Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Dynamic Competition and Productivity Advances BURTON H. KLEIN It seems no accident that the successful industries in We 1970s were either those embodying new technology or those in which plants tended to be relatively small and unspecialized. Moving toward more flexible and less specialized Arms will be the principal long-term adjustment required of U.S. industry if it is to thrive in an era of international competition. During the 1970s a large number of U.S. industries performed splendidly when dealing win a greater degree of foreign competition than they had faced before. Their performance was attributable to an impressive gain in the rate of productivity advance, which made possible Me selling of com- parable products at lower cost or better products at more or less the same cost. It seems to be no accident Nat the successful industries were either those embodying new technology, such as semiconductors and computers, in which flexibility was all-important, or those in which plants tended to be relatively small and unspecialized, such as manufacturers of athletic equip- ment or scientific instruments. In the first part of this chapter, I discuss the essential features of a dynamic model, particularly the relevance of the mode! for explaining how productivity gains come about. Next, I present some statistical results of applying the model to groups of U.S. industries. Finally, I argue that moving toward more flexible and less specialized firms is the principal long-term adjustment required if U.S. industry is to thrive in an era of international competition. 77
78 BURTON H. KLEIN MOTIVATED COMPETITION Dynamic competition consists of developing better or less expensive al- ternatives, and it requires taking risks to ascertain how particular hypotheses will work out. Not all risk taking will be successful. But if one or another firm is successful, say, every 2 or 3 years, that industry can be described as being '~predictably unpredictable." And inasmuch as firms in the industry cannot buy insurance from Lloyds against competitive intrusions, they are motivated to take technological risks to avoid competitive nsks. Such com- petition is measured in teas of the degree of technological risk taken by fibs. If, as in a classical economic world, firms took no nsks, the menu of alternatives would remain unchanged. If, say, In the mousetrap industry fins were motivated to take risks, such as might be involved in obtaining slightly sponger spnn=,s, the menu of alternatives would change, but ever so slowly. On the other hand, if firms in such an industry were more or less simultaneously taking bolder technological risks and imposing larger com- petitive risks on their nvals, then the menu obviously would change more rapidly. The main purpose of the dynamic model proposed here is to relate the degree of competitive risk taking to the rate of technological progress. Continuing productivity gains presuppose advances in relevant technolo- gies and a keen desire- to make good use of this progress. Indeed, without continuing technological advances, productivity gains would sooner or later die out. Luck is involved in making a technological advance, because the larger the attempted improvement, the more diverse are the hints required to bring it about. Luck is also involved in bringing about productivity gains because, generally speaking, the technological advances (e. g., robots, syn- thetic fibers) on which productivity gains are premised, whether in the form of reduced labor costs or reduced input costs, come about in industries other than those in which the gains occur. However, productivity performance seldom depends on one critical technological ingredient; ordinarily it depends on several. [Ience, just as firms search for ideas with which to bring about technological advances, so must they search for ways to bring about pro- ductivi~ gains. Contrary to neoclassical models of productivity growth, in which Finns make instantaneous adjustments to take into account changes in factor prices, there is not, so to speak, a substitute hiding in every bush. Also, contrary to such models, a factor of necessity must be involved. As will be seen, it is simply impossible to make reasonably good predictions without taking the factor of necessity into account. Dynamic economics relates both necessity and luck to the rate of progress. Three critical assumptions were made in developing He dynamic model. First, differences in "the propensity to engage in risk taking" within an industry are assumed to be smaller Man those between industries. I define "the average propensity to engage in risk taking" (PERK) as the degree of
