Research Issues Relating to Structure, Competition, and Performance of Small Technology-Based Firms

Research Issues Relating to Structure, Competition, and Performance of Small Technology-Based Firms

David B. Audretsch

Small Business Economics

16: 37–51, 2001. 2001 Kluwer Academic Publishers. Printed in the Netherlands.

ABSTRACT. The role that small firms play in industrialorganization has evolved considerably since the second worldwar. This paper seeks to document how and why smallbusiness plays a very different role in industrial organizationresearch today than it did some three decades ago.

1. Introduction

The most important development in the field ofindustrial organization is a shift in the frameworkfor analyzing small business. While there werealways a few studies around analyzing smallbusiness through a dynamic lens, a much moreprofound and comprehensive shift in the literaturebegan in the later 1980s and early 1990s. The mostsalient characteristic feature of this literature wasthe introduction of a dynamic or evolutionaryframework.

In the second section of this paper, the view ofsmall business emerging from the static frame-work of the post-war industrial organization isdescribed. This view identifies small business asbeing sub-optimal in terms of scale of production.The resulting impact on performance is negative,in terms of productivity and wages. The policyimplication is that, at least in terms of economicefficiency, small business exerts a drag oneconomic welfare.

In the third section of the paper, the view ofsmall business emerging from the dynamic frame-work of the last decade is introduced. This viewprovides a striking contrast to the role and con-

tribution of small business. When viewed throughthe lens of a dynamic framework, small businessesare seen as an agent of change. The empiricalevidence supporting this dynamic view of smallbusiness is presented in the fourth section. Theimplications for public policy towards business arepresented in the fifth section. In particular, we findthat the role of public policy towards business hasshifted from constraining the power of largecorporations to enabling the creation and com-mercialization of knowledge, particularly bysmall business. A summary and conclusions areprovided in the final section.

2. Small business in industrial organization: 2. the static view

In linking entrepreneurship to economic growth,the obvious starting point is the theory of the firm.The field of economics that focuses the most onlinks between the organization of firms in indus-tries and the resulting economic performance hasbeen industrial organization. The ascendancy ofindustrial organization in the post-war period asan important and valued field economics camefrom the recognition not only by scholars but alsoby policy makers that industrial organizationmatters. The widespread fear vis-à-vis the SovietUnion pervasive throughout the United States atthe end of the 1950s and early 1960s was not justthat the Soviets might bury the Americans becausethey were the first into space with the launchingof the Sputnik, but that the superior organizationof industry facilitated by centralized planning wasgenerating greater rates of growth in the SovietUnion. After all, the nations of Eastern Europe,and the Soviet Union in particular, had a “luxury”inherent in their systems of centralized planning -

Final version accepted on September 15, 2000

School of Public and Environmental AffairsIndiana UniversityBloomington, IN 47405-2100U.S.A.E-mail: [email protected]

a concentration of economic assets on a scalebeyond anything imaginable in the West, wherethe commitment to democracy seemingly imposeda concomitant commitment to economic decen-tralization.

Although there may have been considerabledebate about what to do about the perceived Sovietthreat some three decades ago, there was littledoubt at that time that the manner in whichenterprises and entire industries were organizedmattered. And even more striking, when onereviews the literature of the day, there seemed tobe near unanimity about the way in which indus-trial organization mattered. It is no doubt an ironyof history that a remarkably similar version of thegiantism embedded in Soviet doctrine, fueled bythe writings of Marx and ultimately implementedby the iron fist of Stalin, was also prevalentthroughout the West. This was the era of massproduction when economies of scale seemed tobe the decisive factor in dictating efficiency. Thiswas the world the world so colorfully describedby John Kenneth Galbraith (1956) in his theoryof counterveiling power, in which the power of bigbusiness was held in check by big labor and bybig government.

It became the task of the industrial organizationscholars to sort out the issues involving this per-ceived trade-off between economic efficiency onthe one hand and political and economic decen-tralization on the other. The scholars of industrialorganization responded by producing a massiveliterature focusing on essentially three issues:(i)What are the economic gains to size and large-scale production? (ii) What are the economicwelfare implications of having an oligopolisticmarket structure, i.e. is economic performancepromoted or reduced in an industry with just ahandful of large-scale firms? and (iii) Given theoverwhelming evidence from (ii) that large-scaleproduction resulting in economic concentration isassociated with increased efficiency, what are thepublic policy implications?

