FIGURE 16.1
U.S. private industry R&D expenditures, 1953–98: nominal vs. real (1982–84) dollars. Sources: National Science Board (2000) and Bureau of Labor Statistics (www.bls.gov).
Feedback: Innovation as a Self-Nourishing Process
Can capitalism survive? No, I do not think it can.
—Schumpeter, 1947, p. 61
The future is like everything else. It isn’t what it used to be.
—Attributed to Simone de Beauvoir
Economists can predict everything except the future.
—Recently cited by Alan Blinder
The previous chapter ended on a rather inauspicious note, showing at least one course of events that could end up with the incentives for routine invention activity attenuated, and the rising trajectory of resources devoted to R&D petering out or ceasing altogether. But that scenario was meant primarily to represent a possibility that by no means constitutes an immediate threat or even one for the foreseeable future. Chapter 15 was not meant to draw attention to any imminent crisis but, rather, to call attention to the dangers of the fallacy of composition, which, in this case, involved generalizing from the behavior of individual firms to the behavior of the economy as a whole.
The message of this chapter is that there are good reasons for optimism about the prospects for future innovation. One reason, simply put, is that innovation breeds innovation. In addition, I will draw attention to some very powerful countervailing forces that serve at least to mitigate the innovation-depressing process described in chapter 15’s macroeconomic model of endogenous innovation. These offsets, related directly to the structure of the model, will be described later in the chapter. But they are not the central focus of the discussion, which is, rather, the innovation feedback process, in which innovation leads to economic developments that, in turn, stimulate and facilitate the innovation process. And, finally, even though long-range economic forecasting is notoriously untrustworthy and perhaps worthless, I will conclude with the sanguine personal view that any slowdown in the pace of innovation is probably still far away, and that there is little evidence that the growth trend in innovative activity is about to lose steam.
THE INNOVATION FEEDBACK PROCESS
Recognized Ways in Which Innovation Induces Further Innovation
The idea that successful innovation leads to further innovation is hardly new. Earlier, I cited Nelson’s (1996) cogent observations on this issue. There are various clear ways in which innovation breeds further innovation. Most obviously, one new idea frequently suggests another, particularly when one invention calls for another to make it more effectively workable. The computer gave rise to the need for a mouse, the airship to the need for a parachute. The examples are many.1 One idea leads to another, as in the invention of transistors and the numerous applications that followed. A new product also invites R&D devoted to improvement of that product (as in the computer) or to the creation of superior substitutes (as in the Salk and the Sabin polio vaccines), or even to improvement of an old product that is threatened with replacement. A standard example is the great improvement in sailing vessels after introduction of the steamboat. Note the role of Schumpeterian competition here: the competitors of an initial innovator are driven to protect themselves by seeking to replicate or, better yet, leapfrog the previous technical improvement.
But that is only the beginning. A newly invented product leads to increased understanding not only of the need for additional and possibly supplementary products, but also of the means by which the new products can be created. That is, innovation also helps to facilitate research and development. And it does so in at least two ways—by teaching us new and more effective ways to carry out research, and by creating new instruments for use in the research process. The electronic computer and the Internet (which so substantially facilitates collaborative research) are but two recent and dramatic examples.
Successful innovation also widens the acceptance of innovative products by business enterprise. This is crucial as an encouragement to innovators and, even more important, as a means to ensure that such products are effectively and rapidly put into use. Earlier, in chapter 14, I pointed out that the problem that beset such activity in ancient Rome and medieval China was not a lack of invention, but the absence of systematization of the rest of the innovation process—the dissemination, adoption, and utilization of the new processes and products. In a market economy, the demonstrated success of innovation in a particular sector is all that is needed to attract the attention and the energies of entrepreneurs. And that, patently, is yet another crucial contribution of successful innovation to further innovation.
Finally, there is the most obvious component of the feedback process—the demonstrated monetary rewards of innovation as a stimulus to further innovation. In its crassest form, the publicized rich rewards of some innovators can be counted upon to attract others into this type of enterprise as surely as the discovery of gold can stimulate a gold rush. The financial success of one innovator makes entry into the activity more attractive to others, and makes it easier to raise the requisite capital. The waves of innovation that have recurred throughout recent economic history probably have been driven at least in part by this influence.
