional and national economies were beginning to lurch into a deep recession. I have argued elsewhere that Ford was alone in his industry in paying such wages when he introduced the five-dollar day because he was alone in the way he made automobiles.(15) That the Ford company was unique in this constellation of methods is clear from the contemporary trade journal literature on new factories, quite independent of the dramatic output figures for the Highland Park plant. Given that Ford had been making money for some time, this raises obvious questions. News articles brought the Highland Park plant prominence and renown. Yet there is no evidence of sympathetic wage increases in the five-dollar day's immediate (and even medium-term) aftermath, suggesting that even in the medium term (into the years of the First World War), the Ford production process went substantially unimitated.(16)
It may seem implausible that such a conspicuously profitable innovation did not diffuse swiftly. There is evidence, however, that the key to Ford's production success was not understood by contemporaries. Even cursory research reveals that moving assembly lines were in place in a number of other motor vehicle firms around 1916. Articles in the trade press suggested that the moving lines were responsible for Ford-style quantity production, generally constituted mass production methods in themselves, and offered all of Ford's opportunities for centralized control over pace.(17) There is good reason to doubt each of these claims. Output statistics by manufacturer directly rebut the first; my earlier discussion of the significance of routinization of tasks should discredit the second; a case study may address the third.(18)
Studebaker, with roots as a major producer of carriages and wagons, had early, though limited, experience of American System methods.(19) In the pre-First World War period, the company's managers and directors were consciously thinking about productivity improvements in their auto plants--for example, implementing progressive assembly and otherwise minimizing employees' excuses for wandering around the plant, attempting to plan plant workloads farther in advance, and introducing conveyor belts.(20) Surviving minutes of 1915 Finance Committee meetings, which detail rationales for each of these initiatives, indicate that Studebaker's managers did not think of the conveyors as production control devices. The meeting on conveyor belts approved putting in a number of them, including one for chassis. Among the advantages cited were eliminating five hundred workers from the payroll (about one in seven of the average monthly work force in the Detroit operation that year) and lowering works and inventory expense, as well as certain vaguely stated advantages of flexibility. It was remarked that "|t~his runway and these conveyors will more than pay for themselves in the first year's labor savings." Thus conveyor belts were seen as labor-saving devices rather than as sources of coordination and control. In the springtime of 1914, Studebaker had wanted to ramp up production. The company put its work force on piece rates and crowed to the directors the following summer that the output of three departments had grown by 66,141, and 223 percent, respectively. This was not a production system whose central principle was coordination.(21)
Surviving evidence suggests that, to the extent that conveyors were being used for control in this period, it was exerted only over the final assembly of completed subunits. Traces of mechanically controlled routinization reaching deep into the production process, into the production of components as at Ford, are scarce. We know from the Transactions of the American Society of Mechanical Engineers that the jigs and fixtures of American System production were well known in the supply industry of the day.(22) But this says nothing about how broadly and how intensively they were used and absolutely nothing about the extent to which the work on them was paced by central management. Ford certainly seems to have pursued the tolerances--so putting the company in a position to pursue the main point--very much farther than his competitors. Moreover, how much did the spread of jig-and fixture-work bear on the activities of firms that produced final product--that is, actually assembled cars? Many want ads from assembly firms in 1915 explicitly asked for "fitters," or, more vividly still, stated that "None but experienced men with tools need apply."(23) At no time subsequent to the publicity about Highland Park were a majority of the Ford workers doing the sort of handwork involved in making bodies and interiors; most fed or otherwise tended machines. The proportions were probably very different at firms whose competitive advantage lay more in fit and finish than in low price.
The claims about conveyors are wrong, then, because the central innovation at Ford was not the conveyor belt.(24) The ideas that made Ford production possible were progressive assembly ("moving the work to the men," in the phrase of the day) and American System (interchangeable) parts production. Progressive assembly and use of some conveyors were clearly compatible whether or not the parts required fitting. Even when many parts required a great deal of fitting, the two combined would expand production considerably.(25) Parts moved more efficiently; employees did not wander. But workers wandering the shop floor represented only the grossest form of shirking. The real source of the huge Ford output lay in pushing the limits of the American System by using large-scale investment in the requisite machines and tools to do away with most direct production tasks that required discretion. And that does not seem to have been common.
The Situation after the War
The diffusion of Ford's production idea--as against merely the diffusion of some of the machines--appears to be a distinctly post-First World War phenomenon.(26) Indeed, the most obvious outward feature of the industry's progress through the 1920s is GM's rise to Ford-scale production. It will be helpful to place that growth in its competitive context.
Table I Concentration Indexes (Output) 1919 Ford 43.7 Chevrolet Division 6.6 General Motors |Corporation~ 18.9 Top four 62.2 1925 Ford 39.4 Chevrolet Division 10.7 General Motors |Corporation~ 15.3 Top four 62.5 1929 Ford 28.5 Chevrolet Division 17.9 General Motors |Corporation~ 21.9 Top four 56.7 Source: The American Car since 1775 (New York 1971), 138-41, and Census of Manufactures, various years.