A fundamental characteristic of the industrialorganization literature was not only that it wasobsessed with the oligopoly question but that itwas essentially static in nature. There was con-siderable concern about what to do about theexisting firms and industrial structure, but littleattention was paid to where they came from and

where they were going. Oliver Williamson’sclassic 1968 article in the American EconomicReview, “Economies as an Antitrust Defense: TheWelfare Tradeoffs,” became something of a finalstatement demonstrating this seemingly inevitabletradeoff between the gains in productive efficiencythat could be obtained through increased concen-tration and gains in terms of competition thatcould be achieved through decentralizingeconomic policies, such as antitrust. But it did notseem possible to have both, certainly not inWilliamson’s completely static model.

One of the most striking findings emerging inthis static view of industrial organization is thatsmall firms generally operate at a level of outputthat is too small to sufficiently exhaust scaleeconomies, even when the standard definition ofa small firm employing fewer than 500 employeesis applied. A large number of studies found thatbecause the minimum efficient scale (MES) ofoutput, or the lowest level of output where theminimum average cost is attained, large-scale pro-duction is typically required to exhaust scaleeconomies in manufacturing. Any enterprise orestablishment that was smaller than required bythe MES was branded as being suboptimal orinefficient, in that it produced at average costs inexcess of more efficient larger firms. Weiss(Audretsch and Yamawaki, 1991, p. 403) assumedthat “The term ‘suboptimal’ capacity describes acondition in which some plants are too small to beefficient”.

The importance of scale economies in thetypical manufacturing industry relegated mostsmall firms to being classified as suboptimal. Forexample, Weiss (1964) found that suboptimalplants accounted for about 52.8 percent of industryvalue-off-shipments, Scherer (1973) found that58.2 percent of value-of-shipments emanated fromthe suboptimal plants in twelve industries, andPratten (1971) identified the suboptimal scaleestablishments accounting for 47.9 percent ofindustry shipments. After reviewing the literatureon the extent of suboptimal firms, Weiss (inAudretsch and Yamawaki, 1991, p. xiv) concludedthat, “In most industries the great majority of firmsis suboptimal. In a typical industry there are, let’ssay, one hundred firms. Typically only about fiveto ten of them will be operating at the MES levelof output, or anything like it.”

38 David B. Audretsch

What are the economic welfare implications?Weiss (1979, p. 1137) argued that the existenceof small firms which are sub-optimal representeda loss in economic efficiency and thereforeadvocated any public policy which “. . . createssocial gains in the form of less sub-optimalcapacity.” This actually translated into aningenuious argument against market power, sinceempirical evidence suggested that the priceumbrella provided by monopoly power encour-aged the existence of suboptimal capacity firms.Weiss (1979) went so far as to argue that thelargest inefficiency associated with market powerwas not the higher prices charged to consumersbut rather that it facilitated the existence of sub-optimal scale small firms.

Wages and productivity would be expected toreflect the degree to which small firms are lessefficient than their larger counterparts. There is alarge body of empirical evidence spanning a broadrange of samples, time periods and even countriesthat has consistently found wages (and non-wagecompensation as well) to be positively related tofirm size. Probably the most cited study is that ofBrown, Hamilton and Medoff (1990, pp. 88–89),who conclude that, “Workers in large firms earnhigher wages, and this fact cannot be explainedcompletely by differences in labor quality industry,working conditions, or union status. Workers inlarge firms enjoy better benefits and greatersecurity than their counterparts in small firms.When these factors are added together, it appearsthat workers in large firms do have a superioremployment package.”

Seen through the static lens provided throughtraditional industrial organization and labor eco-nomics the economic welfare implications of therecent shift in economic activity away from largefirms and towards small enterprises is unequivocal– overall economic welfare is decreased sinceproductivity and wages will be lower in small thanin large firms. As Weiss (1979) argued in termsof efficiency and Brown, Hamilton and Medoff(1990) in terms of employee compensation, theimplication for public policy is to implementpolicies to shift economic activity away fromsmall firms and towards larger enterprises.

