Some of the major qualifications and refinements that have emerged from recent research on the technical progress question are considered here. Essentially the argument comes down to this: Research and development is not one large, undifferentiated whole. Distinctions with respect to productivity, major versus minor inventions, other market structure dimen- sions, research responsiveness, basic versus applied research, large, complex versus “normal” R & D, and invention versus development would appear to be especially relevant.
It goes without saying that if one chooses among the data virtually anything can be “proven.” But however useful conspicuous observations may be as a basis for formulating hypotheses, they constitute a hazardous basis for claiming generality. Bell Labs, General Electric, and Du Pont each possess impressive research facilities and personnel, but they do not, apparently, represent average tendencies. (For studies in contrast, consider Western Union and U.S. Steel.) Only a second glance at the statistics is sufficient to dispel any initial impression that the concentration and progressiveness conditions observed in a few glamorous industries are representative (Jewkes et at., 1959, pp. 167-73). The studies reported above reinforce this judgement. Possibly, however, a more refined treatment of the data will better support the Galbraithian position.
1. Productivity
If, over some range, there are scale economies (or diseconomies) as- sociated with R & D, the productivity of R & D will vary with expenditures accordingly. Recall in this connection the intra-industry results which show that absolute firm size and absolute R & D spending move together. As shown in Figure 8, however, R & D intensity (expressed as a percentage of sales) is observed to level off. if not turn down, as firm size increases. But might the productivity of R & D programs in very large firms be greater than are those of their smaller rivals, in which case this latter relation is explained by scale economies associated with R & D?
The evidence, though not extensive, does not support this proposition. If productivity can be judged by patents issued per dollar R & D expenditure, the performance of the very largest firms in most industries is somewhat below that of their large but somewhat smaller rivals (Scherer, 1965b). Similar results, using a somewhat more refined measure of productivity, albeit for a smaller sample of industries, are reported by Mansfield. “When the size of R & D expenditures is held constant, increases in the size of firm are [usually] associated with decreases in inventive output” (1968, p. 216). Comanor found that diseconomies of scale in the pharmaceutical industry were encountered at even moderate firm size (1965, p. 190). The Gal- braithian hypothesis, regarded as a statement of average tendencies, evidently derives little comfort from productivity considerations.
2. Major Versus Minor Inventions
Just as the productivity of R & D is a relevant distinction, so is the economic importance of inventions. If the importance varies systematically by the source of the invention, the.weight to be assigned to the several sources of invention needs appropriately to be adjusted.
Hamberg has set forth and evaluated the hypothesis that large, industrial laboratories are apt to account for a larger proportion of minor than they are of major inventions (1963). As he observes, the hypothesis is not altogether original. Some of the reasons why such a relation might be expected, however, had previously been neglected. Both economic and organizational factors are adduced in favor of the hypothesis.
Basic to the argument is the proposition that major new inventions are usually high-risk activities. The large corporation may be structurally and/or constitutionally ill-suited to bear these. But then a firm need not itself be the source of major, new inventions in order to participate in the commercial development of such items: it may be able to buy into successful new developments originated elsewhere—an aspect of the argument developed in Section 4, below. That it is not especially well-suited to perform high-risk inventive research is, necessarily, a comparative rather than an absolute proposition. It relies mainly on the contention that large organizations, business or otherwise, experience internal organizational failures. Moreover, while claims that the large firm possesses offsetting financial advantages in relation to small firms have merit, the argument requires, at least with respect to high-risk activities, certain qualifications.
Hamberg takes the position that “The one feature of large business size and monopoly power that appears to carry potential weight is associated with the usually large — absolute — profits of the big monopolistic firms as a source of funds to finance research and innovation” (1966, p. 43). Small firms, perhaps especially successful ones, tend often to find that internal resources are simply inadequate to meet their growing financing needs. These internal financing advantages of large, established firms are apt to be reinforced by their superior access to debt financing. Thus, whereas the established firm has tangible assets and a demonstrated operating capability to offer as security, the small firm requiring research funding has only an uncertain prospect of success to offer. Debt financing for the small firm in these circumstances is not apt even to be feasible.
Internal and debt financing do not, however, exhaust the sources of possible funding for inventive activity. The equity capital market, and in particular that part of the equity capital market that is referred to as venture capital (to emphasize its high-risk character), needs also be considered. Here small enterprises may actually enjoy a structural advantage. Investors in tax brackets for which high-risk capital gains investments are attractive may, because of inability to select among the investment projects in a large firm and appropriate the undiluted gains of risky activities, direct their funds to small, specialized enterprises instead. The financing disadvantages of the small firm for venture capital purposes may thus be less substantial than is often suggested.