In short, there are many ways in which innovation facilitates innovation and begets further innovation, notably as new technology offers profitable new opportunities for such outlays. But, so far, I have described only the most obvious components of the feedback process. I turn now to elements that are less widely recognized, starting with a brief recapitulation of the critical role of the oligopolistic competition that, this book suggests, is so vital a component of the market economy’s innovation machine.
Competition Stimulates Innovation, and Innovation Stimulates Competition
Probably the most powerful force that may well lead to continuation of the remarkable growth in innovative activity is the adoption of innovation as the prime weapon of competition in many of the leading oligopolistic sectors of the economy. As has already been emphasized, the resulting “arms race,” in which each firm is forced to keep up with its rivals and to strive to outdo them, is like a Red Queen game, in which all players are forced to keep running as fast as they can in order to stand still. This is an engine of growth of enormous power, and it is a game that participants cannot easily quit. It is also the prime support for optimism about the prospects for continuation of the free-market economy’s historically unprecedented innovation explosion and its acceleration during the twentieth century. But recapitulation of this argument, though central to the analysis of this book, is not the point here. What is at issue, rather, is the feedback process, and it arises here because, as I will argue next, innovation facilitates and stimulates the very type of competition that drives the free-market innovation machine. That is, innovation encourages competition, while competition, in turn, is a key driver of the innovation process.
Competition, Innovation, and Foreign Trade as Mutual Stimuli
In the economic literature on antitrust issues, the contribution of innovation to competition is well recognized. For example, it is observed that new products and processes tend to shorten the life of a monopoly and to bring the possession of monopoly power by a dominant enterprise to an earlier end. Entrants that can leapfrog the incumbent monopolist with a better product or a more efficient process will often succeed in sharing the market with the incumbent, or in replacing the incumbent with a new dominant firm whose monopoly power is likely to be equally transitory, as still later innovators enter the market.2
But a second way in which innovation has contributed enormously to competition has clearly been its role as the key to the astonishing growth in trade among nations. And that, in turn, has greatly increased the power of competition in many industries. That enhancement of competitive pressures has further stimulated the innovative activities of the affected firms. Transportation’s growing speed and reliability and its falling costs have played an important part in the intensification of competition and have internationalized the innovation arms race. Since the Industrial Revolution, world output of goods and services has, as we know, expanded enormously. But the rate of expansion of trade has far outpaced that of production. Maddison (1995, p. 38) estimates that, since 1950, U.S. exports have risen nearly three times as a share of GDP. For the world as a whole, exports as a share of GDP expanded nearly fourteenfold since 1820.
All this was made possible by innovation, which resulted in a revolutionary reduction in the costs, time, and perils of transportation as well as communication. By the end of the eighteenth century, for example, navigators could determine longitude using the chronometer, making it far safer to sail without hugging the shoreline and to enter far nearer harbors that previously were too perilous. Steam propulsion, the end of wooden hulls, the enormous growth in size and speed of vessels, wireless communication, and a host of other profound technical changes were patently the key to the explosion of trade and communication.
The growth of the volume of trade has clearly enhanced competition. In the United States, industry after industry feels the threat of foreign rivalry to a degree never experienced before. Industries such as consumer electronics have to a considerable degree become a preserve of other nations. U.S. leadership in aircraft production is threatened for the first time since the end of World War II. And the automobile market is no longer primarily a domestic affair. The rising competitive pressures from foreign sources are not very different from domestic rivalry in their consequences for the behavior of the firm. Many examples could easily be cited to confirm their effect in intensifying the technological arms race that dictates the innovative activities of the firm.
This chain of relationships indicates yet another way in which the innovation process tends to counteract any tendencies toward diminishing returns and diminishing innovative activity that may affect that process. Innovation acts as a stimulator of the competition that tolerates no letup in innovative activity and that may force its further intensification.
Innovation Extends the Supply of Limited Resources
I turn next to yet another way in which innovation facilitates innovation, one that is also not widely recognized.