DYNAMIC COMPETITION AND PRODUCTIVm Al)VANCES 79 risk a firm takes to achieve minor or major gains in productivity. A firm desiring to generate less costly or better-quality alternatives may introduce robots into its production process gradually; or, like Japanese auto Grins, it may attempt to develop its own highly specialized robots for the purpose of promoting reliability by substituting robots for workers in the most monot- onous tasks. Or, the Of may introduce several new technologies simulta- neously, which would imply an even greater degree of risk. Nevertheless, firms are not free to choose which of these nsk-takin~ strategies to adopt. A firm can jeopardize its competitive position by taking either greater or smaller risks than its rivals. Hence, the first assumptionassumption because there is no way to establish its reasonableness definitely is that firms are in a NASH dynamic equilibrium an equilibrium wherein if one finn does not change its strategic policies with respect to risk taking other firms will have no incentive to change theirs. The second critical assumption is that a greater degree of risk taking will lead to a wider distribution of outcomes as measured by the degree of success fibs have in bringing about the cost reductions that, in turn, generate pro- ductivity gains. A finn that takes only little risk when incorporating a new technology into its production process obviously will be able to obtain only a small reduction in cost. But the larger the attempted cost reduction, the greater will be the variety of technological sources the firm must draw upon and the more the cost reductions will depend on luck. Indeed, if fimns' costs were identical, this would indicate that progress had stopped. Even if all productivity gains were embodied in new capital equipment, costs would not be the same for all finns in an industry, because not all would buy new capital equipment simultaneously. Typically, the timing of investment de- cisions depends on luck in selling new models and on luck in finding machines that will provide significant advantages and fit into an ongoing production process. Moreover, productivity gains, broadly defined, also depend on the outcome of a search to find less expensive inputs. This is not to say, however, that cost differences are necessarily reflected in price differences: if demand is relatively inelastic if increases in prices do not result in a commensurate reduction in salesthere will be little incentive to hold costs in check, and little incentive, therefore, to take risks to generate productivity gains. On the other hand, if demand turns out to be elastic if large output penalties are associated with increases in costs and if finns cannot find ways to reduce their costs, they risk going out of business. The third assumption is Hat progress can be measured in terms of the advantage provided to buyers: namely, the advantage in being able to choose from a wider distribution. As was just indicated, when the elasticity of demand is fairly high, the ability of firms to raise prices is highly constrained. And high-demand elasticities also imply buyers who are bow knowledgeable and motivated.
80 - ci ~ \ 91~H ~lAllOnCOUd \ I ~ cD G I tL 31~8 ~lAll~tl0OUd I , _ In G O H _ ,,, ~ o Id \ \ \ CO \ 1 31~H ~L'l\llonaoua en llJ O ~ _J C ~ C,0 _ O G ~ Q C' LL ~ O ._ :e - ._ - a; 5 c o c) - :t Cat bled N1 PONCHO ~ ~ , 1~H All~llonaoud · ·- · e _ _ bled N1 PONCHO ._ C) 3 en CD ~ fir cat Cat - O :,_ _! ~ IS CO I,L O cr: ~ I C' ~ ~0
DYNAMIC coMpETrrIoN AND pRoDucTrkrlY ADvANCES 81 A highly oversimplified diagrammatic version of the dynamic model is shown in Figure 1. Note, as shown in the top panels, the greater Be degree of risk taking, the more steeply sloping will be the productivity curve. This follows from the fact that the slope of the curve is directly proportional to the width of the distribution. Also note that when PERK is zero, there is no distribution. When PERK is positive, in some industries engaged in making standard- ized products, major emphasis is placed on reducing costs. In others, emphasis is on providing more effectiveness per dollar of expenditure, semi- conductors, for example. In still others, such as television and, especially, computers, bow cost and quality are emphasized. But assuming that the name of Be game in those industries is to generate competitive products, they are, in effect, demanding productivity gains from each other. If firms fall behind in such a dynamic game, they will be "taxed" by their competitors. Hence, when firms are in a dynamic equilibrium (a state in which behavior remains predictably unpredictable), it can be predicted that the rate of productivity gain will depend on the value of PERK. If, as in the bottom panels of Figure 1, PERK declines, then the produc- hvity curve will slope off; that is, if we examine the slope of the curve for a variety of industries in which dynamic competition has declined, we can expect Mat the average rate of productivity gain will dampenbut not in every industry, because luck is also involved. Conversely, when Be value of PERK increases, we can expect the rate of productivity gain to accelerate but not in all industnes, because, again, the factor of luck is involved. This model was tested against data for all U.S. ~ndustnes in four-digit SIC