3. Small firms in industrial organization: 2. the dynamic view

Coase (1937) was awarded a Nobel Prize forexplaining why a firm should exist. But whyshould more than one firm exist in an industry?One answer is provided by the traditional eco-nomics literature focusing on industrial organiza-tion. An excess level of profitability induces entryinto the industry. And this is why the entry of newfirms is interesting and important – because thenew firms provide an equilibrating function in themarket, in that the levels of price and profit arerestored to the competitive levels. The new firmsare about business as usual – they simply equili-brate the market by providing more of it.

An alternative explanation for the entry of newfirms was provided for by Audretsch (1995), whosuggests that new firms are not founded to besmaller clones of the larger incumbents but ratherto serve as agents of change through innovativeactivity.

The starting most for most theories of innova-tion is the firm. In such theories the firms areexogenous and their performance in generatingtechnological change is endogenous. For example,in the most prevalent model found in the literatureof technological change, the model of the knowl-edge production function, formalized by ZviGriliches (1979), firms exist exogenously and thenengage in the pursuit of new economic knowledgeas an input into the process of generating innov-ative activity.

The most decisive input in the knowledge pro-duction function is new economic knowledge. Andas Cohen and Klepper (1991 and 1992) conclude,the greatest source generating new economicknowledge is generally considered to be R&D.Certainly a large body of empirical work hasfound a strong and positive relationship betweenknowledge inputs, such as R&D, on the one hand,and innovative outputs on the other hand.

The knowledge production function has beenfound to hold most strongly at broader levels ofaggregation. The most innovative countries arethose with the greatest investments to R&D. Littleinnovative output is associated with less developedcountries, which are characterized by a paucityof production of new economic knowledge.Similarly, the most innovative industries, also tend

Research Issues of Small Technology-Based Firms 39

to be characterized by considerable investments inR&D and new economic knowledge. Not only areindustries such as computers, pharmaceuticals andinstruments high in R&D inputs that generate neweconomic knowledge, but also in terms of innov-ative outputs (Audretsch, 1995). By contrast,industries with little R&D, such as wood products,textiles and paper, also tend to produce only anegligible amount of innovative output. Thus, theknowledge production model linking knowledgegenerating inputs to outputs certainly holds at themore aggregated levels of economic activity.

Where the relationship becomes less com-pelling is at the disaggregated microeconomiclevel of the enterprise, establishment, or even lineof business. For example, While Acs andAudretsch (1990) found that the simple correla-tion between R&D inputs and innovative outputwas 0.84 for four-digit standard industrial classi-fication (SIC) manufacturing industries in theUnited States, it was only about half, 0.40 amongthe largest U.S. corporations.

The model of the knowledge productionfunction becomes even less compelling in viewof the recent wave of studies revealing that smallenterprises serve as the engine of innovativeactivity in certain industries. These results arestartling, because as Scherer (1991) observes, thebulk of industrial R&D is undertaken in the largestcorporations; small enterprises account only for aminor share of R&D inputs. Thus the knowledgeproduction function seemingly implies that, as theSchumpeterian Hypothesis predicts, innovativeactivity favors those organizations with access toknowledge-producing inputs – the large incumbentorganization. The more recent evidence identi-fying the strong innovative activity raises thequestion, “Where do new and small firms get theinnovation producing inputs, that is the knowl-edge?”

One answer, proposed by Audretsch (1995), isthat, although the model of the knowledge pro-duction function may still be valid, the implicitlyassumed unit of observation – at the level of thefirm – may be less valid. The reason why theknowledge production function holds more closelyfor more aggregated degrees of observation maybe that investment in R&D and other sources ofnew knowledge spills over for economic exploita-tion by third-party firms.