A cross-section of the industry in 1919 shows one firm Ford--selling the great bulk of the output and many relatively small firms each selling a little. Corporately, the two-firm concentration ratio was significantly higher than the one-firm (see Table 1); but the second firm, General Motors, was a true holding company. The manufacturing operations of its divisions apparently had essentially nothing to do with one another, constituting simply a collection of small firms owned and financed in common. The bestselling car after the Ford had roughly one-seventh its market share.
In 1919 the Ford Motor Company still seems to have been alone in using Ford methods root and branch. Nothing in the trade papers suggests otherwise, and two facts strongly support this conclusion. The first is that even the second best-selling brand that year had a volume less than the Ford output in 1912--that is, after Ford had begun implementing American System production seriously but before moving assembly lines enabled them to exploit it thoroughly. Second, the trade papers are full of reports of labor unrest in the industry during the late 1910s, and a close reading of the articles strongly suggests that these were wildcat strikes by highly skilled craft workers.(27)
There was, however, some investment being made with an eye to Ford principles; and Ford's senior production manager, William Knudsen, was getting restless. In 1921 he was hired as head of production on the GM corporate staff, and soon thereafter he became chief of GM's Chevrolet division. But all GM's operations, even Chevrolet's, remained small-scale relative to Ford's. Only in 1923 did Chevrolet make as many cars as Ford had in its watershed year a decade earlier (see Table 2). Indeed, a 1923 article about GM entitled "Eliminating Needless Cost and Confusion" was not about progressive assembly but about standardizing nut and bolt sizes.(28)
Table 2 Ford and Chevrolet Production Year Ford Chevrolet 1912 170,211 2,999 1913 202,667 5,987 1914 308,162 5,005 1915 501,462 13,292 1916 734,811 62,898 1917 622,351 110,839 1918 435,898 80,434 1919 820,445 123,371 1920 419,517 121,908 1921 903,814 61,717 1922 1,173,745 208,848 1923 1,817,891 415,814 1924 1,749,827 262,100 1925 1,643,295 444,671 1926 1,368,383 588,962 1927 356,188 1,749,998 1928 633,594 888,050 1929 1,507,132 950,150 Source: Jerry A. Heasley, The Production Figure Book for U.S. Cars (Osceola, Wis., 1977), 33 and 117.
That this state of affairs would not remain static was clear even in 1920, however. William C. Durant, the entrepreneur who had assembled GM but had little interest in the details of running it, became financially embarrassed in the course of Wall Street speculations. It was arranged that he be rescued by the du Pont interests, who had for some time been major shareholders. The price of the rescue was yielding control of GM to the investors. The du Ponts had recently begun contemplating radical innovations in their own company's organizational structure, including multidivisional management and centralized strategic planning staffs, and they were enthusiastic about reforming GM.(29) Their first act at GM was to install as president a man of their own, that epitome of planning and organization, Alfred P. Sloan.
By the middle of the decade the industrial structure was changing. GM was thriving and Chevrolet's output level was poised for its approach to Ford's. There were no signs as yet of another mass producer, though Walter P. Chrysler was leading a company that took over and restructured failing auto manufacturing operations, and he would be heard from shortly. The rest of the industry's output came in production runs one or even two orders of magnitude less than Ford's.(30)
More interestingly from the present perspective, the industrial substructure was changing as well. The "car for every purse" policy was announced to the world in GM's 1923 Annual Report. The annual model changes stood in strong contrast to the Ford policy of a single, basically unchanging, product.(31) But Sloan was already at work as well on a set of production innovations, whose traces could be seen in the changes Knudsen was wreaking on the Chevrolet shop floor.
Diffusion of Mass Production to General Motors and the Drive to Lower Costs at Chevrolet
Sloan's autobiography is scarcely more illuminating than car-buff books on how mass production came to GM, but, as it happens, we have an account from Knudsen's own hand. Writing in 1927, he is succinct but perfectly clear. He speaks of the preceding five years as the period in which Chevrolet adopted many of the changes in shop-floor practice that Ford had made famous, fruitful, and profitable a decade before. The beginning of his account is worth quoting in detail:
Previous |total cost standards~ had to be more or less discarded, as it was evident that volume was the only means by which the necessary earnings could be produced; and with volume established, accuracy of our workmanship and uniformity of materials would be the parents of the speed which was to produce the required cost.
The machine equipment was first tackled, with the result that all the old machines were discarded, new heavy type standard machines (not single purpose) were installed, and the fixtures strengthened so as to withstand the spring, which is the greater factor than wear. Sequence lines were established so as to pave the way for the conveyors which were to follow.... Gauges and indicators . . . were devised for all operations of major importance, and the inspection system was given full opportunity to come into its own.... And last, but not least, what was finally decided as scrap went to the scrap heap without qualifications or attempts to rework it for use in any form.(32)
This replicates the familiar Ford system, save that individual machine tools were not permanently dedicated to a single part or operation. Instead, their operations were dedicated through jigs and fixtures, which are much less expensive to replace or update. This, of course, would make model changes relatively cheaper.
