FS I V 96 - 7
New Firms and Creating Employment
David B. Audretsch
May 1996
IS S N N r. 0 7 2 2 -6 7 4 8
Forschungsschwerpunkt Marktprozeß und Unter nehmensentwicklung Research Unit
Market Processes and Corporate Development
Zitierweise/Citation:
David B. Audretsch, New Firm s and C reating Em ploym ent, Discussion Paper FS IV 96 - 7, Wissenschaftszentrum Berlin, 1996.
Wissenschaftszentrum Berlin fur Sozialforschung gGmbH, Reichpietschufcr 50, 10785 Berlin, Tel. (030) 2 54 91 - 0
New Firms and Creating Employment
Policy makers across Europe, and certainly in Germany, as well as N orth America have recently proclaimed that small firms are the engines o f job creation and should be encouraged.
This claim provides a striking contrast to government policies pursued only a few years ago based on the assumption that large firms create the most jobs. The purpose o f this paper is to pull together what scholars have identified as the relationship between firm size and job crea
tion. Such studies have been undertaken across a broad spectrum o f countries, industries and time periods, creating fairly conclusive evidence linking job creation and firm size. M ost o f the job creation literature has focused on the question, D o large or small firm s create the most jo b s? but has generally ignored the more fundamental question o f Why should jo b creation vary across firm size and firm age? Therefore, this paper introduces theories o f corporate downsizing and entrepreneurship which provide a theoretical lens through which the appro
priate public policy, implied by the plethora o f job generation studies, is evaluated.
ZUS AMMENF AS SUNG
Neue Unternehmen und die Schaffung von Arbeitsplätzen
Wirtschaftspolitiker in Europa und besonders in Deutschland, aber auch in Nordamerika haben in jüngster Zeit verkündet, daß kleine Unternehmen die Triebkraft für die Schaffung von Arbeitsplätzen seien und sie deshalb zu unterstützen sind. Dies steht nun ganz im Gegensatz zu der noch vor wenigen Jahren verfolgten Wirtschaftspolitik, die davon ausging, daß Großunternehmen die Mehrzahl der Arbeitsplätze schaffen würden. Im vorliegenden Beitrag werden bisherige Forschungsergebnisse über die Beziehung zwischen Unternehmensgröße und der Schaffung von Arbeitsplätzen erörtert und systematisiert. Die bisherigen Ergebnisse beziehen sich auf ein breites Spektrum von Ländern, Industriezweigen und Zeiträumen und sind ziemlich schlüssig im Hinblick auf den Zusammenhang der Schaffung von Arbeitsplätzen und der Unternehmensgröße. Der überwiegende Teil der Untersuchungen zur Schaffung von Arbeitsplätzen hat sich auf die Frage konzentriert, ob große oder kleine Unternehmen die meisten Arbeitsplätze schaffen. Ignoriert wurde dabei die Frage, warum die Schaffung von Arbeitsplätzen sich hinsichtlich der Unternehmensgröße und des Alters der Unternehmen unterscheiden sollte. Ausgehend von dieser Problemlage wird in diesem Beitrag das Konzept für eine neue Theorie des Schrumpfens von Unternehmen und des Unternehmertums vorgestellt, die einen Bezugsrahmen bietet, um die Angemessenheit wirtschaftspolitischer Maßnahmen zu beurteilen.
As unemployment in Germany surpassed four million, and stood at 10.8 percent o f the labor force, it is not surprising that Chancellor Helmut Kohl would undertake some action to spur the creation o f new jobs. Perhaps what is more surprising is the main emphasis announced by the Chancellor in the January 30, 1996 Initiatives f o r Investment and Em ployment1 on new and small firms. The first and main point o f this program con
sists o f a commitment to the "creation o f new innovative firms."2 The rationale underly
ing this commitment is stated factually by the Chancellor in the Program: "New jobs are created mainly in new firms and in small- and medium-sized enterprises."3
This Weltanschauung apparent in the Kohl jobs program represents a sharp departure from that found both in Germany as well as throughout Europe in earlier years. For example, in his widely read, Le D efi American, Servan-Schreiber (1968, p. 159) advo
cated, "the creation o f large industrial units which are able both in size and management to compete with the American giants (...). The first problem o f an industrial policy for Europe consists in choosing 50 to 100 firms which, once they are large enough, would be the most likely to become world leaders in their fields." And as is clearly documented in the 1988 Cecchini Report, there were significant forces in Europe prepared to imple
ment Servan-Schreiber's policy advice.
How could the Kohl Administration draw an inference about the source o f jobs, growth and international competitiveness that is seemingly at odds with both the scholarship and policy thinking o f just a few years prior? In fact, the last decade has seen a virtual explo
sion o f studies trying to identify what types o f firms create new jobs and which ones tend to destroy them. The particular focal points o f this literature have been on two main enterprise characteristics — size and age. These studies have been undertaken across a wide spectrum o f countries, industries and time periods and have attempted to answer a simple but important question, "Who creates new jo b s? " The nature o f what has become known as the literature on job generation is perhaps exceptional in economics. The debate has typically been on measurement, methodology, and interpretation — that is, getting the facts straight — rather than on developing theories to explain why new and small firms or alternatively, large and incumbent enterprises should have a comparative advantage at generating employment.
The purpose o f this paper is to draw out and portray what has been learned from the job generation studies and to propose several theories why a number o f stylized facts are emerging in this literature. In the following section the most salient findings o f the job
This was announced as the Aktionsprogramm fü r Investitionen und Arbeitsplätze ( "Soziale Ein
schnitte und Steuerreform sollen Wirtschaftswachstum anregen: Bundesregierung beschließt Akti
onsprogramm für Investitionen und Arbeitsplätze," Der Tagesspiegel, 31 January, p. 1).
2 The original text of the Aktionsprogramm states, "Offensive für unternehmerische Selbständigkeit und Innovationsfahigkeit" ("Ein Kraftakt zu Rettung des Standorts Deutschland," Frankfurter Allgemeine, 31 January, 1996, p. 11).
3 Ibid. The original text reads, "Neue Arbeitsplätze entstehen zumeist in neugegründeten Unter
nehmen und im Mittelstand."
generation studies are integrated together in a coherent picture. Theories o f corporate downsizing and entrepreneurship are provided in the third and fourth sections. Policy implications from the evidence and theories are suggested in the fifth section. Finally, a summary and conclusion are provided in the sixth section.