The organizational limitations of the large firm as a source of major, new inventions are attributable to its bureaucratic character. This matter is treated more extensively in the discussion of organizational failure in Section 4. Here a simpie sketch of the argument is provided as it applies to inventive research. It relies in large part on the proposition that “in the large firm, the team system and an ‘average man’ mentality generally prevail” (Hamberg, 1963, p. 107). Hamberg cites the research views of the research coordinator of Standard Oil of Indiana, Daniel P. Barnard: “We find the self- directed individual being largely replaced by highly organized team attack in which we employ many people who, if left entirely to their own devices, might not be research- minded. In other words, we hire people to be curious as a group We are undertaking to create research capability by the sheer pressure of money” (1963, p. 107).
Such an approach to research places a high premium on harmony. An emphasis on research results develops which tends to favor support for short- reach programs (Hamberg, 1963, pp. 109-110). As Jewkes, Sawers, and Stillerman point out, such attempts at organization may threaten the spon- taneity which is often regarded as vital to a successful research atmosphere (1959, pp. 132, 182-183). Creative scientists and inventors may be driven to locate elsewhere.
So much for the arguments, what does the evidence support? What, historically, have been the sources of invention? What is the professional standing of corporation scientists? What types of research are most commonly undertaken in the large corporation?
Hamberg’s survey of the sources of invention makes it evident that major new developments have historically (and, although perhaps in somewhat reduced degree, continue to) come preponderately from sources outside the large corporation. Grosvenor, in a study published in 1929, found that only twelve of seventy-two major inventions made since 1889 had originated in industrial laboratories. Jewkes, Sawers, and Stillerman traced the origins of sixty-one major inventions of the twentieth century, most of which occurred between 1930 and 1950. Of these, twelve could be attributed to the laboratories of large corporations, thirty-three were the work of independent inventors, five were from laboratories of small firms, and the remainder were unclassifiable. Hamberg examined twenty-seven inventions made during the period 1946-1955. Seven were the products of large industrial laboratories, twelve were from independent inventors, and the remaining were from small firms, universities, and an agricultural experiment station (1963, p. 96; 1966, p. 16). Schmooklers (1957) study of patent statistics over the period 1950-1957 disclosed that between fifty to sixty percent of the recent inventions were from inventors working outside the organized research groups of the corporate industrial laboratories. Although the relative and possibly absolute importance of the independents has been gradually declining (Nelson et al., 1967, p. 57), the base is too large and the rate of decline too slow for the inventive influence of this group to pass quickly to extinction.
The suggestion that the corporate research atmosphere is not calculated to attract eminent scientists also finds support (Hamberg, 1963, p. 108; Scherer, 1970, p. 354). This is not necessarily a particularly surprising or even undesirable result—especially in view of the systems approach to the innovation question that is advanced in Section 4. It suffices here to point out that, as a matter of perspective, industrial scientists in the large corporation are not, typically, ten feet tall. Again, there are occasional exceptions, but the focus here is on average tendencies. If smaller firms are able merely to attract competent engineers and scientists, they are not at a serious talent disadvantage.
Consider finally the research emphasis of industrial laboratories. Nelson, Peck, and Kalachek conclude: “Outside defense and space-related R & D . . . and possibly some segments of the civilian electronics and chemical indus- tries, the bulk of corporate R & D is modest design improvement work not reaching very far—the type of work that results in yearly changes in auto- mobile design, gradual improvements in the automaticity, speed, and capacity of machine tools, rather than radically new products and processes” (1967, p.54). It might not be unfair to characterize industrial R & D as mainly mundane rather than exotic, contrary to the conventional wisdom. Indeed, the majority of what gets reported as R & D expenditure is concentrated in the production process —including design, engineering, and tooling, as well as manufacturing and marketing startup costs (Mansfield, 1968, p. 106). A study by Mansfield andHamberger concluded that “the bulk of R & D carried out by large corporations is relatively safe and aimed at fairly modest advances in the state of the art” (Mansfield, 1969, p. 66). Even in a corporation as large and competent as Du Pont, of the major inventions that it implemented from 1920 to 1949, many more, proportionately, of the improvement inventions than of the new product inventions were originated in Du Pont’s laboratories (Mueller, 1962, pp. 342-343).
Although this may appear to relegate the large corporations to a position of secondary importance, this is not intended. For one thing, an improvement emphasis by the large firms may represent a rational allocation of resources in a systems sense. For another, while there may be a “good deal of truth” in the improvement hypothesis, “technical change in many industries may be due in considerable measure to the cumulative effect of many ‘improvement’ inventions” (Mansfield, 1968, p. 93). Put differently, technical change in some industries is mainly of the improvement sort. Finally, in some industries and for certain types of research, the reported average tendencies between new product and improvement innovations do not hold. This is discussed, below.