It is in this context of reaching for large-volume production, lower unit costs, and, ultimately, larger sales that one should read Sloan's remarks in his autobiography about agitating in GM Executive Committee meetings and in the company more widely, starting no later than 1924, for designing more common parts into the cars of the various divisions. He advanced this enterprise in part by establishing institutions. In 1923, for example, he set up a General Technical Committee of engineers from the various divisions to discuss problems of common interest. Some of these were certainly matters of technical engineering detail, but Sloan also seems to have addressed the engineers from time to time in a hortatory mode:
The future problem of the automobile industry from the financial or business viewpoint is the great question of volume. |A company can achieve it~ ... by price reductions. ... This is one of the points in which General Motors has not made the gain ... that it has got to make from now on. |Superior profits come~ about largely through an increased spread between selling price and the cost, and through economies injected into manufacturing by lower over-head, increased turn-over, and all that sort of thing.(33)
He also tried to advance his project through concrete engineering policies. To get the Chevrolet car to be a slightly upmarket but very close substitute for the Model T and still yield or better his target rate of return on investment, Sloan needed scale economies in excess of those yielded by the current volume.(34) He formed the idea of selling major Chevrolet parts to people who did not want to buy Chevrolets. Sloan argued to his board and managers that, discreetly done, increasing the content of common parts in this way need not compromise the models' distinct identities. Discerning a gap in the GM model spectrum between the Chevrolet and the Oldsmobile, Sloan decided to make a car largely from Chevrolet parts that would sit between the two. To ensure that this plan was carried out in the proper spirit, he resisted the entreaties of various other divisions and insisted that the prototype be designed by Chevrolet engineers.(35) Thus was the Pontiac automobile born.
It is clear from Sloan's autobiography that he commissioned studies of the cost advantages this would bring, and the results were apparently summarized in a memorandum to Sloan from Donaldson Brown late in 1924.(36) The magnitude of the savings involved might indicate whether this production strategy was seen simply as a clever ploy that might feature sporadically in the company's planning or whether it was understood to be a principle that could exercise a long-term influence on the economics of automobile manufacturing. Although Brown's memorandum does not seem to have survived, one can make a crude approximation to its numbers.
Chevrolet was producing only one model. From Ford, we know something about the structure of costs involved in producing substantial numbers of a single, relatively inexpensive automobile in 1926. In March of that year, for example, the unit cost of a Model T was $421.99. Of this sum, $253.55 was allocated to materials expenses, $79.29 to labor, and $89.15 to overhead.(37) Model T output for 1926 was 1,368,383, so the total overhead expense was $122,401,859.35. This number is the key to the indicative comparison.
Suppose, just for rough magnitudes, that the underlying cost structures for producing the Chevrolet was the same as for the Ford Model T.(38) As of mid-1926, Chevrolet output appears to have been planned on a schedule of one million cars per year. Allocating the overhead costs over the smaller Chevrolet production run produces a unit cost of $455.24. The question then is what difference the common parts would have made to costs. We can infer that Pontiac production for 1926 was expected to be about 300,000. Allocating the overhead over the combined production runs yields a unit cost of $427.00, eliminating most of the initial gap.(39)
This outcome might have been different. The Ford cost structure is not the only possible one, and the dotted lines in Figure 1 illustrate two generic alternatives. In one, corresponding to higher fixed costs and lower variable costs, the change in the length of the production run in question has an even sharper effect on unit costs. In the other, lower fixed costs and higher variable costs render the effect slight. But the actual Ford numbers lead to a cost curve that is quite steep in the relevant region. Substituting the combined Chevrolet and Pontiac numbers shows that the economies of the additional scale dissipate roughly 85 percent of the cost disadvantage that GM faced relative to the Model T. In terms of the costs required to support the competitive strategy, this projection would have put Chevrolet precisely where Sloan wanted it: extremely cost-competitive with Ford on a comparable parts basis, a little better furnished, and a bit more profitable.(40)
This calculation is obviously a crude approximation. The Chevrolet, with its slightly but discernibly superior features, may have been a more expensive car to make at any volume, even holding weight constant. Furthermore, the calculation assumes all parts were common in the Chevrolet and Pontiac, which was certainly not the case. But neither of these observations seriously undermines the calculation's two main points. Given the shape of the short-run average cost curves, these production runs were long enough that only large differences in materials and labor input-which did not in fact exist--would have made a substantial difference to the unit costs. The Pontiac component of the production runs constituted a crucial element of their length; Pontiac, it appears, made the Chevrolet's initial cost-competitiveness possible.
The foregoing is suggestive but hardly conclusive evidence regarding the ongoing significance of such economies. We do have documents, however, in which well-informed participants describe indirectly but clearly the motives or interrelationships just sketched and their importance both at the time and for the longer run.