2 The Job Generation Debate 2.1 W hy Does It Matter?
Some fifteen years ago, David Birch revealed some startling findings from his long-term study o f U.S. job generation. Despite the prevailing conventional wisdom at that time, Birch (1981, p. 8) reported that, "whatever else they are doing, large firms, are no longer the major providers o f new jobs for Americans." Rather, Birch(1981 and 1987) claimed to have discovered that most new jobs emanated from small firms.
Birch's findings triggered a storm o f controversy that has only intensified in the 1990s.
There are at least three major debates underlying this controversy. First, at the time Birch announced his results, the preoccupation o f the literature in industrial organization, the field most directly associated with studying markets, was to identify the extent o f con
centration in markets and its effects on economic performance. The long-term trend had clearly been identified as an increased concentration in economic activity both at the ag
gregate as well as at the market level. For example, the percentage o f total U.S. manu
facturing assets accounted for by the largest 100 corporations increased from about 36 percent in 1924, to 39 percent after the Second World W ar to over 50 percent by the end o f the 1960s, causing F.M. Scherer to conclude (1970, p. 44) that, "Despite the statisti
cal uncertainties, one thing is clear. The increasing domestic dominance o f the 100 largest manufacturing firms since 1947 is not a statistical illusion."
Consistent with the trend towards increased concentration was the shift in economic activity away from small firms and towards large enterprises. The small-firm share o f employment decreased in every major sector o f the U.S. economy between 1958 and
1970. Perhaps most striking was the decrease in the share o f employment accounted for by small firms o f nearly one-quarter in the manufacturing sector (Audretsch, 1995).
Thus, given the state o f knowledge in the industrial organization literature, it was not readily apparent how to reconcile Birch's claim that 80 percent o f the new jobs were created in small firms with the empirical evidence pointing towards an ever-increasing concentration o f economic activity.
A second debate emerged regarding the exact methodology, application and interpreta
tion o f the underlying data used to make inferences in the job-generation studies (Armington and Odle, 1982; Storey and Johnson, 1987; FitzRoy, 1989; and Brown, Hamilton and Medoff, 1990). As Brown, Hamilton and M edoff (1990) pointed out, job generation may be a deceptive measure because many o f the newly generated jobs sub
sequently disappear. That is, without consideration o f the number o f job disappearances, focusing solely on the amount o f job generation emanating from small firms is misleading
and results in an overstatement o f the amount o f economic activity actually stemming from small firms (Storey, 1990). In addition, Birch's analysis was based on Dun and Bradstreet records. This data source has been criticized for not being particularly accu
rate with respect to employment counts and for providing only partial coverage o f small plants and firms.
A third debate involving anecdotal evidence has emerged in the popular press. For example, The Economist reports that, "Despite ever-larger and noisier mergers, the big
gest change coming over the world o f business is that firms are getting smaller. The trend o f a century is being reversed. Until the mid-1970s, the size o f firms everywhere grew;
the numbers o f self-employed fell. Ford and General M otors replaced the carriage- maker's atelier; McDonald’s, Safeway and W.H. Smith supplanted the corner shop. N o longer. N ow it is the big firms that are shrinking and small ones that are on the rise. The trend is unmistakable — and businessmen and policy-makers will ignore it at their peril."4
2.2 The Evidence
Perhaps in response to the controversy triggered by Birch's study, a wave o f studies emerged during the 1980s examining the links between firm size and age on the one hand and job creation on the other. In a 1989 article on "The Disciples o f David Birch," IN C magazine reported that "The debate keeps raging: How important are small, growing companies to the U.S. economy? Do they really create most o f the new jobs? There's a lot o f bad information on the subject being bandied about — but increasing numbers o f researchers are assembling hard evidence, even going out and counting companies them
selves."5
Such job generation studies were also undertaken outside o f the United States. For example, a study for Italy (Invernizzi and Revelli, 1993) found that overall job creation and destruction accounted for 8 percent o f annual manufacturing employment during the 1980s. The entry and exit o f firms accounted for between one-third and one-quarter o f gains and losses in employment. As in the case o f the United States, small firms were found to contribute positively to net employment, while large firms were found to be net job destroyers.
Gallagher, Daly and Thomason (1991) used Dun and Bradstreet credit rating data to measure job generation in the United Kingdom over the 1980s. They found that between 1985 and 1987 small firms provided 48 percent o f all new jobs, greatly exceeding their share o f employment, which was 21 percent in 1985. In contrast, the largest enterprises, defined as having at least 1,000 employees, provided only 13 percent o f all new jobs, although they accounted for 37 percent o f 1985 employment. Their results were gener
ally consistent with two previous studies they had undertaken for the 1971-81 and 1982- 84 periods.
5
"The Rise and Rise of America's Small Firms," The Economist (21 January, 1989, 73-74).
"The Disciples of David Birch: A New Generation of Researches is out to Discover How the U.S.
Economy Really Works," INC, January 1989, 39-45.
While the propensity for large and small firms to create jobs was clearly found to vary across sectors o f the economy, countries, and time periods, Birch's results seemed to hold up qualitatively, in that most studies found that new and small firms generate more than their share o f new jobs. Still, there was no consensus. For example, an OECD study
"expresses skepticism about the assumption that small enterprises will be the m otor o f job creation."6 What has become known as the OECD Jobs Study stresses that, "the large claims made for the job creation ability o f small enterprises are often based on faulty statistics...a more correct statement is that small establishments are disproportion
ately responsible for both gross job gains and losses."7
Still, the findings in most o f the job generation studies (Loveman and Sengenberger, 1991) show that when rates o f job creation, job destruction, and net employment changes are calculated by size class, smaller plants and firms are generally found to have the highest rates o f job creation and job destruction. M ost strikingly, while the rates o f net job change, or job creation minus job destruction, tend to be the greatest in small producers and the lowest in large producers, small firms tend to exhibit a positive rate o f net job change, while their larger counterparts tend to exhibit a negative rate o f net job change.
Just when the debate seemed to be converging towards a consensus — that small firms provide the engine o f job creation— Davis, Haltiwanger and Schuh (1996) injected con
siderable doubt by arguing that the findings o f most job generation studies are based on a statistical fallacy. According to Davis, Haltiwanger and Schuh (1996), "Most longitudi
nal studies o f the relationship between employer size and job creation suffer from a sta
tistical pitfall known as the regression fallacy or regression-to-the-mean bias. The poten
tial for bias arises whenever employers experience transitory fluctuations in size, or whenever measurement error introduces transitory fluctuations in observed size. Both phenomena are important features o f longitudinal data on employers." They show that when various corrections are made for regression-to the mean, it is the large firms and not the small enterprises that account for higher rates o f net job change (measured as gross job creation minus gross job destruction).