manufactunng categories. The data cover He period 1960 to 1980. How He factor of necessity was measured in making the tests is described below, and the results are presented in terms of weighted averages. The statistical test that I employed was Wilcoxson's Signed Rank Test, which takes into account all observations. One group of 11 industries failed to meet the test, but all the others came within the acceptance region, measured at the 5 percent confidence interval. ~ STATISTICAL RESULTS All of the industries analyzed (Table 1) were separately classified in teens of their price performance during the periods 1960 to 1970 and 1970 to 1980. The industries were then combined into three groups based on Heir rank distribution. The best group of industries in each of the periods were Hose whose prices increased at ~ rate that was one standard deviation or more below the average, and the worst group were those whose prices increased at a rate that was one standard deviation or more above He average. The industries whose ranking in the distribution remained the same during both
82 ~ e ,,` .= ~ O ~ ~ 5\ C~ o ._ C) - o ~:: S: ._ C) C) ~: CC C) ~ 00 O ~ ~ C5 C~ {_ ~ 0 C) C~ ~ C5 _ _ J ~ ~ C) E~ C~ O .- COO O 3 o: ox ~2 O~ o e~ _ =~. o {_ ~ 4,,) ;> e,5 _ s: ~5 _ o x - _ o es ~ o _ ~ _ ' C $ _ ~ ~ . . . X t_ ~ oo ~ ~ ~ o o o _ C ~ . . . ~ ~ o _ _ _ o C5` =o o ~ ~ °. .> o ~. ° =- o =, <-, $ . . . o _ {_, _ _ _ $ $ $ u, cq O' .~{ o ~ ~ - - o o - x o~ ~ . ~ - D X C) t_ ~ Cr~ C~ 11 ~ \~D 00- 0 t~ S~ · ~ C: X C~ X . . . _ _ ~ C~ O o ~ _ . . V~ U) ~0 11 ~ ._ - _ ~ e`0 ·= · tr) 3 ~ U, o Cr~ CO o o ox - o o C~ C~ li _ _ _ _ c,o . _ o ~ . . o o ~D o o o ~ - ._ o C~ o C~ U: o o o ._ o C~ oo C~ - ~C - :e C,: ._ - ~: ~ 1 .= o :C - . o C~ _ CC U: - - c CC _ - :, .o _ _ Ct ._ _ ~ C~ _ o o V ~ ~ D . . ;3 o ~q
DYNAMIC COMPET1710N AND PRODUCT7Vm ADVAI`ICES 83 periods are Nose shown in Group lI; that is, their relative prices remained He same during bow penods. On the other hand, prices increased the least in the industries shown in Group I, and, if the calculations had been made in terms of constant prices, the prices for those industries would have actually declined. Conversely, industries shown in Group III are those that exercised less price restraint during the 1970s than they did during He 1960s. Now, think of competition simply in terTns of the elasticity of demand, as nearly all down-to-earth economists do. Though the elasticity of demand relies in part on consumer preferences, it depends mainly on the availability of substitutes. The greater the availability of substitutes, the less freedom of choice firms have with respect to raising prices; the smaller the availability, the more likely it is that prices will rise. As it happens, this concept of competition, in which the availability of substitutes is taken as a given, is a special and limiting, case of dynamic competition, because in the short non the availability of substitutes must be taken as a given (i.e., when the value of PERK is, by definition, 01. However, as already shown, in the longer run the rate at which new substitutes are developed depends on the factors of necessity and luck. The term "unitary elasticity of demand" means that, say, a 10 percent increase in prices will be followed by an equal reduction in output. When this proposition is expressed in terms of rates, it can be seen that the industries in Group II exhibited greater demand inelasticity than Hose in Group I, and the Group Ill industries exhibited a far greater degree of demand inelasticity Can those in Group lI. Thus, while in the last group there was a 40 percent penalty in terms of output performance, this was far outweighed by prices increasing 27 times as rapidly during the 1970s as Hey did during the 1960s. Why demand turned out to be so inelastic for the third group of industries is easy to explain: in these indusmes, substitutes were far from plentiful. The main industries included basic materials of one kind or anothersteel, aluminum, copper, cement, and a variety of metal products were the main ones. For the most part, these industries have not featured dynamic com- petition for some years. Moreover, the increasing degree of protection of the steel industry from foreign imports beginning in the late 1960s protected not only that industry's ability to raise prices, but also Hat of industries making substitutes. During He 1 970s, increases in the costs of steel inputs alone had quite as large an impact on manufacturing prices as did increases in the prices of oil inputs.2 Indeed, because steel inputs are used in many industries, the impact generated a source of deadweight drag by making American industries less competitive in foreign as well as in domestic markets. Moreover, because the price shocks created by those industries occurred mainly during economic upturns, He generation of deadweight drag not only jeopardized economic growth but made necessary the checking of inflation by very substantial . , . Increases In Interest rates.