The contemporary traces are in Executive Committee discussions concerning the location of the new Pontiac production. When the project was first seriously discussed, Sloan was blunt: an Oldsmobile engine and a Chevrolet chassis, to be assembled in Chevrolet plants.(41) As he said at a later meeting, "|Pontiac~ will give us everything for which we have been waiting, namely, the lowest priced 6-cylinder car that is possible constructed with Chevrolet parts." Donaldson Brown, the senior finance executive, then stated his confidence that the plan would result in a generous profit for the corporation because of the very small capital investment involved, there being few additional tools, jigs, and fixtures required; he said that it would be a help with the overhead burden of both Oldsmobile and Chevrolet. Some months later, John L. Pratt, a persistent representative of the du Pont interests on the GM board, wanted second-sourcing for the various parts.(42) He was explicitly worried about labor agitation and its power to interrupt the stream of GM's sales revenues and profits. The committee concluded, at Knudsen's explicit urging, that the scale economies were more important. In June 1926, the Oakland Division head, desperate for production to lower his own unit overhead, pleaded for Pontiac parts productions. He was voted down, for the same reason.(43)
The retrospective view is documented as well. In a memorandum to the Executive Committee written (to judge from internal evidence) in 1934, Pratt returned to his earlier theme, complaining bitterly. When the du Ponts had bought in (1918-20) and for some time thereafter, he noted, production operations had been quite decentralized. Since then, the trend had been dangerously the other way. The policy had had real merits, he acknowledged: the scale economies had indeed yielded great profits. But with the emerging political situation, Pratt thought, operations were even more vulnerable than before to hostile labor activities. He urged the committee to think very seriously about what it should do now to minimize this problem. Much more revenue than necessary, in Pratt's view, was being supported by a small number of plants and even shops.(44) Apparently, opinions about priorities differed.
Fortune, not long thereafter, sided with Sloan. An apparently obscure passage in its famous series of articles on GM in 1938-39 forcefully makes the point about common parts and asserts that this policy was implemented on a large scale during the Depression. The passage makes the connection to the subtle competitive aspect of the annual model changes as well.
The new tools, dies, jigs, fixtures, and other equipment that GM must replace when it changes models cost for all the divisions between $25 million and $35 million annually. These items are written off during the model year .... |I~f the consumer is not to feel them in the price of his car, they must be spread over as large a production as possible. An independent like Packard or Stude-baker, for example, must sell something like 100,000 cars |for any but the least expensive cars an extraordinarily large number~ to absorb tools and dies at a practical figure per car. But a family of cars like GM's can reduce this cost even further by sharing the same tools and dies. We have already noted the fact that some chassis and engine parts are made |c. 1937~ by Buick for- Pontiac and Oldsmobile. Certain models of these three cars also use essentially the same tools for body shells, so much so that in corporation manufacturing circles they are still thought of |in their Depression-consolidated form~ as the B-O-P line. |In GM plants in Southgate, California, and Linden, New Jersey, at the time the article was written, the same assembly lines even turned out all three cars.~ Other savings come from the fact that rear-fender stampings are essentially the same on the Oldsmobile 6, Pontiac 6, and Chevrolet, and on the Oldsmobile 8, Pontiac 8, and Buick 40. To Chevrolet, with a volume that often runs over a million, the savings from such sharing are unnecessary. |In 1937, its direct competitors, Ford and Plymouth, sold 848,608 and 514,061, respectively.~ But to the smaller volume cars--especially in poor years--they are vital. Neither Pontiac nor Oldsmobile could really afford to bring out its lowest-priced 6 this year if Chevrolet were not carrying part of the body-tooling expense. And the many mutual chassis and engine parts between Chevrolet and LaSalle, together with the body parts that LaSalle shares with B-O-P, are a major reason why the Cadillac Division |whose annual output at that time was roughly 50,000~ can earn a profit at all. This trick is not peculiar to General Motors. Chrysler, in the manufacture of the Chrysler, DeSoto, and Dodge, has perhaps carried it even further.(45)
General Motors started with parts-sharing across the Chevrolet, Oldsmobile, and Pontiac divisions in the mid-1920s. The strategy spread slowly through the other divisions during the decade and received a major boost at both the extensive and the intensive margins in the redivisionalization of March 1931 that was GM's main formal response to Depression operating conditions.(46) Walter Chrysler hardwired all of this in from the start: his corporation was also organized into divisions, but Chrysler made a conscious decision to have one central engineering staff rather than independent groups in each division in order to foster the economies of parts-sharing.(47)
The independents who fell by the wayside in the Great Depression had far fewer of these economies of scope available and, on average, far smaller plants.(48) As the Fortune writer observed, a depression is a time that makes these economies count, particularly for the relatively expensive cars with relatively income-elastic demands. The idea that annual model changes were invented by Sloan in the mid-1920s can be exploded by perusing the winter Auto Show issues of many earlier years' Automotive Industries.(49) Model changes did, however, become a competitive issue in the mid-1920s, when alone in the industry, Chrysler, Dodge, and Hudson enjoyed counter-cylical success with the introduction of new models.(50) Most manufacturers made one basic model at any one time. The horizontally much more integrated GM then made annual but superficial model and style changes and, more broadly, progressive product innovation a source of competitive advantage by implementing the link to scale economies.(51) This was a much more powerful weapon than the model changes, which alone would simply drive smaller competitors to seek customers with very inelastic demands. Such a clientele might exist. The common parts strategy enabled GM to meet even that competition with lower costs. GM's rise to dominance really began with economies of scope in production.