A large literature has emerged focusing on whathas become known as the appropriability problem.The underlying issue revolves around how firmswhich invest in the creation of new economicknowledge can best appropriate the economicreturns from that knowledge (Arrow, 1962).Audretsch (1995) proposes shifting the unit ofobservation away from exogenously assumedfirms to individuals – agents with endowments ofnew economic knowledge. As J. de V. Graafobserved nearly four decades ago, “When we tryto construct a transformation function for societyas a whole from those facing the individual firmscomprising it, a fundamental difficulty confrontsus. There is, from a welfare point of view, nothingspecial about the firms actually existing in aneconomy at a given moment of time. The firm isin no sense a ‘natural unit’. Only the individualmembers of the economy can lay claim to thatdistinction. All are potential entrepreneurs. Itseems, therefore, that the natural thing to do is tobuild up from the transformation function of men,rather than the firms, constituting an economy. Ifwe are interested in eventual empirical determi-nation, this is extremely inconvenient. But it hasconceptual advantages. The ultimate repositoriesof technological knowledge in any society are themen comprising it, and it is just this knowledgewhich is effectively summarized in the form of atransformation function. In itself a firm possessesno knowledge. That which is available to itbelongs to the men associated with it. Its produc-tion function is really built up in exactly the sameway, and from the same basic ingredients, associety’s.”

But when the lens is shifted away from focusingupon the firm as the relevant unit of observationto individuals, the relevant question becomes, Howcan economic agents with a given endowment ofnew knowledge best appropriate the returns fromthat knowledge?

The appropriability problem confronting theindividual may converge with that confronting thefirm. Economic agents can and do work for firms,and even if they do not, they can potentially beemployed by an incumbent firm. In fact, in amodel of perfect information with no agency costs,any positive economies of scale or scope willensure that the appropriability problems of thefirm and individual converge. If an agent has an


idea for doing something different than is currentlybeing practiced by the incumbent enterprises –both in terms of a new product or process and interms of organization – the idea, which can betermed as an innovation, will be presented to theincumbent enterprise. Because of the assumptionof perfect knowledge, both the firm and the agentwould agree upon the expected value of the inno-vation. But to the degree that any economies ofscale or scope exist, the expected value of imple-menting the innovation within the incumbententerprise will exceed that of taking the innova-tion outside of the incumbent firm to start a newenterprise. Thus, the incumbent firm and theinventor of the idea would be expected to reacha bargain splitting the value added to the firmcontributed by the innovation. The payment to theinventor – either in terms of a higher wage orsome other means of remuneration – would bebounded between the expected value of the inno-vation if it implemented by the incumbent enter-prise on the upper end, and by the return that theagent could expect to earn if he used it to launcha new enterprise on the lower end. Or, as FrankKnight (1921, p. 273) observed more than seventyyears ago, “The laborer asks what he thinks theentrepreneur will be able to pay, and in any casewill not accept less than he can get from someother entrepreneur, or by turning entrepreneurhimself. In the same way the entrepreneur offersto any laborer what he thinks he must in order tosecure his services, and in any case not more thanhe thinks the laborer will actually be worth to him,keeping in mind what he can get by turning laborerhimself.”

Thus, each economic agent would choose howto best appropriate the value of his endowment ofeconomic knowledge by comparing the wage hewould earn if he remains employed by an incum-bent enterprise, w, to the expected net present dis-counted value of the profits accruing from startinga new firm,

π. If these two values are relativelyclose, the probability that he would choose toappropriate the value of his knowledge through anexternal mechanism such as starting a new firm,Pr(e), would be relatively low. On the other hand,as the gap between w and π becomes larger, thelikelihood of an agent choosing to appropriate thevalue of his knowledge externally through startinga new enterprise becomes greater, or

Pr(e) = f(π – w). (1)

This model analyzing the decision of how bestto appropriate the value of new economic knowl-edge confronting an individual economic agentseems useful when considering the actual decisionto a new firm taken by entrepreneurs. Forexample, Chester Carlsson started Xerox after hisproposal to produce a (new) copy machine wasrejected by Kodak. Kodak based its decision onthe premise that the new copy machine would notearn very much money, and in any case, Kodakwas in a different line of business – photography.It is perhaps no small irony that this same entre-preneurial startup, Xerox, decades later turneddown a proposal from Steven Jobs to produce andmarket a personal computer, because they did notthink that a personal computer would sell, and, inany case, they were in a different line of business– copy machines (Carrol, 1993). After seventeenother companies turned down Jobs for virtuallyidentical reasons, including IBM and HewlettPackard, Jobs resorted to starting his owncompany, Apple computer.