Ultimately, one of the prospective impediments to innovation is the finite character of the economy’s natural resources that are critical to utilization of the new products and processes that innovation provides. Thus, it may well be suspected that, in a world in which the economy’s finite natural resources can only be depleted, this will raise problems for innovation that will grow more serious with the passage of time. After all, most inventions must be embodied at least partly in concrete material objects, requiring metals, fuels, and other scarce resources for their utilization, and this may well become an impediment to innovation—both to its production and to its contribution to the economy’s output.
However, it can be argued that, in an important sense, the available quantities of the economy’s natural resources can be expanded and, indeed, that this has actually happened. This notion may seem bizarre, yet the fact that, over the decades, the real prices of so many of these resources have not been increasing, and that at least some of them have actually declined, must surely suggest that there is something to this notion. And there is a straightforward explanation.
Suppose that inventions constantly decrease the percentage of petroleum that is lost in the process of extraction from a well, and that another stream of inventions constantly increases the number of miles a given quantity of petroleum will enable a vehicle to travel. Suppose both of these developments move far faster than the rate at which the earth’s fixed physical stock of oil is used up. Then it should be clear that the inventory of prospective miles of transportation by petroleum-driven vehicles can actually have been expanded. The intellectual input will, in effect, have served to increase the supply of the physical input in the one sense that really counts: the still available capacity of that input to contribute to future output of the economy.3
The Acceleration Effect of Innovation on Production
It is important here to note again a relationship between the flow of innovation and the rate of economic growth that is not entirely obvious.
Each successful innovation contributes to the growth of the nation’s GDP. Thus, an economy in which R&D produces, for example, an unchanging output of one innovation per month 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 innovation flow that fuels output growth remains unchanging at one invention per month. In other words, the rate of growth of the value of GDP is an increasing function of the level of innovation. This acceleration relationship applies to innovation generally, so that, if the competitive market mechanism leads firms to devote a constant quantity of resources to R&D, we would expect continued growth of GDP to result.
Of course, the innovation arms race described in chapter 4 tends not only to prevent firms from decreasing their inventive activities, but also encourages them to increase their expenditures on innovation with the passage of time. The acceleration relationship just described tells us about the effects of this, too. It says 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. In other words, from a once-and-for-all increase in the level of expenditure on R&D we can expect a permanent increase in the rate of growth of GDP.
THE ROLE OF PRICE INFLATION: REAL VS. NOMINAL TARGETS FOR R&D EXPENDITURE AND THE RISING RELATIVE COST OF R&D
So far, I have discussed a general feedback mechanism that encourages a sanguine view of the prospects for innovation. But I have not yet addressed specifically the possible offsets to the less promising prospects suggested by chapter 15’s extended cost disease model.4 That analysis implied that one of the probable consequences of the innovation arms race and the growth of innovation stimulated by the feedback process is an ever-rising cost of R&D that can ultimately choke this activity off. The cost disease model raised the prospect of steady long-term decline in the real resources devoted to R&D, under the pressures of constantly rising relative costs of this activity—a result of its own success in raising productivity growth in other sectors of the economy. This story rests on two premises: first, money illusion, which leads firms to sustain only their nominal outlays on R&D (with no adjustment for the effects of price inflation), thereby allowing the real outlays to decline as price inflation grows; and, second, a paucity of price-insensitive innovation activity.
The role of money illusion here is the possibility that the rising dollar spending on innovation that is induced by the arms-race process will be insufficient to prevent reductions in the amount of purchasing power used for this purpose, as rising price inflation in the economy erodes the value of the dollar. For example, if private industry raises its dollar R&D spending by 20 percent but, at the same time, inflation cuts the purchasing power of the dollar by, say, 45 percent, then evidently the net result will be a decline, not an increase, in innovative activity.
