The Free-Market Innovation Machine: Analyzing the Growth Miracle of Capitalism

The discussion uses a model of an oligopolistic industry, in which, as already suggested, most routinized innovation takes place. The model here takes no account of risk and uncertainty. Of course, the R&D devoted to any particular project—to the creation of any special new product or process—involves great uncertainties about cost, the time that will be required, or even whether anything useful at all will emerge. But here I will assume that the R&D activity of any firm with which we are concerned is on a scale sufficiently large to include a considerable number of such projects. Then, to simplify the discussion, one can assume that something like a law of large numbers applies, and that the overall value of the innovations that will emerge from the firm’s R&D is moderately predictable, even if that of any individual R&D undertaking is not.¹

If it is true that in a number of high-tech industries innovation is the prime weapon of competition, a simple story emerges. The profit-maximizing firm will adopt the quantity of R&D expenditure at which expected marginal profit yield is zero. But this magnitude will depend on the behavior of the other firms in the industry. A firm that falls behind in its innovations for a substantial period of time can expect considerable erosion of its market either because its product is considered inferior by customers or because its costs, and therefore its prices, are higher than those offered by rivals for products of comparable quality. Consequently, no firm will in the long run dare to underspend its competitors systematically.⁴ Thus, one can expect an industry norm to emerge, with firms that engage in substantial R&D outlays generally making sure that their R&D expenditures are up to that norm.5 That norm will constitute an equilibrium in the arms race, but one that is only temporary—a truce, not a full end to hostilities.⁶

Still, firms will be tempted to exceed the equilibrium norm, and such enhanced spending would very likely be profitable if rivals could be counted upon to ignore the challenge and keep to their previous levels of expenditure. For that would enable our spoiler firm—the firm that upsets the tacit arrangement under which no player previously violated the industry spending norm—to outperform its rivals in terms of price and/or product quality and take customers away from them. However, each firm is aware that such passive reaction cannot be expected from competitors. Rather, the rivals are likely to respond by matching any spoiler’s increased rate of investment. The likely consequence of the spoiler’s initiative, then, is that the industry will end up with a new and higher R&D investment norm, but with none of the firms having gained any (relative) advantage over its competitors.

Total profits cannot be expected to be the same as they were before, once the spoiler firm has caused a general increase in R&D investment in the industry. Very plausibly, as in the case of rivalrous inflation of an industry’s advertising, the industry’s firms, including the spoiler, may well end up with less profit than before. This does not mean that increases in R&D expenditure will never occur. Rather, it implies that they will take place when the R&D effort of some firm has provided a highly promising breakthrough, calling for an increase in investment outlay whose profit prospects are unusually bright.

The story is described graphically in figure 4.1, which shows a marginal cost (MC) curve and two marginal revenue (MR) curves, MRJ and HMR, all as functions of the amount of R&D spending by Company X. For simplicity, the curves are drawn as straight lines. The MR curves reflect the two possible competitor responses. If each time Company X expands its spending its rivals do the same, the MR resulting from X’s enhanced R&D will be quite low (line HMR), because Company X will fail to pull ahead of its rivals. On the other hand, if competitors let X increase its R&D spending without increasing their own, then X can expect its increased product quality to put it well ahead of the others, and so its MR from R&D spending will be relatively high (line MRJ).

1. The cumulative character of many innovations. Many innovations do not merely replace older technology and make that technology obsolete; rather, they add to what was previously available, thus constituting a net increase in the economy’s store of technical knowledge. Such innovations can be said to entail creative knowledge accumulation rather than creative destruction.

2. The well-known public-good property of information in general, and of innovation in particular. Improved technology, once created, need not contribute to the output only of the firm that made the breakthrough. At relatively little additional cost, it can also add to the outputs of other enterprises. Thus, although this public-good property can impede attainment of the optimal amount of innovation expenditure, it also has a beneficial side, constituting an economy of scope in the generation of technological improvements for a multiplicity of firms.

3. An” accelerator” feature of innovation. A level stream of innovation usually means that output will be not level but growing. An economy whose R&D produces a level output of one innovation per month, with a given productivity contribution, will obtain a GDP that is higher each month than it was in the previous month. The economy’s ability to produce output will grow constantly even though the flow of innovations that fuels that output growth remains level at one invention per month.11 This acceleration relationship applies to innovation generally, so that, if the competitive market mechanism described in the previous graphs leads firms to devote a constant level of resources to R&D, we would expect continued growth of GDP to result.

Of course, as noted, the ratchet principle tends to increase the expenditure of resources on innovation, not just to leave them level. Our acceleration principle tells us about the effects of this, too, if there are no diminishing returns to R&D innovation productivity (but note the Jones argument described in footnote 9 above). The acceleration relationship indicates that, if the level of R&D spending were to increase just once, for example, and then stay at that new higher level forever, the growth rate of GDP would also move upward. GDP would then grow at a faster rate forever.

All of this reinforces the role of the competitive market mechanism and its stimulation of innovation as a contributor to the extraordinary growth that characterizes the world’s free-enterprise economies. But at this point one may well pause to ask whether the analysis does not exaggerate the facts. It tells us that R&D investment norms can be expected to increase from time to time, and that the ratchet serves to resist retreat to the previous lower investment levels. This then implies that the GDP growth rate can be expected to keep rising, and the natural question is whether history is consistent with such a picture. The answer is, in fact, yes, but before we show this, let us be clear that the analysis never asserts that this must continue forever. For example, R&D personnel may find it increasingly difficult to come up with good new ideas, so that, even if more and more is spent on R&D, it is conceivable that, some day, its results will be worth less and less.