Similarly, IBM turned down an offer from BillGates, “the chance to buy ten percent of Microsoftfor a song in 1986, a missed opportunity thatwould cost $3 billion today.” IBM reached itsdecision on the grounds that “neither Gates norany of his band of thirty some employees hadanything approaching the credentials or personalcharacteristics required to work at IBM.”

Divergences in beliefs with respect to the valueof a new idea need not be restricted to what isformally known as a product or even a processinnovation. Rather, the fact that economic agentschoose to start a new firm due to divergences inthe expected value of an idea applies to the sphereof managerial style and organization as well. Oneof the most vivid examples involves Bob Noyce,who founded Intel. Noyce had been employed byFairchild Semiconductor, which is credited withbeing the pioneering semiconductor firm. In 1957Noyce and seven other engineers quit en massefrom Schockley Semiconductor to form FairchildSemiconductor, an enterprise that in turn isconsidered the start of what is today known asSilicon Valley. Although Fairchild Semiconductorhad “possibly the most potent management andtechnical team ever assembled” (Gilder, 1989,


p. 89), “Noyce couldn’t get Fairchild’s easternowners to accept the idea that stock options shouldbe part of compensation for all employees, not justfor management. He wanted to tie everyone, fromjanitors to bosses, into the overall success of thecompany . . . This management style still sets thestandard for every computer, software, and semi-conductor company in the Valley today . . . EveryCEO still wants to think that the place is run theway Bob Noyce would have run it” (Cringley,1993, p. 39). That is, Noyce’s vision of a firmexcluded the dress codes, reserved parking places,closed offices, and executive dining rooms, alongwith the other trappings of status that werestandard in virtually every hierarchical andbureaucratic U.S. corporation. But when he triedto impress this vision upon the owners of FairchildSemiconductor, he was flatly rejected. The for-mation of Intel in 1968 was the ultimate result ofthe divergence in beliefs about how to organizeand manage the firm.

The key development at Intel was the micro-processor. When long time IBM employee TedHoff approached IBM and later DEC with his newmicroprocessor in the late 1960s, “IBM and DECdecided there was no market. They could notimagine why anyone would need or want a smallcomputer; if people wanted to usecomputer, theycould hook into time-sharing systems” (Palfremanand Swade, 1991, p. 108).

The model proposed by Audretsch (1995)refocuses the unit of observation away from firmsdeciding whether to increase their output from alevel of zero to some positive amount in a newindustry, to individual agents in possession of newknowledge that, due to uncertainty, may or maynot have some positive economic value. It is theuncertainty inherent in new economic knowledge,combined with asymmetries between the agentpossessing that knowledge and the decisionmaking vertical hierarchy of the incumbent orga-nization with respect to its expected value thatpotentially leads to a gap between the valuation ofthat knowledge.

How the economic agent chooses to appropriatethe value of his knowledge, that is either withinan incumbent firm or by starting or joining a newenterprise will be shaped by the knowledgeconditions underlying the industry. Under theroutinized technological regime the agent will tend

to appropriate the value of his new ideas withinthe boundaries of incumbent firms. Thus, thepropensity for new firms to be started should berelatively low in industries characterized by theroutinized technological regime.

By contrast, under the entrepreneurial regimethe agent will tend to appropriate the value of hisnew ideas outside of the boundaries of incumbentfirms by starting a new enterprise. Thus, thepropensity for new firms to enter should be rela-tively high in industries characterized by theentrepreneurial regime.

Audretsch (1995) suggests that divergences inthe expected value regarding new knowledge will,under certain conditions, lead an agent to exercisewhat Albert O. Hirschman (1970) has termed asexit rather than voice, and depart from an incum-bent enterprise to launch a new firm. But who isright, the departing agents or those agentsremaining in the organizational decision makinghierarchy who, by assigning the new idea a rela-tively low value, have effectively driven the agentwith the potential innovation away? Ex post theanswer may not be too difficult. But given theuncertainty inherent in new knowledge, the answeris anything but trivial a priori.