The regression-to-the-mean objection to the job generation studies was first made by Leonard (1986), who pointed out that if plants and firms have a long-run equilibrium size that is different from their current level, due to a stochastic shock or random shock, they will tend to adjust intertemporally towards their long-run equilibrium level. Large firms will tend to have grown as a result o f such stochastic fluctuations, and small firms will tend to have been adversely influenced by such random shocks. An implication is that employment statistics will tend to identify small producers growing on average and large producers declining.
Based on U.S. Census Bureau data for manufacturing plants between 1972 and 1988, Davis, Haltiwanger and Schuh (1996) find that small firms and plants dominate the crea-
6 "Scepticism on Job Creation Role of Small Business," Financial Times, 20 July, 1994, p. 7.
7 Ibid.
tion and destruction o f jobs in the U.S. manufacturing sector. Smaller manufacturing firms and plants were found to exhibit sharply higher gross rates o f job creation but not higher net rates o f job creation.
However, after correcting for the regression-to-the-mean problem, Baldwin and Picot (1995) and Baldwin (1995) find that the net job creation for smaller establishments in Canadian manufacturing exceeds that o f large establishments. In a related paper, Picot and Dupuy (forthcoming) correct the longitudinal data base for the influence o f the tran
sient component o f employment change and find the same results.
Joachim Wagner (1995) also corrected for the regression-to-the-mean problem in analyz
ing the relationship between firm size and job creation in the German federal state o f Lower Saxony between 1978 and 1993. As in the American and Canadian studies, Wagner found that gross job creation and destruction rates tend to be negatively related to firm size. However, the net rate o f job creation does not appear to be systematically related to the size o f the firm, as measured by average employment in the base and final year. M ost recently, a government study found that between 1977 and 1994 establish
ments with fewer than 20 employees created 1.2 million new jobs, which accounted for two-thirds o f all new jobs created.8
Does creative destruction, as captured by the extent o f market turbulence, lead to higher job creation, in the manner posited by Joseph Schumpeter in 1911? While some prelimi
nary evidence for the United States does, in fact, provide support for the Schumpeterian Thesis (Audretsch, 1995), Audretsch and Fritsch (1996) find no evidence that job crea
tion is associated with a turbulent environment, at least in the case o f West Germany during the late 1980s. In fact, in both the manufacturing and service sectors a high rate o f turbulence in a region tends to lead to a lower and not a higher rate o f growth. This negative relationship is attributable to the fact that the underlying components — the birth and death rates — are both negatively related to subsequent job creation. That is, those regions with higher rates o f new-firm startup activity tend to experience less job creation in subsequent years. M ost strikingly, the same is true for the exit rates. That is, those regions that experience higher exit rates also tend to experience less job creation in sub
sequent years.
Konings (1995) uses the Workplace Industrial Relations Surveys o f 1980, 1984 and 1990 o f plant level employment data to analyze the relationship between establishment size and gross job creation and destruction in the United Kingdom during the 1980s.
Konings finds that the gross job creation rate is higher in small establishments and lower in large establishments. By contrast, the gross job destruction rate is lowest in the small plants and highest in large establishments. Thus, in contrast to the American, Canadian and German studies cited above, Konings finds that, in absolute numbers, the larger plants create most jobs but also destroy the most jobs.
Broersma and Gautier (forthcoming) use longitudinal data from the Netherlands Central Bureau o f Statistics between 1978 and 1991 to identify the link between firm size and job
8 "Studie: Kleinbetriebe schaffen mehr neue Jobs," Die Welt, 15, March, 1996, p. 11.
flows. They find that both job creation and job destruction tend to be negatively related to firm size in the Netherlands. However, the evidence still suggests that small firms contribute the most to net job growth.
2.2 Economic Welfare Implications
Brown, Hamilton and M edoff (1990) draw a number o f economic welfare conclusions regarding the recent shift in job creation away from large corporations and toward smaller enterprises. After documenting the systematically lower wages and poor working conditions associated with smaller enterprises, in terms o f hours worked and the safety and health environments, they conclude that there is a net economic welfare loss associ
ated with such a shift in economic activity. Thus, Brown, Hamilton and M edoff (1990, pp. 88-89) conclude that, "Workers in large firms earn higher wages, and this fact cannot be explained completely by differences in labor quality, industry, working conditions, or union status. Workers in large firms also enjoy better benefits and greater job security than their counterparts in small firms. When these factors are added together, it appears that workers in large firms do have a superior employment package." This observation is anything but new. Nearly two decades ago, Weiss (1979) argued that sub-optimal scale firms represent a net welfare loss in terms o f lower efficiency.9
The positive relationships between wage and non-wage benefits on the one hand, and firm size on the other, have been confirmed to hold across a wide spectrum o f countries.
For example, Oosterbeek and van Praag (1995) found that the positive relationship between firm size and employee compensation holds in the Netherlands. Wagner (forthcoming) similarly finds along with wages, quality o f job training apprenticeships tend to be positively related to firm size in Germany.10
However, the conclusions by Brown, Hamilton and M edoff (1990), Weiss (1991) and others are based on a static analysis. Audretsch, van Leeuwen, Mankveld and Thurik (1995) examine the link between employee compensation and firm size through the dynamic lens provided by a longitudinal base o f Dutch manufacturing workers. They find that firm age exerts a positive impact on productivity and employee compensation.
Apparently as firms mature they become more productive and raise the level o f employee compensation, even after controlling for the size o f the firm. This new finding suggests that not only will some o f the small and sub-optimal firms o f today become the large and optimal firms o f tomorrow, but that there is at least a high tendency for the low produc
tivity and low wage firm o f today to become the high productivity and wage firm o f tomorrow. Thus, the links between firm size and employee compensation may be very different when viewed through a dynamic lens rather than through a static lens.
9 Not only did Weiss (1991) find that the minimum efficient scale (MES) level of output exceeds that of most firms (enterprises) and plants (establishments), but that, "On the average, about half of total shipments in the industries covered are from sub-optimal plants. The majority of plants in most industries are sub-optimal in scale, and a very large percentage of output is from sub-optimal plants in some unconcentrated industries."
io Similar evidence for Germany has been provided by Gerlach and Hiibler (1995).
3 A Theory of Corporate Downsizing
The Labor Department recently reported that as a result o f corporate downsizing, „more than 43 million jobs have been erased in the United States since 1979."ll This includes 24.8 million blue collar jobs and 18.7 million white collar jobs. Between 1980 and 1993, the 500 largest U.S. manufacturing corporations cut 4.7 million jobs, or one quarter o f their work force (Audretsch, 1995). Recent downsizing announcements by U.S. corpo
rations include 123,000 job cuts by AT&T, 122,000 by IBM, and 99,400 by Boeing.