But what about the facts? Of course there have been decades during which catastrophes such as wars and depressions have kept GDP from rising or have even reduced it. But ten-year periods are too short for the evaluation of the long-run performance of market economies. We can say that there have now been about six half-centuries since the modern free-market economy made its debut. The first half of the eighteenth century is generally taken to have preceded the birth of the British Industrial Revolution, with GDP per capita consequently rising more slowly than in the second half. And, during the century as a whole, GDP per capita in Western Europe surely rose a substantial amount, but one that was less than 50 percent.12 In the nineteenth century we observe a similar pattern, with relatively modest growth even in England until after 1830, and in France, Germany, and the United States until well after that. Thus, growth in the second half of the century was presumably higher than in the first, and in the century as a whole per capita GDP in Western Europe did something between doubling and trebling—far better than the previous century. The twentieth century’s growth, too, sped up during its second half and, remarkably, per capita GDP ended up seven or eight times as large as it began.¹³

This chapter has provided a concrete model that can help us to understand the role of competition among high-tech oligopoly firms in the remarkable growth record of the free-enterprise economies. It is an arms-race model with the addition of the ratchet attribute of innovation investment. The first of these features—the arms-race analogue—can explain the presence of powerful norms for R&D investment in leading oligopoly industries. The ratchet, in turn, implies a propensity of such an industry not to retreat from the current standard for R&D spending, and occasionally to move to a new and higher standard that all major firms in the industry are apt to feel they must match.

1. The applicability of the law of large numbers here must not be exaggerated. The law would apply fully only if the chance of success of one R&D project does not depend on the success of others. Where they are interdependent, failure in one may sink many others, making innovations something nearer to a “house of cards” activity.

2. However, one should note that only monopolies, notably the old AT&T with its Bell Laboratories, seem to be in a position to carry out substantial basic research, particularly if they are regulated and regulation permits them to recoup their research outlays. Another part of the explanation lies in the heavier spillovers that basic research generates, discouraging such activity by competitive firms, which fear that too much of the benefit will go to competitors. I also recognize that innovations are often contributed by small firms (see the next chapter) but, except when it is carried out by trade associations, it is rarely routinized. Here, Schumpeter’s views on the role of large firms in innovation should also be recognized.

3. And oligopolistic industries are important for the economy not only because of their role in innovation. In addition, they probably constitute the largest share of U.S. output, and the revenues of the largest oligopoly firms outstrip the gross domestic products of entire economies such as Denmark and Norway.

4. Witnesses in patent litigation lawsuits or antitrust cases involving competing technology repeatedly testify (and internal firm documents confirm) that the firms try avidly to keep track of the innovative activities of their rivals in making their own plans. The market-leader firm’s management is often concerned about the catch-up plans of its nearest rival, and that rival measures its performance and its reputation in terms of quality of technology against that of the market leader. I offer no specific citations because of court-imposed confidentiality rules.

5. Recall Schumpeter’s characterization of the role of competition: “in capitalist reality as distinguished from its textbook picture … [the] kind of competition which counts [is] the competition from the new commodity, the new technology … competition which strikes not at the margins of the profits and the outputs of the existing firms but at their foundations and their very lives” (1947, p. 84). Or, as Nelson puts it, “given that its rivals are induced … to invest in R&D a firm may have no choice but to do likewise” (1996, p. 52).

6. The nature of this “norm” is deliberately kept vague here, because in practice firms follow one another’s spending practices only roughly and approximately. This is partly because no two firms are perfectly alike, nor are the attitudes of their managements. Partly it is so because firms may be imperfectly informed about the R&D spending of their rivals. The norms may be vaguely based on sales, assets, market shares, or other criteria, or on a combination of these.

7. In chapter 14, it will be assumed that these outlays are fixed in nominal terms. This does not affect the argument here, but it does play an important role in the macro growth model in which the cost of R&D can change cyclically.

8. This statement somewhat exaggerates the effectiveness of the ratchets in preventing the economy from ever sliding backward in its expenditure on R&D. After all, even in machinery, ratchets sometimes slip. In R&D investment, firms may, for example, be forced to cut back their R&D expenditure if business is extremely bad; or they can simply make mistakes in planning how much to spend on investment; or they may be discouraged by repeated failures of their research division to come up with salable products. The economy’s ratchets are indeed imperfect, but they are there. They cannot completely prevent backsliding in R&D expenditure, but they can be a powerful influence that is effective in resisting such retreats.

9. Jones (1995) has argued that there are sharply diminishing returns to R&D effort, on the basis of data for research employment in the period 1950 to 1988. These apparently expanded more than fivefold while total factor productivity growth was “relatively constant or even declining.” However, McGrattan (1998) has disputed this conclusion. She provides longer-term data suggesting that returns to R&D effort do not decline nearly as sharply as Jones implies, if they decline at all. In any event, the argument on R&D spending here does not depend on the outcome of that discussion, but does assume that the marginal return to R&D activity is positive. For studies showing this, see Griliches (1992, p. 45) and Nadiri (1993, p. 11).

10. Here the word “steady” means “continuing” and is not meant to deny the possibility of diminishing returns.

11. This does not, of course, mean that there is a fixed relationship between R&D spending and rate of growth of the economy’s output. All that is assumed here is that a rise in R&D will generally increase growth to some degree relative to what it would have been otherwise, other things being equal.

12. Indeed, according to Maddison’s calculations (2001), this is all per capita GDP rose between 1500 and 1820.

13. There are noted analysts who argue that the true figure was far greater than this. See, for example, the work of William Nordhaus (1997). DeLong (2001) adds: “William Nordhaus brackets the growth of real wages over the past century as somewhere between a 20-fold and a 100-fold increase. Alan Greenspan—Chairman of the Federal Reserve—has [suggested adjustments of the statistics that lead] to an estimate of a thirty-fold increase in material wealth over the past century.”