Thus, when a new firm is launched, itsprospects are shrouded in uncertainty. If the newfirm is built around a new idea, i.e., potential inno-vation, it is uncertain whether there is sufficientdemand for the new idea or if some competitorwill have the same or even a superior idea. Evenif the new firm is formed to be an exact replicaof a successful incumbent enterprise, it is uncer-tain whether sufficient demand for a new clone, oreven for the existing incumbent, will prevail in thefuture. Tastes can change, and new ideas emergingfrom other firms will certainty influence thosetastes.

Finally, an additional layer of uncertaintypervades a new enterprise. It is not known howcompetent the new firm really is, in terms of man-agement, organization, and workforce. At leastincumbent enterprises know something about theirunderlying competencies from past experience.Which is to say that a new enterprise is burdenedwith uncertainty as to whether it can produce andmarket the intended product as well as sell it. Inboth cases the degree of uncertainty will typicallyexceed that confronting incumbent enterprises.


This initial condition of not just uncertainty, butgreater degree of uncertainty vis-à-vis incumbententerprises in the industry is captured in the theoryof firm selection and industry evolution proposedby Boyan Jovanovic (1982). Jovanovic presents amodel in which the new firms, which he termsentrepreneurs, face costs that are not only randombut also differ across firms. A central feature ofthe model is that a new firm does not know whatits cost function is, that is its relative efficiency,but rather discovers this through the process oflearning from its actual post-entry performance. Inparticular, Jovanovic (1982) assumes that entre-preneurs are unsure about their ability to managea new-firm startup and therefore their prospectsfor success. Although entrepreneurs may launcha new firm based on a vague sense of expectedpost-entry performance, they only discover theirtrue ability – in terms of managerial competenceand of having based the firm on an idea that isviable on the market – once their business is estab-lished. Those entrepreneurs who discover thattheir ability exceeds their expectations expand thescale of their business, whereas those discoveringthat their post-entry performance is less thancommensurate with their expectations will contactthe scale of output and possibly exit from theindustry. Thus, Jovanovic’s model is a theory ofnoisy selection, where efficient firms grow andsurvive and inefficient firms decline and fail.

The role of learning in the selection process hasbeen the subject of considerable debate. On theone hand is what has been referred to as theLarackian assumption that learning refers toadaptations made by the new enterprise. In thissense, those new firms that are the most flexibleand adaptable will be the most successful inadjusting to whatever the demands of the marketare. As Nelson and Winter (1982, p. 11) point out,“Many kinds of organizations commit resources tolearning; organizations seek to copy the forms oftheir most successful competitors.”

On the other hand is the interpretation that therole of learning is restricted to discovering if thefirm has the right stuff in terms of the goods it isproducing as well as the way they are beingproduced. Under this interpretation the new enter-prise is not necessarily able to adapt or adjust tomarket conditions, but receives information basedon its market performance with respect to its

fitness in terms of meeting demand most effi-ciently vis-à-vis rivals. The theory of organiza-tional ecology proposed by Michael T. Hannanand John Freeman (1989) most pointedly adheresto the notion that, “We assume that individualorganizations are characterized by relative inertiain structure.” That is, firms learn not in the sensethat they adjust their actions as reflected by theirfundamental identity and purpose, but in the senseof their perception. What is then learned iswhether or not the firm has the right stuff, but nothow to change that stuff.

The theory of firm selection is particularlyappealing in view of the rather startling size ofmost new firms. For example, the mean size ofmore than 11,000 new-firm startups in the manu-facturing sector in the United States was found tobe fewer than eight workers per firm (Audretsch,1995). While the minimum efficient scale (MES)varies substantially across industries, and even tosome degree across various product classes withinany given industry, the observed size of most newfirms is sufficiently small to ensure that the bulkof new firms will be operating at a suboptimalscale of output. Why would an entrepreneur starta new firm that would immediately be confrontedby scale disadvantages?