~4 BURTON H. KLEIN By contrast, Me first group of industries exercised a high degree of price restraint not because of their benevolence but, rawer, because substitutes were becoming more plentiful. ~ the 1970s, American Grins faced a large in- crease in foreign competition: well over one-half of Me industries in Group I were heavily involved in foreign competition. In other industnes, there was an increase in domestic competition. For example, the TV news media be- came a better and better substitute for newspapers. In the farm machinery industry, there was a substantial increase in dynamic competition due to the efforts of John Deere & Company to become something other than a tradi- tional farm equipment manufacturer. (International Harvester, which re- mained highly inadaptive, was He main victim of the competition.) Primarily because of the increase in domestic competition, the farm implement industry was able to export almost one-half of its output during the 1970s. Finally, also contributing to higher demand elasticities were some industries in which there were autonomous declines in demand, manufactured milk products and brassieres, for example. As Table 1 shows, the average movements in wage rates, labor produc- tivity, input costs, and equipment expenditures were more or less as might be predicted, given the decidedly different demand elasticity pressures to which the industries were subjected. Inasmuch as Group I industries suffered a smaller decline in equipment expenditures than either Groups II or III, while experiencing a sharp increase in He average rate of productivity gain and, as a consequence, a relatively small loss in He average output rate, it can be assumed Hat they were economizing on capital as well as labor inputs. In his diary, Sadi Carnot, who is now given credit for having discovered the Second Law of Thermodynamics, proposed a tax on the rent of French fanns, the purpose of which was to encourage large landholders to sell parcels to individual farmers, who would have an incentive to improve productivity in order, as Carnot wrote, "to excape the tax."3 Apparently, as of that time productivity on French farms was increasing so slowly that it was almost in a static equilibrium. Note in Table 1 that Group I industries were improving productivity at such a sufficiently rapid rate that they were able to minimize a large positive "tax" imposed on them by Heir competitors. But, as was also seen, when industries are able to base their expectations on an inelastic demand curve, there is no need to bring about productivity advances to minimize the tax, because it is a negative "tax." Hence, the rate of pro- ductivity advance depends on the "tax" difference and it does because a positive tax encourages a search for a wider diversity of technological inputs. As it happens? this is close to the reasoning contained in He Second Law of Thermodynamics, which is also concerned with explaining movements away from a static equilibrium (i.e., the entropy of the physical world is always . · ~ Increasing'. The same reasoning also provides a good argument for foreign competition.
DYNAMIC COMPEII7lON AND PROD UC;TIVIIY ADVANCES 85 As was pointed out above, the increasing presence of foreign competition played an important role in increasing the rate of productivity gain, with the consequence that output losses were much smaller than they otherwise would be. And, despite what some labor union leaders and ill-advised politicians have had to say, it was not foreign competition that was robbing Americans of jobs. In the Group I industries, which were the most "taxed" by foreign competition, output during the 1970s declined by about 10 percent compared with the 1960s, but in Group III industries the decline was about 40 percent. NEED FOR LONGER-RUN ADJUSTMENTS In the short run, large public deficits may result in serious economic problems because of their well-known effect on keeping interest rates high. But just imagine how much more serious the problems would be in the maintenance of macros/ability if the phenomenon of deadweight drag had continued! True, with voluntary quotas some increases in the prices of steel inputs may occur if the current recovery gains further momentum. These increases, however, will be small compared with those involved with the previous method of control (i.e., the target-price mechanism). Nevertheless, in the longer run some industries face serious adjustment problems. For many years some industries featuring scale economies that were associated with an almost unbelievable degree of specializationno- tably, automobiles and steel were able to escape from foreign competition. But that is no longer the case. Why, then, have Japanese firms done so well? In particular, why over the past 20 years have productivity gains in the Japanese auto industry averaged about three times those of the U.S. auto industry? Essentially, they have evolved a practice that can be described as "dynamic flexibility.'' As con- trasted with static flexibility, dynamic flexibility is not concerned with pro- ducing more than one product (e.g., cars and light trucks) on a single production line although the Japanese do this, too. Rather, it is concerned with de- signing production lines in a way that they can quickly evolve in response to changes in either the product or production technology. In other words, Me central preoccupation is to get ideas into action quickly. For example, when new Japanese models develop bugs, some companies make changes on the production line before the next model year. The main purpose of dynamic flexibility, however, is to make rapid changes in production technology for the purpose of lowering costs and thereby improving productivity. Almost every Japanese auto company has a large machine-tool operation in which 200 to 400 people do nothing but create new tools, which are quickly introduced into the production process. In turn, this requires a highly nonspecialized method of operation, which the Japanese often liken to the approach followed in American farming. In particular,