Corroborative Evidence from GM's Labor Market Behavior
About what precisely had John Pratt been worried? In 1934, the obvious threat derived from New Deal legislation encouraging blue-collar workers to form unions. But Pratt's comments were largely retrospective. Had labor been restive--in Flint or Detroit, at any rate--in the mid-1920s or between those years and the Depression? An alternative explanation for Pratt's concern would be increased interdependence of production and so the vulnerability of profits. If shared-parts production had indeed been on the increase, that fact could well have been a source of worry to those responsible for the organization of production--the coordination of factor flows--on behalf of the shareholders. These agents could have tried to buy the peace (as cheaply as possible, of course). Relatively generous rent-sharing--when there were rents to share and where the comparison is made to appropriately situated firms--would thus be indirect evidence that production was increasingly interdependent.
It is worth examining this labor question at the grass-roots level as well as at the level of formal organizations. If the production innovations were utilized in a way that made the company's production much more interdependent, the syndicalist spirit in itself would pose an increased threat to the operation.
There is singularly little evidence that labor was stirring in the 1920s. If syndicalists or even traditional craft unionists had been particularly active in Flint, one would expect to see some traces in the local newspapers. Yet the industrial history of the city written from the Flint Journal's backfiles indicates nothing of the sort in its chapters on the 1920s. The subject is not treated at any length until the author deals with the subsequent decade, where his account of the great sit-down strike of 1937 begins with the words "Traditionally free from labor troubles...."One paragraph is worth quoting whole:
There had always been unions in Flint but never to any extent in the vehicle and automobile industries until 1933. A union in the Buick |plant~ in 1919 lasted but a short time. Over the years the carriage factories and then the automobile plants had little departmental strikes that seldom lasted more than a few hours. The first of any note was at Fisher Body Plant No. 1 from June 29 to July 8, 1930, an almost spontaneous walkout.(52)
The situation in Detroit was equally tranquil. A history of the Packard Company, written at a time when ephemera had not begun to crumble and participants were still available for interviews, records that
|t~he prosperity of the middle 1920's was shared by Detroit labor to an extent sufficient to make |the~ city "a sterile field for organized labor." ... The decline in power |of organized labor~ is indicated in the Auto Workers' News, the organ of labor in the automobile industry, which dwindled from an eight page newspaper given to reporting strikes in |1919~ to a four page paper of much smaller dimensions by 1927, and the news of |Packard~ Local 127 was typical of labor news generally, being confined largely to membership drives and reports of outings and dances held by the organization. There were no reports of large-scale unrest.(53)
The recent monographs by Ronald Edsforth and Steve Babson present a similar picture.(54)
This relative calm was in fact roughly contemporary with the Colorado coal agitation, in which the Wobblies (International Workers of the World) played their last prominent part. But there is no evidence of IWW industrial activity in Michigan in this period. Communist fly-sheets occasionally appeared at factory gates, but these groups were monitored, and no evidence has survived to suggest that they had much popular support.(55)
It is true that in the mid-1920s the American Federation of Labor (AFofL) had authorized organizing efforts in the industry, and GM management was certainly aware of this intention. A note dated 15 October 1926 from U.S. Rubber executive C. S. Ching was circulated among the membership of the Special Conference Committee (and so to a senior GM representative). Its subject is the new AFofL leadership, and it refers explicitly to the resolution about organizing automobiles.(56) But it was soon an open secret that no agents were being sent and no monies appropriated. In retrospect, the episode looks more like an effort to placate craft unions worried about membership poaching than like a response to the rise of mass production work in the industry.(57)
Thus the conventional view of a generally quiescent labor force--particularly for all of the Michigan trade--seems to be the right one at both the upper and grass-roots levels. The main element of Pratt's picture that could have been changing in this period, then, was the vulnerability of GM's stream of profits. The pattern of compensation of the relevant production employees, absolutely and relative to employees at Ford, at the time of the production and common-part innovations is thus a question of some interest. Was management increasingly eager to forestall labor unrest through higher wages, even in the absence of union agitation? There were definitely rents to distribute: GM profits were expanding rapidly even relative to its buoyant past. Unit margins were on the increase, and though part of this was undoubtedly attributable to marketing, part stemmed from cost reduction.(58)
GM employee earnings in this transition caught up with and surged ahead of Ford's. Both were superior to the rest of American manufacturing.(59) Ford was declining as the Model T grew unpopular. At the same time, Chevrolet was finding scale economies for itself and its siblings, and in due course finding a mass market of its own. By the time Ford had another model, Chevrolet was well established, and Chrysler was close to bringing out the Plymouth, the last successful interwar entrant into true mass production. When Ford brought out the Model A, there was great public excitement, and demand ran vastly ahead of supply. In such a situation, the costs to the firm (tangible and intangible) of disruptions to production were growing rapidly. Ford compensation leaped ahead--but only briefly.(60) As Ford's profitability fell relative to GM's, so (again) did the attractiveness of its compensation relative to that of the steadily advancing GM. At the end of the decade, both Fordist firms paid much better than American industry at large. By 1930, the Depression was hitting hard, but wages at the two auto producers were still good, and one did distinctly better working for GM.(61)
Table 3 Average Annual Employee Earnings ($) American Year Manufacturers Ford General Motors 1922 1,129 1,660 1,456 1923 1,242 1,742 1,515 1924 1,242 1,683 1,500 1925 1,267 1,743 1,642 1926 1,292 1,637 1,705 1927 1,292 1,593 1,724 1928 1,317 1,749 1,748 1929 1,339 1,806 1,670 1930 -- 866 1,616
Source: Column 1 is derived from the Federal Reserve Bulletin for Nov. 1929 (711 and 716) and Nov. 1930 (685). Column 2 is calculated from monthly reports on production and nonproduction hours, rates, and amounts for the Rouge Plant in AR-65-106, box 1, of the Ford Industrial Archive. The GM figures are given on p. 546 of the Federal Trade Commission, Report on the Motor Vehicle Industry (Washington, D.C., 1939). They are probably close to measuring what the Ford numbers measure, but they certainly do not get it exactly right. There is no superior alternative.