This issue came up early in the book. The basic microeconomic model of chapter 4 yielded the conclusion that oligopolistic competition can be expected to enforce standards for R&D spending by firms in high-tech industries. Each firm is driven to live up to industry spending norms by the fear that, if it fails to do so, its rivals will come up with superior products and processes and thereby cut into its sales or drive it from the market altogether. Now the behavior of chapter 15’s model rests on the premise that these industry norms for R&D expenditures are set in terms that are nominal (not adjusted for inflation) rather than real (that is, inflation adjusted). The reason this assumption plays such a role in the model is easy to see. If firms raise their nominal expenditures sufficiently every time that relative productivity growth in other sectors increases the cost of innovative activities—that is, if they automatically prevent any fall in real expenditure—then the cost disease, by definition, will have no effect on the flow of innovation. But if the firms think in terms only of the number of dollars they devote to innovation activities, regardless of the purchasing power of those dollars, then rising prices and, in particular, any rising cost of R&D resulting from the cost disease will tend to offset and even undermine altogether the arms-race growth mechanism.
The question, then, is whether the expenditure norms in my analysis are fixed in real or in nominal terms, or something in between. Observation of business behavior suggests that business persons do adjust their expenditure to inflation, but that there is often a lag and often failure to recognize the full amount by which a rising price level has reduced the purchasing power of the amounts they budget. The available statistical evidence, as we will see, also indicates that the answer is somewhere in between the two extremes. Money illusion does appear to play a role in R&D funding, sufficient to preserve the relevance of the chapter 15 story. But there also appear to be competitive pressures that are strong enough to offset much of this price-inflation effect and thus to sustain the growth in the resources devoted to innovation. The relevant data for the United States seem not to be broadly consistent either with business behavior that is governed entirely by money illusion, at the one extreme, or with total absence of that influence, at the other.
Data provided by the National Science Board (2000) indicate that, since World War II, total U.S. nominal expenditure on R&D has risen in a near-exponential pattern. But the evidence indicating that the innovation arms race is not undermined by money illusion is that real U.S. R&D expenditure (deflated by the consumer price index for all items [CPI-All Items]) also rose at an average annual growth rate of more than 4 percent (from some $19 billion 1982–84 dollars in 1953 to nearly $140 billion in 1998). The pattern of R&D spending by private industry alone is very similar (figure 16.1). Nominal industry spending increased almost fortyfold (the curve labeled “nominal expenditures” in the graph) and real industry spending, calculated as before, rose more than sevenfold (the broken curve) over the forty-five year period (a remarkable average rate of growth of more than 5 percent per year).
Yet there were exceptions that seem consistent with some role of money illusion. Total U.S. real expenditures on research and development actually fell in eight of the forty-five years for which data are reported. The first significant instance was the era of the 1970s, when the economy experienced inflation extraordinary for peacetime in the United States. The National Science Board writes, “Starting in 1969 … and for nearly a decade thereafter, R&D growth failed to keep up with either inflation or general increases in economic output” (2000, p. 2–7). Real R&D spending declined in the years 1969, 1970, 1971, 1974, and 1975. In these years, then, firms corrected very imperfectly for inflation in their R&D outlays. The second period with very slow or even declining real R&D outlays included the end of the 1980s and the beginning of the 1990s, with actual declines in 1992, 1993, and 1994. But here the story does not fit in directly with a simple cost-inflation scenario, since the compensation of scientists and engineers slowed markedly between 1989 and 1993, and began to rise rapidly only after that (see figure 16.2).5
FIGURE 16.1
U.S. private industry R&D expenditures, 1953–98: nominal vs. real (1982–84) dollars. Sources: National Science Board (2000) and Bureau of Labor Statistics (www.bls.gov).
The generally upward trend in R&D spending is clearly consistent with the ratchet model of chapter 4, while the cut in real spending in the 1970s is consistent with the assumptions underlying the model of chapter 15. The implications of the second period of slow growth and real decline, also nearly a decade in duration, are less clear. Taken as a whole, the evidence indicates that R&D outlays by industry do not respond reliably to inflation by an offsetting rise in nominal spending.6
FIGURE 16.2
Median annual salaries for U.S. Ph.D. scientists and engineers (in nominal dollars) and U.S. private industry R&D expenditures (in millions of nominal dollars), 1973–97. Sources: National Science Board (2000) and personal correspondence, October 23, 2000, from Rolf Lehming, Director, Integrated Studies, Science and Engineering Indicators, National Science Foundation.