Since 1986 IBM has reduced employment by about 45 percent.12 Perhaps most discon
certing, the rate o f corporate downsizing has apparently increased over time. During most o f the 1980s, about one in 25 workers lost a job. In the 1990s this has risen to one in 20 w orkers.13
Such downsizing has not been at all unique to the United States but has become increas
ingly rampant throughout Europe. Consider the case o f Sweden. Some 70 percent o f Sweden's manufacturing employees work for large companies, most o f them multina
tionals, such as Volvo, which have been constantly shifting production out o f the high- cost Standort, Sweden, and into lower cost countries, through outward foreign direct investment. Between 1970 and 1993 Sweden lost 500,000 private sector jobs, and un
employment is currently 13 percent o f the work force. And Sweden is not an exceptional case. For example, every third car that is manufactured by a German company is actually produced outside o f Germany.14 Similar corporate downsizing has taken place in Germany.15 For example, the German chemical industry is once again profitable and ex
hibiting strong growth. At the same time, the largest firms in the industry continue to downsize and reduce employment. In 1994 employment fell by 4.7 percent to 531,000, and it is predicted that an additional 30,000 jobs will be lost to downsizing.16 Corporate downsizing has not been isolated in the chemical industry. As Newsweek observes, "For the men who run the Siemens Corp., the very heart o f Germany's electronics industry, these are the years o f blood and anguish."17 The reason? By the end o f 1994 Siemens had
12,600 fewer employees than in 1992.18
The theory o f the industry life cycle may be useful in explaining the recent wave o f cor
porate downsizing that has occurred throughout the developed industrialized countries recently. While the theory o f the product life cycle dates back at least to 1950, a more 11 "The Downsizing of America," New York Times, 3 March, 1996, p. 1.
12 "Big Blue's White-Elephant Sale," Business Week, 20 February, 1995, p. 26.
13 "Out One Door and In Another," Business Week, 22 January, 1996, p. 41.
14 Globalisierung: Auslandsproduktion deutscher Autohersteller," Handlesblatt, 31 January, 1994.
15 "Wir Wollen Geld Sehen," Der Spiegel, 20 February 1995, 100-102.
16 "Chemie: Höhere Gewinne, weniger Arbeitsplätze," Die Welt, 21 January, 1995, p. 12.
17 "Lost on the Infobahn: Europe is Losing the Technology Business to U.S. and Japanese Firms,"
Newsweek, 31 October, 1995, 40-45.
18 Similar waves of downsizing have been reported in Japan ("Gentle Downsizing in Japan," Interna
tional Herald Tribune, 24 February, 1996, p. 11).
recent and modem description o f the industry life cycle has been posited by Oliver Williamson (1975): "Three stages in an industry's development are commonly recog
nized: an early exploratory stage, an intermediate development stage, and a mature stage.
The first, or early formative stage involves the supply o f a new product o f relatively primitive design, manufactured on comparatively unspecialized machinery, and marketed through a variety o f exploratory techniques. Volume is typically low. A high degree o f uncertainty characterizes business experience at this stage. The second stage is the inter
mediate development stage in which manufacturing techniques are more refined and market definition is sharpened; output grows rapidly in response to newly recognized applications and unsatisfied demands. A high but somewhat lesser degree o f uncertainty characterizes market outcomes at this stage. The third stage is that o f a mature industry.
Management, manufacturing, and marketing techniques all reach a relatively advanced degree o f refinement. M arkets may continue to grow, but do so at a more regular and predictable rate. Established connections with customers and suppliers (including capital market access) all operate to buffer changes and thereby to limit large shifts in market shares. Significant innovations tend to be fewer and are mainly o f an improvement vari
ety."
What is often overlooked is that the life cycle theory is as much a theory o f the evolution o f technological knowledge and its diffusion as it is about the flows o f exports, imports and foreign direct investment. At the heart o f the evolution o f technological knowledge over an industry's life cycle was the notion that the return accruing to innovative activity from a given investment in creating new technological knowledge diminish as the indus
try evolves. This means that the cost o f innovating relative to the cost o f imitating also diminishes as the industry matures.
An important implication o f the life cycle model is that as an industry matures, the cost o f imitating falls relative to the cost o f innovating, so that it becomes increasingly eco
nomical to transfer that technological knowledge to less costly locations o f production, either through trade or foreign direct investment.
Giersch, Paque and Schmieding (1992) point out that the German Wirtschaftswunder, or economic growth miracle, was fueled to a considerable extent by relatively low labor unit costs and an undervalued currency. Thus, technology developed in the United States could simply be adopted with the end result o f lower unit costs o f production and inter
national competitiveness. But as the European countries caught up to the United States, and the unit cost o f labor began to even surpass that o f the United States in countries such as Germany, simply following a strategy o f technology adoption was no longer suf
ficient to ensure international competitiveness in Europe. The 1994 mean manufacturing employee compensation (including insurance and other employee benefits) was the high
est in Germany, at $25.71 per hour. By contrast, the mean hourly manufacturing wage was just $19.01 in Japan and $16.73 in the United States.
Many o f the industries which have been the traditional strengths in W estern Europe have evolved towards the mature and declining stages o f the life cycle. This means that the high-cost Standort in countries such as Germany becomes increasingly vulnerable, since the production o f rather standardized technologies can be shifted to locations in central
and Eastern Europe, or in Asia, with lower production costs. The daily earnings o f labor (1992) have been estimated to be $78.34 in the European Union, but only $6.14 in Poland, and $6.45 in the Czech Republic, $1.53 in China, $2.46 in India and $1.25 in Sri Lanka.
While labor cost disadvantages can be offset through productivity increases and the substitution o f technology for labor, innovative activity itself tends to become more in
cremental in nature as the industry evolves over the life cycle. Therefore, it becomes in
creasingly difficult for European firms in mature industries to maintain their international competitiveness through innovative activity.
What has been named by the media in Germany as the Innovationskrise and the Standort Problem are really just twin horns o f the same dilemma. The divergence o f real living standards between the developed and developing countries combined with the maturation o f traditional industries in the developed countries dictates that either domestic firms lose market shares to foreign companies which are increasingly able to clone existing tech
nologies and apply them to cheaper location-specific costs, such as wages, or else they maintain their competitiveness through foreign direct investment and by shifting produc
tion to lower-cost foreign locations.