The core of this article began by describing the well-known shop-floor methods at Ford about 1914 and by clarifying the peculiarities of Ford's innovations. It went on to adduce evidence that mass production was not under way in any important sense in other American automobile companies before the First World War or immediately thereafter, casual comments in the historical literature notwithstanding.(62)
The first major postwar implementation of a Ford-style production system was at Chevrolet. It was carried out in an idiosyncratic way compared to the Ford exemplar, and the article identified the differences and the opportunities that the Chevy version opened for GM's competitive strategy. GM used common parts for various models rather than Ford's single model to achieve economies of scale, and it relied on jig and fixture changes rather than on entirely dedicated machine tools to minimize the costs of variety. The resulting opportunities stemming from the economies of common parts bore most directly on the marketplace competition among Chevrolet, Ford, and any potential entrants to the mass market. The article provided some crude calculations to support this interpretation: the shape of the short-run average cost curve highlighted the considerable economies of having a product line designed as GM's was.
Sloan left clear traces of his goals, which involved scale economies in production and changes in product design instigated by the manufacturer. The production innovation was central to successful flexible specialization. It was no accident that GM began directing a great deal of attention to its model changes at this time, but the competitive strategy was not about responding to evanescent shifts in demand at all.(63) Annual model changes were not new to the American automobile industry in the mid-1920's. GM's real competitive innovation lay in enforcing the alliance between them and scope economies. Only then did model changes, and the multi-product line along with them, become an important force in the evolving industry and firm structure.
1 John B. Rae, The American Automobile Industry (Boston, Mass., 1984); James J. Flink, The Automobile Age (Cambridge, Mass., 1889); E. D. Kennedy, The Automobile Industry: The Coming of Age of Capitalism's Favorite Child (New York, 1941); Allan Nevins and Frank Ernest Hill, Ford: Expansion and Challenge, 1915-1933 (New York, 1957); Alfred D. Chandler, Jr., ed., Giant Enterprise: Ford, General Motors, and the Automobile Industry (New York, 1964).
2 Alfred P. Sloan, Jr., My Years with General Motors (Garden City, N.Y., 1964).
3 Chandler, Giant Enterprise, 150 and 155.
4 Flink, The Automobile Age. The point here is underlined by the drift of Flink's first full paragraph on p. 242.
5 See, for example, R. P. Thomas, "Style Change and the Automobile Industry during the Roaring Twenties," in Business Enterprise and Economic Change: Essays in Honor of Harold F. Williamson, ed. Louis P. Cain and Paul J. Uselding (Kent, Ohio, 1973), and Harold Katz, The Decline of Competition in the Automobile Industry, 1920-1940 (New York, 1977).
6 For example, David A. Hounshell, From the American System to Mass Production, 1800-1932: The Development of Manufacturing Technology in the United States (Baltimore, Md., 1984), and Stephen Meyer, "The Persistence of Fordism: Workers and Technology in the American Automobile Industry 1900-1960," in On the Line: Essays in the History of Auto Work, ed. Nelson Lichtenstein and Stephen Meyer (Urbana, Ill., 1989), 73-99.
7 Most of the facts drawn on in my analysis are reasonably well known; the contribution here lies principally in what is done with them. Thus both Katz, The Decline of Competition, and James P. Womack et al., The Machine That Changed the World (New York, 1990), describe the logical connections explored here. But neither gives the reader any reason to believe that the effects were large enough to matter.
8 Alfred D. Chandler, Jr., and Stephen Salsbury, Pierre S. du Pont and the Making of the Modern Corporation (New York, 1971), 517, and "General Motors I," Fortune, Dec. 1938, 148
9 Sloan, My Years with General Motors, 66.
10 General Motors Corporation, Annual Report (1923), 6.
11 See Stephen Meyer, The Five-Dollar Day: Labor Management and Social Control in the Ford Motor Company, 1908-1921 (Albany, N.Y., 1981), chap. 2.
12 For example, Ransom Olds and Henry Leland had made early experiments with relatively unskilled assembly of high-tolerance interchangeable parts c. 1904. But the experiments were not pursued effectively.
13 Chapter 6 of Hounshell, From the American System, gives an authoritative account.
14 Ford Additional Tax Case materials in the Automotive History Collection of the Detroit Public Library and the Ford Archives, Dearborn, Mich.
15 Daniel M. G. Raff, "Wage Determination Theory and the Five-Dollar Day at Ford" (Ph.D. diss., Massachusetts Institute of Technology, 1987), chap. 13.