That, in principle, the distinction between major and minor inventions is important would be freely conceded by most of those who have been involved in the industrial R & D controversy. That, in most industries, the large indus- trial research laboratory appears mainly to be a source of improvement inventions and that other sources play so substantial a role in the major new development aspects of research comes as some surprise to many—including myself. Not that small firms and inventors would be expected to play no role in this activity, but that they are so important is almost astonishing.
3. Other Dimensions of Market Structure
With the exception of concentration, the above discussion runs exclu- sively in terms of absolute size. The focus, thus, is mainly on the bigness rather than the monopoly claims of the neo-Schumpeterian hypothesis. If, however, the latter is also to be addressed, the frame of reference needs to be expanded. Relevant to a judgment of the influence of monopoly on innovation would be the condition of entry and changes in relative shares. These in turn may be influenced by such matters as product differentiation, extent of vertical integration, and the degree to which parallel action is observed.
Scherer concludes that the evidence from case studies supports the view that actual and potential competition, present at least in moderate degree, stimulate technical progress, “both as direct sources of innovation and as spurs to existing industry members” (1970, p. 377). This appears to be a majority opinion. Studies by Comanor (1967) and Phillips (1966) address other aspects of market structure, but the results have not been wholly conclusive or the welfare implications obvious. Perhaps for the present it is sufficient to rest on the observation that, on the average, concentration does not appear to have a significant influence on innovation and that, in a systems sense, an active market in inventions favors progressiveness. The latter is considered further in Section 3.
4. Research Responsiveness
Two questions are raised by the matter of research responsiveness First, are oligopolistically organized industries more adept at exploiting their technological potential? Second, are there science related industrie! which, by reason of the opportunities for commercial development affordec by their close connections with a sophisticated external science base, stanc apart as exceptions from the general argument? Consider the oligopoly argument first.
Ideally, one would like to test whether there is any systematic relatioi between market structure and the proclivity of an industry to take advantage of its technological opportunities. This would require that realized achieve ment be compared with technological potential. Characterizing the latter ii a sufficiently precise way to make meaningful comparisons does not, how ever, appear to be feasible. Nevertheless, those who espouse the case fo oligopoly on account of its superior progressiveness properties appear t< rely in large part on the proposition that oligopoly—by reason of the ric network of interfirm relations among the member firms (in which bot offensive and defensive research considerations play a role)—induces mor complete exhaustion of an industry’s innovative potential. Even if onl indirect evidence can be brought to bear on the argument, this presumabl would be useful.
If, as seems plausible, the richness of oligopolistic interaction shoul usually be strongest among the largest firms in the industry, then oligop( listically induced proclivities to innovate ought to be evident by comparin the relative research performance of the largest firms in less concentrated industries with those with greater concentration. The largest firms in the more concentrated industries ought to perform relatively better. The data, though limited, contradict the argument (Williamson, 1965). Scherer contends that there is “considerable evidence” that dominant firms are “slow innovators but aggressive followers” (1970, p. 371). For the present it would appear to be judicious to regard the oligopolistic interaction hypothesis at best to be unproven and probably doubtful.
Consider now the possibility that in certain science based industries, market structure is the outcome of at least partly exogenous technological change. Phillips is the leading exponent of this view. He argues that the adaptations by firms which operate in an environment in which a “related science creates more or less continuous opportunities for innovation” can alter market structure, with the result that only a few successful firms survive (1970, p.15). He makes a compelling case for this in the commercial airline industry (1970). Default and stochastic market failure considerations not unlike those discussed in Chapter 11, below, help to explain his results.
There is, more generally, broad agreement that the technological oppor- tunities in the chemical and electrical industries are greater than in the mechanically based industries (Scherer, 1965b, pp. 1100, 1121; Nelson et al., 1967, pp. 40-43). Also, Nelson, Peck, and Kalachek observe that “science- based inventions have been the major factor differentiating the products and processes of the twentieth from the nineteenth centuries” (1970, p. 40). This subset of industries is thus of special significance in assessing the market structure-innovation issues. Do such industries display quite different innovation in relation to firm size characteristics than come through from the average tendencies reported earlier?
The matter, unfortunately, has not been extensively studied. Casual references to what is often referred to as the “chemical exception” seem to suggest that certain industries do violate the usual rules, and it will be recalled that Mansfield did find that the largest firms in the chemical industry spent more on R & D, relative to sales, than did somewhat smaller firms (1968, p. 94). Judging from patent statistics, however, Scherer reports: “For both chemicals (with stone, clay, and glass) and electrical equipment, there may be slightly increasing returns up to sales of roughly $500 million, but beyond that point a definite flattening out of the patenting function is evident” (1965b, p. 1110). What would seem to be required really, is a series of intensive industry studies — such as Phillips’ study of the commercial airline industry—of the potentially exceptional cases. Which industries genuinely display exceptional tendencies, and what are the limits of the argument?