The point for the discussion of this chapter is that, although money illusion and related influences can indeed plausibly be assumed to result in some cut in real R&D spending when its cost increases, one must not take this observation to extremes. People in business do recognize inflation and they also, at least eventually, recognize when it has eaten substantially into real outlays. In the longer run, we can expect that the oligopolistic competition discussed in chapter 4 will lead to some adjustment in nominal outlays on innovative activities by business firms that may well continue to make up for the periods of erosion by rising inflationary costs and more. Accordingly, the trajectory of R&D spending reported by the data just cited does show an overall and persistent trend in the real resources devoted to R&D. The cost disease
may indeed pose some threat to growth in real outlays on R&D, but not nearly as severely or as persistently as the simplest form of the model may suggest.7
THE PROSPECTS: WHAT CAN WE EXPECT?
After this description of some of the forces that may conceivably terminate the extraordinary growth achievements of capitalism, as well as the powerful forces that work in the other direction, the reader is entitled to ask: Where does the balance lie? My answer is that there is certainly no evidence of any imminent weakening, much less termination, of human ingenuity, or of the flow of products it creates, or, perhaps more to the point, of the competitive mechanism that drives it on.8 This book has described a powerful mechanism that is built into the capitalist economy and that spurs it relentlessly to innovation and growth. The fierce oligopolistic competition in innovations, the serendipity between investment in the innovation process and the markets in licenses that ensure rapid dissemination of new technology, along with financial rewards to its proprietors, and the incentives that entice entrepreneurs into productive rather than rent-seeking activities—these are all powerful and continuing forces. These forces are supplemented strongly by the feedback process described in this chapter: the stimulus that innovation provides to further innovation. I see no evidence that any of these mechanisms is about to weaken.
The facts seem to support this inference from the theory. Recent decades have introduced a flood of extraordinary new products. After a period of slowdown during the 1970s and 1980s, which less sanguine observers described as the “end of the golden age,” productivity growth once more went on the march. There are even signs that the computer revolution is at last beginning to make a discernible contribution to economic welfare. U.S. economic growth in the last several years attained a pace that astonished everyone who has studied the general subject, though, predictably, it has since been slowed once again by recession. Of course, future progress can also be expected to continue to have its declines and recoveries. But I have yet to see any portents for the longer run suggesting that the economic accomplishments of the free market are about to undergo permanent decline.
However, it is not the purpose of the book to provide forecasts about prospective innovation and growth. My goal has been to suggest useful explanations of the incredible growth in output and innovation that has already occurred. I have offered a set of explanatory hypotheses along with some evidence and some analysis pertinent to each of them. Yet it must be recognized that the very nature of the subject condemns the evidence to be spotty and unsystematic, for how can one really hope to prove what has stimulated such things as invention and entrepreneurship, when neither of these can even be measured? Moreover, the set of explanations offered here is surely far from complete, though one can hardly aspire to completeness in such an arena.
None of this is meant to be apologetic; it is intended only to suggest the limitations inherent in my undertaking with some degree of frankness. On the contrary, I believe that, when the set of candidate explanations I have proposed is considered carefully, the importance of their role will appear self-evident.
Surely, competition with innovation as a weapon must be a powerful stimulus for invention and its utilization in the economy.
Surely, the profits offered by dissemination via the market mechanism must materially speed the replacement of obsolete products and production processes.
Surely, routinization must help to reduce the fortuitous component in the stream of innovation.
Surely, success in innovation has led to further innovation.
And, surely, such developments as the rule of law must have vastly enhanced the incentive for independent innovative activity.
That this cannot be the entire story I am readily prepared to acknowledge. It nevertheless seems evident that the forces to which I have drawn attention have played a crucial role and, by themselves, could have contributed a substantial explosion in innovation and growth. I claim no more, but with this claim my tale must end.
___________
1. Landes (1998, pp. 191–92) writes, “Innovation was catching because the principles that underlay a given technique could take many forms, find many uses. If one could bore cannon, one could bore the cylinders of steam engines. If one could print fabrics by means of cylinders (as against the much slower block printing) one could … print word text faster than by the up-and-down strokes of a press and turn out penny tabloids and cheap novels by the tens of thousands. Similarly, a modified cotton-spinning machine could spin wool and flax. Indeed, contemporaries argued that the mechanization of cotton manufacture forced these other branches to modernize. … New, new, new. Money, money, money.”