16 The first references to widespread rising wages in the automotive sector in Detroit in Automotive Industries, the most relevant trade paper of the day, are contemporaneous with the arrival of orders connected with the First World War. This seems unrelated to my argument.
17 For an example, with some photographs and shop-floor diagrams. see "Conveyor System Aids Big Production," The Automobile, 20 July 1916, 100-104.
18 See, for example, Jerry A. Heasley, The Production Figure Book for U.S. Cars (Osceola, Wis.), various years. Chapter 10 of the forthcoming revision of Raff, "Wage Determination Theory," gives a more detailed treatment and a quantitative analysis drawing comparisons to Ford.
19 Hounshell, From the American System, 146-51.
20 See, for example, the minutes for 2 Sept., 23 March, and 19 Jan. 1915, respectively, in the Studebaker National Museum in South Bend, Ind.
21 On the piece rates, see the minutes for 9 Feb. 1914. On the consequences, see the "Report of the Vice-President in Charge of Construction and Engineering," Secretary of the Corporation files, Supporting Matter for Minutes, Directors' Meeting of 4 Aug. 1914.
22 American Society of Mechanical Engineers Sub-Committee on Machine Shop Practice, "Developments in Machine Shop Practice during the Last Decade" 34 (1912): 858.
23 See, for example, "Newspaper Advertisements of Detroit Car and Parts Factories for More Men," The Automobile, 22 April 1915, 713.
24 Note well that none of the innovations mentioned in this sentence or the following one were engineering innovations. They were well known in other lines of manufacturing, but they had never been tried and exploited with a product as complex and as physically substantial as a car.
25 Estimates are given in Heasley, Production Figure Book for U.S. Cars.
26 It is possible that, GM aside, this diffusion went on principally through the entry and exit of firms rather than through change in the methods of ongoing firms. See Timothy F. Bresnahan and Daniel M. G. Raff, "Technological Heterogeneity, Adjustment Costs, and the Dynamics of Plant Shut-Down Behavior: The American Motor Vehicle Industry in the Time of the Great Depression," unpublished MS.
27 See "Government Analysis of Labor: Detroit Strike Reaches Acute Stage: Shortage Will Reach 35,000 Men," Automotive Industries 38 (25 April 1918): 841, "Detroit Union Workers Demand 44 Hour Week: Formal Demonstration to Claim Wage Increase: Growth of Organized Labor," ibid. 40 (1 May 1919): 972, and "Detroit Now Faces Labor Shortages," ibid. (8 May 1919): 1028.
28 D. G. Baird, "Eliminating Needless Cost and Confusion," Industrial Management 65 (1923): 334-37. The parts in question were purchased from outside GM. It is a measure of how little these initiatives were focused on competitiveness that GM published its specification lists, thereby enabling other manufacturers to share in any upstream economies.
29 See Alfred D. Chandler, Jr., Strategy and Structure: Chapters in the History of the Industrial Enterprise (Cambridge, Mass., 1962). For their frustrations with Durant's management of GM during the preceding period, see Chandler and Salsbury, Pierre S. du Pont, chap. 17. The du Ponts' ongoing inability to make Durant provide basic accounting information is particularly poignant.
30 See Heasley, Production Figure Book for U.S. Cars.
31 See Sloan, My Years with General Motors, for example, 165 and 65, respectively.
32 William S. Knudsen, "For Economical Transportation: How the Chevrolet Motor Company Applies Its Own Slogan to Production," Industrial Management 76 (1927): 65-68.
33 General Technical Committee minutes for the meeting of 12 Aug. 1925, in the General Motors Institute (GMI) Alumni Foundation's Kettering Papers, Flint, Mich.
34 See "1926--Pontiac--How It Got Started" in the Daniel Wilkerson Papers at GMI: "We were, of course, fighting to hold place in this market against the powerful price sales appeal of the Model-T Ford." Sloan writes in his autobiography (154) that "|d~espite the success of the K Model Chevrolet, it was still too far from the Ford Model T in price for the gravitational pull we hoped to exert in Mr. Ford's area of the market. It was our intention . . . to move down in price on the Model T as our position justified it."
35 For traces of the entreaties, see Sloan, My Years with General Motors, 156. Regarding the outcome, Wilkerson writes, on the page following that quoted earlier: "In the design stage, our organization group responsible for the primary development of the budding Pontiac was the Chevrolet engineering staff. This department produced the drawings, production plans, preliminary manufacturing cost figures, and pilot models.... The pilot model Pontiacs delivered by Chevrolet ... made use of a considerable number of standard Chevrolet parts for economy, and also for facilitating future service supply."
36 Sloan quotes from his memorandum of reactions to it in My Years with General Motors, 157.
37 The cost figures for the month are roughly typical for the year. The figures derive from the Model T cost books in the Ford Archives, accession 736. They appear more accessibly, with a sketchy time series of earlier figures, in Wayne Lewchuk, American Technology and the British Vehicle Industry (New York, 1987), 45.