The consequences o f the maturation o f traditional industries in high-cost locations is — as a result o f either a loss in market share or a shift in production out o f the high-cost Standort to lower-cost locations — what has been termed in the press as corporate downsizing. For example, employment by the ten largest companies has generally fallen within Germany between 1984 and 1995. At the same time, employment by these com
panies outside o f Germany has risen, drastically in some cases. Thus, there is consider
able evidence that the largest German companies are shifting jobs outside o f Germany, resulting in a wave o f corporate downsizing within Germany.
As Die Zeit points out in a front page article, "When Profits Lead to Ruin — More Profits and M ore Unemployment: Where is the Social Responsibility o f the Firms?" the German public has generally responded to this corporate downsizing with accusations that corporate Germany is no longer fulfilling its share o f the social contract.19
What is to become o f the dislocated resources, in particular labor? The answer is noth
ing, unless either the price o f those inputs, i.e. wages, fall sufficiently to compete inter
nationally, or they are redeployed in new economic activity that cannot be so costlessly diffused across geographic space.
As an industry matures, technological knowledge also tends to evolve where the content o f that knowledge becomes increasingly based on information and less on tacit knowl
edge. Audretsch and Feldman (1996) argue that the marginal cost o f transmitting infor
mation across geographic space has become virtually zero in an age dominated increas
ingly by e-mail, fax machines and global telecommunications. However, the cost o f
19 "Wenn der Profit zur Pleite führt: Mehr Gewinne — und mehr Arbeitslose: Wo bleibt die soziale Verantwortung der Unternehmer?" Die Zeit, 2 February, 1996, p. 1.
transmitting knowledge and especially tacit knowledge across geographic space is not at all trivial. Thus, Audretsch and Stephan (1996) find that local proximity between uni
versity-based scientists and biotechnology firms matters when the scientist is involved in the transfer o f tacit knowledge. However, when the scientist provides only information to the biotechnology firm geographic proximity does not play an important role.
It is in new and emerging industries where tacit knowledge comprises a relatively large component and information a relatively low component in the production process. Pro
duction in such new industries is less easy diffuse and is therefore compatible with the higher wage levels pervasive throughout the developed countries.
Among the most cited sources o f evidence for the existence o f an Innovations-Krise is the declining share o f patent and R&D vis-ä-vis Japan and the United States exhibited by Europe in General, and Germany in particular. A more careful inspection, however, reveals that the relative technological advantage o f German companies in the industries that Germany has traditionally held a comparative advantage in, such as chemicals, spe
cialized machine tools, and automobiles, has not greatly deteriorated over the last decade.
The continued technological leadership o f German firms in these industries is evidenced by their sound and profitable recoveiy out o f the recession and rapid recovery in global export markets.20 The companies in these industries have not exhibited a loss in either innovative capacity or the capacity to adopt innovations made abroad, at least not within their narrowly defined industries.
The innovation challenge in Germany and the rest o f Europe does not seem to exist in the industries in which Europe has traditionally held a comparative advantage. Rather, the innovation challenge confronting Europe apparently lies in new industries. For example, a recent cover story o f Newsweek was devoted to, "Why Europe is Losing the Technology Race."21 22 As the lead article o f this issue points out, "The problems at Siemens are far from unique. They are, instead, spread throughout much o f Europe's high-tech landscape, and in particular what the Germans like to call "telmatik": the rap
idly converging fields o f computers, telecommunications and televisions...With only a handful o f exceptions, in nearly every segment o f the so-called information-technology industry, there is a rout underway. European competitors are all but invisible."
Similar sentiment can be found in Germany, where D er Spiegel, observed recently that,
"Global structural change has had an impact on the German economy that only a short time ago would have been unimaginable: Many o f the products, such as automobiles, machinery, chemicals and steel are no longer competitive in global markets. A n d in the industries o f the future, like biotechnology and electronics, the German companies are barely participating."72 And the Wall Street Journal recently warned that in Germany, "If 20 "European Economies: Gloom to Boom," The Economist, 24 December 1994, 35-36.
21 "Why Europe is Losing the Technology Race," Newsweek, 31 October 1994.
22 Der Spiegel, number 5, 1994, 82-83. The original text states, "Der weltweite Strukturwandel hat die deutsche Wirtschaft mit einer Wucht getroffen, die noch vor kurzem unvorstellbar schien: Viele
you look at the chip industry, it's a disaster. And the computer industry has been for many years. Energy technology as such is a disaster."23 This echoes the concerns o f one o f Germany's leading politicians, Lothar Spaeth, and the chairman o f the McKinsey &
Co. Germany, Herbert A. Henzler, who argue in their best-selling book, Can the Germans Still Be Saved?, that German's "greatest structural crisis in the postwar period has been the result o f missing the boat on cutting edge technologies."24
Why has it proven so difficult to shift economic activity out o f traditional mature indus
tries and into newly emerging industries? At least some insight is provided by the litera
ture identifying the role that specific technological paradigms play in shaping the nature o f innovative activity. This literature, which spans organizational theory and business history suggests that firm behavior and organizations are shaped by the specific techno
logical environment in which firms are operating (Chandler, 1990). In particular, this literature has generally identified that firm behavior is closely linked to core firm compe
tence. That is, firms are organizations with a specific set o f competencies within a bounded set o f activities. As Nelson and Winter (1982) emphasize, firm core competen
cies typically have a tacit nature and are stored and organized in the routines which guide decision-making. The learning process through which capabilities and routines are devel
oped and shaped is to a large extent local and path dependent.
The concept o f technological paradigms links the technological environment within which the firm has operated to the core competence o f that firm. Innovations that enhance the existing capabilities and routines are generally viewed as falling within the technological paradigm o f the core competence o f the firm. By contrast, innovations that detract and destroy the existing capabilities and routines o f the firm are generally viewed as falling outside o f the boundaries o f the core competence o f the firm.
Whether or not any given firm adopts a new technology will very much depend upon whether that new technology falls within the core competence o f the firm or outside o f the core technological competence. This is because the cost o f adaptation to the organi
zation is considerably lower for competence enhancing innovations than for competence destroying innovations. As Dosi, Pavitt and Soete (1990, p. 45) point out, "Leadership in an old technological paradigm may be an obstacle to a swift diffusion o f the new one, especially owing to the interplay between the constraint posed by the capital stock to readjustment o f productive activities and the behavioral trends in 'old' companies which may embody differential expertise and enjoy high market shares in 'old* technologies."