2. Yet the literature emphasizes that new technology can also contribute to monopoly power, for example via what are called “network effects.” These are cases in which the value of the product is crucially dependent on the number of users and the compatibility of the products they use, as in the case of a computer program that various users employ to communicate with one another. The innovating firm that creates the most popular new program may achieve dominance of the market and the compatibility requirement may make it difficult for rivals to enter.
3. Yet one can expect limits to the extent that innovation can expand what we may call the effective supply of natural resources. On this, see Baumol, Blackman, and Wolff (1989, pp. 357–58).
4. The main issue is whether firms deal with real or nominal costs and that will be discussed below. In addition, we should note that independent R&D pursuits, as related to parameter h (price-insensitive R&D activity), are the nonroutine innovative activities, largely conducted outside the premises of the giant corporations. The magnitude of this parameter, however, is not related to the amount of such nonroutine activity alone. This is because h is a measure of responsiveness of price-insensitive R&D to changes in total R&D. Moreover, independent R&D is surely not entirely price insensitive. Indeed, it is highly plausible that even independent entrepreneurs and researchers are in many cases influenced by the cost of R&D and its implications for the prospect for profit from their activity. It is also very likely that some of their activity will be held back or even terminated if rising costs make it more difficult to raise the funds needed to carry out a successful research and development program.
Still, one suspects that the enthusiasm of at least some independent inventors and entrepreneurs will lead many of them to disregard rising costs and carry on in pursuit of their goals, readily adjusting to the enhanced financial difficulties. The history of invention tells us of many individuals—such as John Harrison, inventor of the first accurate marine chronometer, which enabled navigators to compute their longitude at sea, and Charles Goodyear, who discovered vulcanization, the process that improves the physical properties of rubber—who devoted lifetimes to the monomaniac pursuit of their targets. Their stories include episodes of financial stringency that nonetheless did not stop their activities. In short, there is inventive activity that, for all practical purposes, is not deterred by rising costs—at least if those cost increases are not enormous or unforeseeable.
This implies that parameter h is not zero and may conceivably be well above it, meaning that a rise in successful R&D in the economy can lead to a proportionately significant increase in price-insensitive innovative activity. Basically, if success in innovation is sufficiently effective in breeding further pursuit of innovative success, this can overcome the depressing effects of the systematically rising cost of the process.
5. The declines in figure 16.1 are enhanced by the possibility that the cost of R&D rose substantially more quickly than the overall rate of inflation as a result of the cost disease. This is illustrated by deflating the graph using the assumption that R&D costs rose as quickly as those of health care. Using the CPI-Medical Care price index for this purpose, we obtain the flattest of the curves in figure 16.1. The qualitative properties of the time path still remain the same—a general rise over the forty-five-year period as a whole, more than trebling real expenditures. But, calculated using the medical care index, there are sixteen years of decline in real R&D expenditures (about one-third of the period).
6. The conclusion is supported by a regression using the preceding data for 1953–98, with the growth in nominal U.S. R&D outlays by private industry as the dependent variable, and time and the rate of growth of inflation as the two independent variables. This yielded a very small and statistically insignificant coefficient for the latter.
7. For an excellent characterization of the literature on money illusion, and further empirical evidence that behavior in reality falls somewhere well in between the two extremes of exclusive reliance on either real monetary quantities or nominal quantities, see Shafir, Diamond, and Tversky (1997).
8. We must be careful to avoid prediction based only on extrapolation of temporary slowdowns or leaps of progress and the assumption that they will persist indefinitely. When growth in the United States slowed markedly between about 1972 and 1990, many were prepared to conclude that the end was near for American economic leadership. I cannot resist claiming that my colleagues Sue Anne Batey Blackman and Edward Wolff and I were among the first to gather evidence indicating that there was no basis for this gloomy forecast, though we, too, were initially inclined to accept it. Another noteworthy example is the long period during which the adoption of computers displayed little or no effect on productivity. Paul David (1989) was able to show, in his usual perceptive way, that it had taken a correspondingly long period for electrification to make its mark on output and productivity, warning us that we should not expect major instantaneous economic effects from technical breakthroughs.