5. Basic Versus Applied Research
While rarely explicit, the suggestion is often made that much of industrial research, especially the research conducted by the larger companies, can appropriately be designated as basic research. In a gross sense at least, this view must be rejected. Industrial R & D in 1961 was predominantly develop- ment (seventy-eight percent) and applied research (eighteen percent), with basic research accounting for only four percent of the total (Nelson et al., 1967, p.55). Although large firms are more apt to sponsor basic research than their smaller rivals, major in-house basic research occurs in only a few industries: eighty-three percent was attributable to six industries, with four of these (chemicals, electrical equipment and electronics, aerospace, and petroleum) accounting for over seventy percent (Nelson et al., 1967, p. 56). This raises again the science based exception question considered above.
Lest the tail wag the dog, one should, presumably, regard arguments that favor big business on account of differential basic research propensities with caution. This is not to say that the argument lacks merit in certain of the science based industries which, as pointed out above, have had a dispropor- tionate impact on twentieth-century life .styles. But the argument does not appear to warrant great weight when applied across the board; a discriminat- ing attitude with respect to basic research is thus indicated. Even within the science based subset, giant size is not clearly required.
The study by Mansfield and Hamburger of the R & D programs of twenty- two major firms in the chemical and petroleum industries bears on this matter. They found that although the “largest firms in the sample devote a larger proportion of their R & D budget to more basic, more risky, and longer term projects than their smaller competitors … the differences between the largest firms and firms one-half their size are seldom large, if they exist at all” (Mansfield, 1969,p.71). 9 Thus, even in industries which , in a comparative sense, are known to direct a high percentage of R & D expenditures to basic research, giant size does not appear to favor basic research support of unusual proportions.
6. Large, Complex Versus “Normal” R & D
The romantic notion that the large industrial research laboratory is typically engaged in basic research on major new product developments is frequently buttressed by claims that the imperatives of modern technology are such as to leave little opportunity for any but the very large firms to con- duct research on the requisite scale.125 126 The argument, apparently, is that the so-called chemical exception is becoming the rule. Although there may be some tendencies in this direction, these appear to be limited rather than general and, even where observed, are gradual. For those who take the posi- tion that the near term future is apt to be a simple extrapolation of the past, there appears to be little reason to expect that the relations reported above will soon be drastically upset.
There are, nevertheless, projects the size and complexity of which require unusual research resources. Examination of military and space R & D suggests that great absolute size may be necessary to support much of this research (Nelson et al., 1967, p. 53) — although a case can be made that some of this bigness is attributable to the choice of inferior research and contracting practices by the military services (Kaysen, 1963; Williamson, 1967). If, however, the prevailing approach to military R & D is taken as given, great size may be necessary adequately to support military R & D efforts. Private R & D also, occasionally, involves “complex multi- component systems or complicated chemical processes” for which great size is important (Nelson et al., 1967, p. 53).
One should take care, however, lest the argument be inappropriately generalized. Atomic energy and Apollo programs are exotic and conspicuous, but they are not really typical. It is, moreover, interesting to note that the Apollo program was conducted not by a single, integrated contractor but by a series of independent corporations coordinated by NASA.127 Although this may have been dictated partly by reasons of time pressure, the strategy can also be supplied with an underlying contractual rationale (Williamson, 1967). In any event, projects that require a large, complex R & D capability are less common in the private sector and would appear to constitute an identifiable subset — for which exceptions to the main argument may be necessary — rather than the general case (current or prospective).
7. Invention Versus Development
The distinction between invention and development in the treatment of innovation will be considered in greater detail in Section 4. Suffice it here to point out that while modest resources are frequently sufficient to support invention and even early stage development work (Jewkes et al., 1959, pp. 210- 212), later stage development often incurs much greater expense (Scherer, 1970, p. 356). Although giant size is rarely indicated on this account, large size may be. Individuals and small firms that engage in inventive and early stage development work may, therefore, eventually have to face the prospect of achieving large size if the innovation is to be brought successfully to completion. Firms that are already large and possess extensive research laboratories and supporting facilities presumably enjoy an advantage in the later stages. Thus, unless smaller firms can efficiently secure access — on their own or through market processes — to an equivalent capability, they may be deemed to experience serious innovative weaknesses.
Source: Williamson Oliver E. (1975), Markets and hierarchies: Analysis and antitrust implications, A Study in the Economics of Internal Organization, The Free Press.