Archibugi and Pianta (1992) have analyzed patterns o f patenting for specific industries over a broad spectrum o f countries and concluded that, even when considered over a long period o f time, the technological capabilities o f most countries remain remarkably
ihrer Produkte wie Autos und Maschinen, Chemikalien und Stahl sind international nicht mehr wettbewerbsfähig. Und in den Zukunftsindustrien - der Biotechnik etwa oder der Elektronik - sind die Deutschen nur unzureichend vertreten."
23 "Some Germans Fear They're Falling Behind in High-Tech Fields," The Wall Street Journal, 27 April, 1994, p. 1.
Quoted from The Wall Street Journal, 27 April, 1994, p. 1.
24
specialized. M ost countries, especially smaller ones, tend to specialize in technology in just several industries. Within the technological paradigms associated with these indus
tries, new technological developments apparently tend to diffuse fairly rapidly. However, there has been little tendency for countries, including Germany and other European countries, to broaden their technological bases, suggesting that diffusion across techno
logical paradigms is considerably more complicated and costly.
Cohen and Levinthal (1989) argue that firms can influence their ability to adopt new technologies by expanding the boundaries o f core competence, or what they term ab
sorptive capacity. While R&D is generally considered to generate new technological knowledge, Cohen and Levinthal argue that it also serves a dual purpose -- to assimilate and exploit existing knowledge, or to facilitate the adoption o f existing technology. That is, economists (Arrow, 1962) have long observed that firms invest in R&D in order to internalize knowledge which is external to the firm. Cohen and Levinthal (1989, p. 569) similarly argue that, "While R&D obviously generates innovations, it also develops the firm's ability to identify, assimilate, and exploit knowledge from the environment."
Cohen and Levinthal refer to the two faces o f R&D, but there is perhaps also a third face. The nature and direction o f R&D activities undertaken, while serving to expand the absorptive capabilities o f the firm, may also contribute to defining the boundaries and entrenching those boundaries o f the firm's capabilities. Thus, Cohen and Levinthal make explicit reference to the numerous studies identifying that many o f the important innova
tions in new industries come from outside o f the emerging industry. For example, most o f the computer industry's main innovations originated with developments outside o f the industry, particularly in semiconductors. Similar evidence has been found for the alumi
num industry. And an important study found that o f the twenty-five major discoveries introduced into the United States by DuPont, despite the company's reputation for path
breaking research, fifteen originated with work done outside o f the company. Why weren't these innovations pursued by the firms creating the initial technological knowl
edge? Presumably because their (restricted) core competencies did not easily permit adopting new technologies beyond the boundaries o f the firm's technological paradigm.
The innovation challenge confronting Europe resembles those confronting IBM. It is often overlooked that IBM has never lost its technological superiority in its core product -- the mainframe computer. Rather, the prolonged crisis at IBM is attributed to the com
pany's inability to extend this technological superiority beyond the technological para
digm o f the mainframe computer to the emerging technologies involving new industries in related, but distinct fields. To simply conclude that IBM suffers from an innovation crisis is to overlook the fact that in 1992 IBM registered the most patents o f any firm in the world with the Untied States Patent Office, engaged in greater R&D expenditures than at least any other American corporation, and introduced more new innovative prod
ucts than any other firm in the United States. I f IBM suffers from an innovation problem it does not lie in a lack o f innovative activity, but more likely in the areas, or the indus
tries, associated with that innovative activity. That is, IBM has not suffered from a lack o f innovative activity, p e r se, but rather from a lack o f innovative activity in newly emerging industries, which Microsoft and Intel have succeeded in doing.
And so it may be with Europe — the Innovations Krise is not so much a deficiency o f innovative activity but rather an inability to shift the technological paradigm away from the more traditional industries and towards new emerging industries. It is the inability o f the incumbent firms to shift out o f technological paradigms that are maturing and into the newer technological paradigms associated with emerging industries that almost inevitably leads to corporate downsizing (Baily, Bartelsman and Haltiwanger, forthcoming).
4 A Theory o f Entrepreneurship
The starting point for most theories o f innovation is the firm.25 In such theories the firms are exogenous and their performance in generating technological change is endoge
nous.26 For example, in the most prevalent model found in the literature o f technological change, the model o f the knowledge production function, formalized by Zvi Griliches (1979), firms exist exogenously and then engage in the pursuit o f new economic knowl
edge as an input into the process o f generating innovative activity.
The most decisive input in the knowledge production function is new economic knowl
edge. And as Cohen and Klepper (1992) conclude, the greatest source generating new economic knowledge is generally considered to be R&D. Certainly a large body o f em
pirical w ork has found a strong and positive relationship between knowledge inputs, such as R&D, on the one hand, and innovative outputs on the other hand.
Audretsch (1995) proposes shifting the unit o f observation away from exogenously assumed firms to individuals — agents with endowments o f new economic knowledge.
As J. de V. G raaf (1957) observed nearly four decades ago, "When we try to construct a transformation function for society as a whole from those facing the individual firms comprising it, a fundamental difficulty confronts us. There is, from a welfare point o f view, nothing special about the firms actually existing in an economy at a given moment o f time. The firm is in no sense a 'natural unit'. Only the individual members o f the econ
omy can lay claim to that distinction. All are potential entrepreneurs. It seems, therefore, that the natural thing to do is to build up from the transformation function o f men, rather than the firms, constituting an economy. I f we are interested in eventual empirical de
termination, this is extremely inconvenient. But it has conceptual advantages. The ulti
mate repositories o f technological knowledge in any society are the men comprising it, and it is just this knowledge which is effectively summarized in the form o f a transforma
tion function. In itself a firm possesses no knowledge. That which is available to it be
longs to the men associated with it. Its production function is really built up in exactly the same way, and from the same basic ingredients, as society's.“
25 See for reviews of this literature Baldwin and Scott (1987), Cohen and Levin (1989), Scherer (1984 and 1992), andDosi (1988).
26 See for example Scherer (1984 and 1991), Cohen and Klepper (1991, 1992a and 1992b), and Arrow (1962 and 1983).
A large literature has emerged focusing on what has become known as the appropriabil
ity p ro b le m 21 The underlying issue revolves around how firms which invest in the crea
tion o f new economic knowledge can best appropriate the economic returns from that knowledge (Arrow, 1962). But when the lens is shifted away from focusing upon the firm as the relevant unit o f observation to individuals, the relevant question becomes, How can economic agents with a given endowment o f new knowledge best appropriate the returns fro m that knowledge?