86 BURTON H. Kr FIN workers are rotated from job to job on the production lines every 3 to 6 months, and managers, every 3 to 5 years, often to decidedly different jobs. For example, in one fib I visited, He head of the accounting department was an engineer. Or, when I talked to a Japanese executive in charge of building a new Honda plant to produce Accords in Columbus, Ohio, I found Hat he was not an engineerand never before had been concerned with building a plant. "Why did they choose you?" I asked him. He replied that his superiors wanted someone who was good at asking shard questions and in the Amencan marketing division of Honda, from which he came, "We Japanese specialize in asking sharp questions'" Another respect in which Japanese auto plants are like Amencan farms is Hat there is a high degree of interaction between management and workers. For example, I asked an Amencan worker in charge of the painting line at He Honda motorcycle plant in Columbus, "What is it like to work for a Japanese firm?" He told me in some detail about an experience he had with receiving painting robots from Japan, which, having been designed for paint- ing flat surfaces on automobiles, did a poor job in painting the curved surfaces of motorcycles. He discussed the problem with two managers during lunch in He company cafeteria. Immediately after lunch, the managers went to see o c: UJ Path 1: Minimum average COSt at various points in time Path 11: Minimum average COSt over time \~ - An' YEARS FIGURE 2 TWO ways to reduce costs.
DYNAMIC COMPETITION AND PRODUCTIVITY ADVANCES 87 for themselves that He robots left bubbles In the paint, and by 4:00 p.m. that day a message was sent to Japan that they were going to revert to the former method of painting until a better-designed painting robot was furnished them. This American worker's reaction? Nothing like that ever before had happened to him! A diagrammatic representation of the classical American and Japanese approaches is shown in Figure 2. The Japanese long-run cost paw is described as a "saturation path," because the Japanese claim that they are introducing innovations about as rapidly as they can. Inasmuch as costs cannot be pinned down for a period longer than 6 months or so, there is a greater multiplicity of long-run cost curves. It also should be noted that Japanese plants for producing cars are only about one-third as large as comparable American plants. Why? The Japanese claim that building larger plants would impede their enjoyment of dynamic flexibility. Similar practices to those of the Japanese firms are probably not numerous in American industry, but there are some notable exceptions. John Deere & Company operates in much He same way, as does Boeing. As a result of subcontracting with Mitsubishi for parts to be used on the 757 and 767 airplanes, Boeing executives claim that Mitsubishi is a good parmer, because they operate in much the same way as Boeing. Moreover, engineers from the Sorry television plant located in San Diego claim that they not only make about 90 percent of Heir components, but by utilizing a highly skilled labor force Hey are able to operate quite as flexibly as comparable plants in Japan. And engineers from the Honda motorcycle plant in Ohio also make the same claim. Desperately needed, of course, are the data required to substantiate these claims. CONCLUSIONS Ending the U. S . quotas on Japanese auto imports was a constructive step. lithe Japanese did not evolve into Heir present method of operations overnight; hence, it would be very optimistic to assume that American auto firms will be able to make the adjustment without some serious setbacks. Is moving toward less specialization contrary to the teachings of Adam Smith? By no means. He understood He secret of the success of the Japanese auto ~ndus~y as early as 1776: The man whose whole life is spent in performing a few simple operations, of which the effects are perhaps always the same, or very nearly the same, has no occasion to exert his understanding or to exercise his invention in finding out expedients for removing difficulties which never occur. He naturally loses, therefore, the habit of such exertion, and generally becomes as stupid and ignorant as it is possible for a human creature to become.4
88 BURTON H. K] FIN NOTES 1 For a more detailed discussion of die statistical results, see Burton H. Klein, Prices, Wages, and Business Cycles (New York: Pergamon Press, 1984), pp. 4~127. 2. Ibid., pp. 123-124. 3. N. L. S. Carnot, Reflections on the Motive Power of Heat. Translated by R. H. Thurston (New York: Wiley, 1980), pp. 206 208. 4. Adam Smith, The Wealth of Nations. London: Everyman's Library, J. M. Dent & Sons Ltd., 1947: vol. II, p. 278.