38 The calculation that follows assumes that the two companies could operate with a roughly similar degree of backward integration, It is customary to think of Ford as a much more integrated company than GM in this period. But Nevins and Hill are openly skeptical about this; see Nevins and Hill, Ford: Expansion and Challenge, 9. Academic commentary early in the decade identified engines, gears, axels, crankshafts, radiators, electrical equipment, roller bearings, warning signals, spark plugs, bodies, plate glass, and body hardware as components produced within the corporation; see C. C. Edmonds, "Tendencies in the Automobile Industry," American Economic Review 13 (Sept. 1923): 426. The policy of top management was to have an assured supply of all items essential in assembling an automobile; see Chandler and Salsbury, Pierre S. du Pont, 514. Edmonds also observes (433) that Ford purchased "large numbers of bodies" and "|n~umerous parts" from suppliers. Nor did the Ford and Chevrolet product designs differ in ways that would have required notably different skill structures in the labor forces.
39 "Motor Makers War Is Believed Near," New York Times, 7 June 1926, 34. Since the expectations about Pontiac were stated in terms of expected daily output from a new purpose-built factory, the calculation turns on an, assumption about how many days a year the factory would have been expected to operate. Chevrolet factories seem to have operated about three hundred days per year in 1926. See "Ford Closing Stirs a Crop of Rumors," New York Times, 3 Dec. 1926, 8. Depreciation ("wear-and-tear") is presumably sensitive to utilization. No handbook survives giving definitions and conventions for Ford cost accounting in this period. James O. McKinsey, Budgetary Control (New York, 1922), was the standard text of the day. It makes clear (189) that estimated depreciation of capital equipment should be incorporated in the overhead figures given in cost accounts. The total in the text has therefore not been further adjusted.
40 A representative Chevrolet for these purposes weighed 2,130 pounds complete, the analogous Ford, 1,961. (Much of the Chevrolet's superior furnishing is said, in the folklore of the trade, to show up in this statistic.) Assuming materials costs to be proportional to weight, this suggests a Chevrolet unit cost of $449.82. A very crude calculation of gross profit margins then puts Ford's at roughly 19 percent, the Chevrolet Division's at roughly 30. Superior finish and materials would cut into this difference, of course, though not by much; the same is true of a marginally more complex engine.
The weight and price data for this calculation come from the Automotive Industries annual statistical issue of 18 Feb. 1926. There is some ambiguity to be resolved in matching those numbers with the cost statistics. But the closeness of this estimated Ford markup to the Ford Company's own figure for 1924 (see, for example, Lewchuk, American Technology, 42, noting that 1924 is the last year reported) suggests that this match is the correct one.
41 Executive Committee meeting of 3 Sept. 1924. This reference and some others that follow come from notes that Alfred Chandler made while researching his 1962 book and has kindly made available to me.
42 Ibid., Executive Committee meeting, 23 June 1925.
43 Ibid., meeting of 30 June 1926.
44 Report to the Executive Committee from J. L. Pratt, 25 April |1934~.
45 "General Motors I," 156.
46 For a summary of the minutes of the meeting, see Sloan, My Years with General Motors, 177. Both cost-accountable management and engineering staffs seem to have been merged.
47 Personal communication to author from Bruce Thomas, Chrysler Corporation Engineering, 1991.
48 Timothy F. Bresnahan and Daniel M. G. Raff, "Intra-Industry Heterogeneity and the Great Depression: The American Motor Vehicle Industry, 1929-1935," Journal of Economic History 51 (1991): 317-31.
49 Or see R. P. Thomas, "Style Change and the Automobile Industry."
50 R. P. Thomas, An Analysis of the Pattern of Growth in the Automobile Industry, 1895-1929 (New York, 1977).
51 GM depreciated the jigs and fixtures that temporarily dedicated its machine tools over a three-year working life. Ford appears to have depreciated them over four. This variance does not make for a significant difference in overhead burden per unit output. The competitive significance of annual model changes ran through the basically static economies of common parts.
This production strategy did turn out to have a dynamic aspect, however. The long production runs supported more routinized jobs, which had lower training costs. Companies with lower employee training costs are more inclined to respond to cyclically slack demand, all other things equal, by scaling back rather than retaining their labor force until closure is the only option. This is in essence what happened to the population of firms in this industry in the Depression. See Bresnahan and Raff, "Technological Heterogeneity."
52 Frank Rodolf, An Industrial History of Flint (Flint, Mich., 1949), 366, 367.
53 John Parker, "A History of the Packard Motor Car Company from 1899 to 1929" (MA Thesis, Wayne State University, 1949), 81.
54 Ronald Edsforth, Class Conflict and Cultural Consensus: The Making of a Mass Consumer Society in Flint, Michigan (New Brunswick, N.J., 1987), esp. 114-26, and Steve Babson, Building the Union: Skilled Workers and Anglo-Gaelic Immigrants in the Rise of the UAW (New Brunswick, N.J., 1991).
55 Roger Keeran, The Communist Party and the Auto Workers Union (New York, 1980), chap. 2.
56 Accession 1662, box 8, Hagley Museum and Library, Wilmington, Del.
|Printer friendly Cite/link Email Feedback|
|Author:||Raff, Daniel M.G.|
|Publication:||Business History Review|
|Article Type:||Industry Overview|
|Date:||Dec 22, 1991|
|Previous Article:||Labor and Imperial Democracy in Prewar Japan.|
|Next Article:||The modernization of production in the French automobile industry between the wars: a photographic essay.|