The appropriability problem confronting the individual may converge with that confront
ing the firm. Economic agents can and do work for firms, and even if they do not, they can potentially be employed by an incumbent firm. In fact, in a model o f perfect informa
tion with no agency costs, any positive economies o f scale or scope will ensure that the appropriability problems o f the firm and individual converge. If an agent has an idea for doing something different than is currently being practiced by the incumbent enterprises - - both in terms o f a new product or process and in terms o f organization — the idea, which can be termed as an innovation, will be presented to the incumbent enterprise.
Because o f the assumption o f perfect knowledge, both the firm and the agent would agree upon the expected value o f the innovation. But to the degree that any economies o f scale or scope exist, the expected value o f implementing the innovation within the incumbent enterprise will exceed that o f taking the innovation outside o f the incumbent firm to start a new enterprise. Thus, the incumbent firm and the inventor o f the idea would be expected to reach a bargain splitting the value added to the firm contributed by the innovation. The payment to the inventor — either in terms o f a higher wage or some other means o f remuneration — would be bounded between the expected value o f the innovation if it implemented by the incumbent enterprise on the upper end, and by the return that the agent could expect to earn if he used it to launch a new enterprise on the lower end. Or, as Frank Knight (1921, p. 273) observed more than seventy years ago,
"The laborer asks what he thinks the entrepreneur will be able to pay, and in any case will not accept less than he can get from some other entrepreneur, or by turning entre
preneur himself. In the same way the entrepreneur offers to any laborer what he thinks he must in order to secure his services, and in any case not more than he thinks the laborer will actually be worth to him, keeping in mind what he can get by turning laborer him
self."
Thus, each economic agent would choose how to best appropriate the value o f his endowment o f economic knowledge by comparing the wage he would earn if he remains employed by an incumbent enterprise, w, to the expected net present discounted value o f the profits accruing from starting a new firm, k . If these two values are relatively close, the probability that he would choose to appropriate the value o f his knowledge through an external mechanism such as starting a new firm, Pr(e), would be relatively low. On the other hand, as the gap between w and n becomes larger, the likelihood o f an agent choosing to appropriate the value o f his knowledge externally through starting a new enterprise becomes greater, or
27 See Cohen and Levin (1989) and Baldwin and Scott (1987).
Pr(e) = f ( n - w) (1)
As Knight (1921), and later Arrow (1962) emphasized, new economic knowledge is anything but certain. N ot only is new economic knowledge inherently risky, but substan
tial asymmetries exist across agents both between and within firms. (Milgrom and Roberts, 1987). Which is to say that the expected value o f a new idea, or what has been termed here as a potential innovation, is likely to be anything but unanimous between the inventor o f that idea and the decisionmaker, or group o f decisionmakers,28 o f the firm confronting with evaluating proposed changes or innovations. In fact, it is because in
formation is not only imperfect but also asymmetric that Knight (1921, p. 268) argued that the primary task o f the firm is to process information in order to reach a decision:
"With the introduction o f uncertainty — the fact o f ignorance and the necessity o f acting upon opinion rather than knowledge — into this Eden-like situation (that is a world o f perfect information), its character is entirely changed...With uncertainty present doing things, the actual execution o f activity, becomes in a real sense a secondary part o f life;
the primary problem or function is deciding what to do and how to do it."
Alchian (1950) pointed out that the existence o f knowledge asymmetries would result in the inevitability o f mistaken decisions in an uncertain world. Later, Alchian and Demsetz (1972) attributed the existence o f asymmetric information across the employees in a firm as resulting in a problem o f monitoring the contribution accruing from each employee and setting the rewards correspondingly. This led them to conclude that, "The problem o f economic organization is the economical means o f metering productivity and rewards"
(Alchian and Demsetz, 1972, p. 783).
Combined with the bureaucratic organization o f incumbent firms to make a decision, the asymmetry o f knowledge leads to a host o f agency problems, spanning incentive struc
tures, monitoring, and transaction costs. It is the existence o f such agency costs, com
bined with asymmetric information that not only provides an incentive for agents with new ideas to appropriate the expected value o f their knowledge externally by starting new firms, but also with a propensity that varies systematically from industry to industry.
Coase (1937) and later Williamson (1975) argued that the size o f an (incumbent) enter
prise will be determined by answering what Coase (1937, p. 30) articulated as, "The question always is, will it pay to bring an extra exchange transaction under the organizing authority?" In fact, Coase (1937, p. 24) pointed out that, "Other things being equal, a firm will tend to be larger the less likely the (firm) is to make mistakes and the smaller the increase in mistakes with an increase in the transactions organized."
Holmstrom (1989) and Milgrom (1988) have pointed out the existence o f what they term as a bureaucratization dilemma, where, "To say that increased size brings increased
28 For example, as of 1993 a proposal for simply modifying an existing product at IBM had to pass through 250 layers of decisionmaking to gain approval ("Überfordert und Unregierbar," Der Spiegel, No. 14, 1993, p. 127).
bureaucracy is a safe generalization. To note that bureaucracy is viewed as an organiza
tional disease is equally accurate" (Holmstrom, 1989, p. 320).
To minimize agency problems and the cost o f monitoring, bureaucratic hierarchies de
velop objective rules. In addition, Kreps (1991) has argued that such bureaucratic rules promote internal uniformity and that a uniform corporate culture, in turn, promotes the reputation o f the firm. These bureaucratic rules, however, make it more difficult to evaluate the efforts and activities o f agents involved in activities that do not conform to such bureaucratic rules. As Holmstrom (1989, p. 323) points out, "Monitoring limita
tions suggest that the firm seeks out activities which are more easily and objectively evaluated. Assignments will be chosen in a fashion that are conducive to more effective control. Authority and command systems work better in environments which are more predictable and can be directed with less investment information. Routine tasks are the comparative advantage o f a bureaucracy and its activities can be expected to reflect that."
Williamson (1975, p. 201) has also emphasized the inherent tension between hierarchical bureaucratic organizations and the ability o f incumbent organizations to appropriate the value o f new knowledge for innovative activity outside o f the technological trajectories associated with the core competence o f that organization, "Were it that large firms could compensate internal entrepreneurial activity in ways approximating that o f the market, the large firm need experience no disadvantage in entrepreneurial respects. Violating the congruency between hierarchical position and compensation appears to generate bureaucratic strains, however, and is greatly complicated by the problem o f accurately imputing causality." This leads Williamson (1975, pp. 205-206) to conclude that, "I am inclined to regard the early stage innovative disabilities o f large size as serious and pro
pose the following hypothesis: An efficient procedure by which to introduce new prod
ucts is for the initial development and market testing to be performed by independent investors and small firms (perhaps new entrants) in an industry, the successful develop
ments then to be acquired, possibly through licensing or merger, for subsequent market
ing by a large multidivision enterprise...Put differently, a division o f effort between the new product innovation process on the one hand, and the management o f proven resources on the other may well be efficient.“
The degree to which agents and incumbent firms are confronted with knowledge asym
metries and agency problems with respect to seeking out new economic knowledge and (potential) innovative activity would not be expected to be constant across industries.
This is because the underlying knowledge conditions vary from industry to industry. In some industries new economic knowledge generating innovative activity tends to be relatively routine and can be processed within the context o f incumbent hierarchical bureaucracies. In other industries, however, innovations tend to come from knowledge that is not o f a routine nature and therefore tends to be rejected by the hierarchical bureaucracies o f incumbent corporations. Nelson and Winter (1974, 1978 and 1982) described these different underlying knowledge conditions as reflecting two distinct technological regimes — the entrepreneurial and routinized technological regimes: "An entrepreneurial regime is one that is favorable to innovative entry and unfavorable to
innovative activity by established firms; a routinized regime is one in which the condi
tions are the other way around." (Winter, 1984, p. 297).
Gort and Klepper (1982) argued that the relative innovative advantage between newly established enterprises and incumbent firms depends upon the source o f information generating innovative activity. If information based on nontransferable experience in the market is an important input in generating innovative activity, then incumbent firms will tend to have the innovative advantage over new firms. This is consistent with Winter's (1984) notion o f the routinized regime, where the accumulated stock o f nontransferable information is the product o f experience within the market, which firms outside o f the main incumbent organizations, by definition, cannot possess.
By contrast, when information outside o f the routines practiced by the incumbent firms is a relatively important input in generating innovative activity, newly established firms will tend to have the innovative advantage over incumbent firms. Arrow (1962), Mueller (1976), and Williamson (1975) have all emphasized that when such information created outside o f the incumbent firms cannot be easily transferred to those incumbent enter
prises — presumably due to the type o f agency and bureaucracy problems described above — the holder o f such knowledge must enter the industry by starting a new firm in order to exploit the expected value o f his knowledge.
Thus, when the underlying knowledge conditions are better characterized by the routi
nized technological regime, there is likely to be relatively little divergence in the evalua
tion o f the expected value o f a (potential) innovation between the inventor and the deci
sionmaking bureaucracy o f the firm. Under the routinized regime, a great incentive for agents to start their own firms will not exist, at least not for the reason o f doing some
thing differently. When the underlying knowledge conditions more closely adhere to the entrepreneurial technological regime, however, a divergence in beliefs between the agent and the principal regarding the expected value o f a (potential) innovation is more likely to emerge. Therefore, it is under the entrepreneurial regime where the startup o f new firms is likely to play a more important role, presumably as a result o f the motivation to appropriate the value o f economic knowledge. Due to agency problems, this knowledge cannot be easily and costlessly transferred to the incumbent enterprise.
This model analyzing the decision o f how best to appropriate the value o f new economic knowledge confronting an individual economic agent seems useful when considering the actual decision to a new firm taken by entrepreneurs. For example, Chester Carlsson started Xerox after his proposal to produce a (new) copy machine was rejected by Kodak. Kodak based its decision on the premise that the new copy machine would not earn very much money, and in any case, Kodak was in a different line o f business — photography. It is perhaps no small irony that this same entrepreneurial startup, Xerox, decades later turned down a proposal from Steven Jobs to produce and market a per
sonal computer, because they did not think that a personal computer would sell, and, in any case, they were in a different line o f business — copy machines (Carrol, 1993). After seventeen other companies turned down Jobs for virtually identical reasons, including IBM and Hewlett Packard, Jobs resorted to starting his own company, Apple computer.
Similarly, IBM turned down an offer from Bill Gates, "the chance to buy ten percent o f Microsoft for a song in 1986, a missed opportunity that would cost $3 billion today."29 IBM reached its decision on the grounds that "neither Gates nor any o f his band o f thirty some employees had anything approaching the credentials or personal characteristics required to w ork at IBM."30
Divergence in beliefs with respect to the value o f a new idea need not be restricted to what is formally known as a product or even a process innovation. Rather, the fact that economic agents choose to start a new firm due to divergence’s in the expected value o f an idea applies to the sphere o f managerial style and organization as well. One o f the most vivid examples involves Bob Noyce, who founded Intel. Noyce had been employed by Fairchild Semiconductor, which is credited with being the pioneering semiconductor firm. In 1957 Noyce and seven other engineers quit ert masse from Schockley Semicon
ductor to form Fairchild Semiconductor, an enterprise that in turn is considered the start o f what is today known as Silicon Valley. Although Fairchild Semiconductor had
"possibly the most potent management and technical team ever assembled" (Gilder, 1989, p. 89), "Noyce couldn't get Fairchild's eastern owners to accept the idea that stock options should be part o f compensation for all employees, not just for management. He wanted to tie everyone, from janitors to bosses, into the overall success o f the com
pany...This management style still sets the standard for every computer, software, and semiconductor company in the Valley today...Every CEO still wants to think that the place is run the way Bob Noyce would have run it" (Cringley, 1993, p. 39). That is, Noyce's vision o f a firm excluded the dress codes, reserved parking places, closed offices, and executive dining rooms, along with the other trappings o f status that were standard in virtually every hierarchical and bureaucratic U.S. corporation. But when he tried to impress this vision upon the owners o f Fairchild Semiconductor, he was flatly rejected. The formation o f Intel in 1968 was the ultimate result o f the divergence in beliefs about how to organize and manage the firm.
The key development at Intel was the microprocessor. When long time IBM employee Ted H off approached IBM and later DEC with his new microprocessor in the late 1960s,
"IBM and DEC decided there was no market. They could not imagine why anyone would need or want a small computer; if people wanted to sue a computer, they could hood into time-sharing systems" (Palfreman and Swade, 1991, p. 108).
Coase (1937) was awarded a Nobel Prize for explaining why a firm should exist. But why should more than one firm exist in an industry?31 One answer is provided by the traditional economics literature focusing on industrial organization. An excess level o f profitability induces entry into the industry. And this is why the entry o f new firms is
29 "System Error," The Economist, 18 September 1993, p. 99.
30 Paul Carrol, "Die Offene Schlacht," Die Zeit, No. 39, 24 September 1993, p. 18.
31 Coase (1937, p. 23) himself asked, "A pertinent question to ask would appear to be (quite apart from the monopoly considerations raised by Professor Knight), why, if by organizing one can eliminate certain costs and in fact reduce the cost of production, are there any market transactions at all? Why is not all production carried on by one big firm?"