An implication of the theory of firm selectionis that new firms may begin at a small, evensuboptimal, scale of output, and then if merited bysubsequent performance expand. Those new firmsthat are successful will grow, whereas those thatare not successful will remain small and mayultimately be forced to exit from the industry ifthey are operating at a suboptimal scale of output.

Subsequent to entering an industry, an entre-preneur must decide whether to maintain its outputexpand, contract, or exit. Two different strands ofliterature have identified several major influencesshaping the decision to exit an industry. The first,and most obvious strand of literature suggests thatthe probability of a business exiting will tend toincrease as the gap between its level of output andthe minimum efficient scale (MES) level of outputincreases. The second strand of literature points tothe role that the technological environment playsin shaping the decision to exit. As Dosi (1988)and Arrow (1962) argue, an environment charac-terized by more frequent innovation may also beassociated with a greater amount of uncertainty


regarding not only the technical nature of theproduct but also the demand for that product. Astechnological uncertainty increases, particularlyunder the entrepreneurial regime, the likelihoodthat the business will be able to produce aviable product and ultimately be able to survivedecreases.

An important implication of the dynamicprocess of firm selection and industry evolutionis that new firms are more likely to be operatingat a suboptimal scale of output if the underlyingtechnological conditions are such that there is agreater chance of making an innovation, that isunder the entrepreneurial regime. If new firmssuccessfully learn and adapt, or are just plainlucky, they grow into viably sized enterprises. Ifnot, they stagnate and may ultimately exit fromthe industry. This suggests, that entry and thestartup of new firms may not be greatly deterredin the presence of scale economies. As long asentrepreneurs perceive that there is some prospectfor growth and ultimately survival, such entry willoccur. Thus, in industries where the MES is high,it follows from the observed general small size ofnew-firm startups that the growth rate of the sur-viving firms would presumably be relatively high.

At the same time, those new firms not able togrow and attain the MES level of output wouldpresumably be forced to exit from the industry,resulting in a relatively low likelihood of survival.In industries characterized by a low MES, neitherthe need for growth, nor the consequences of itsabsence are as severe, so that relatively lowergrowth rates but higher survival rates would beexpected. Similarly, in industries where the prob-ability of innovating is greater, more entrepreneursmay actually take a chance that they will succeedby growing into a viably sized enterprise. In suchindustries, one would expect that the growth ofsuccessful enterprises would be greater, but thatthe likelihood of survival would be correspond-ingly lower.

4. The empirical evidence

Not only was the large corporation thought to havesuperior productive efficiency, but conventionalwisdom also held the large corporation to serveas the engine of technological change andinnovative activity. After all, Schumpeter (1942,

p. 106), “What we have got to accept is that thelarge-scale establishment has come to be the mostpowerful engine of progress.” A few years later,John Kenneth Galbraith (1956, p. 86) echoedSchumpeter’s sentiment when he lamented, “Thereis no more pleasant fiction than that technolog-ical change is the product of the matchless inge-nuity of the small man forced by competition toemploy his wits to better his neighbor. Unhappily,it is a fiction.”

Knowledge regarding both the determinants andthe impact of technological change has beenlargely shaped by measurement. Measures of tech-nological change have typically involved one ofthe three major aspects of the innovative process:(1) a measure of inputs into the process, such asR&D expenditures, or the share of the labor forceaccounted for by employees involved in R&Dactivities; (2) an intermediate output, such as thenumber of inventions that have been patented; or(3) a direct measure of innovative output.

The earliest sources of data, R&D measured,indicated that virtually all of the innovativeactivity was undertaken by large corporations. Aspatent measures became available, the generalqualitative conclusions did not change, althoughit became clear that small firms were moreinvolved with patent activity than with R&D. Thedevelopment of direct measures of innovativeactivity, such as data bases measuring new productand process introductions in the market, indicatedsomething quite different. In a series of studies,Acs and Audretsch (1987, 1988, 1990) found thatwhile large firms in manufacturing introduced aslightly greater number of significant new inno-vations than small firms, small-firm employmentwas only about half as great as large-firm employ-ment, yielding an average small-firm innovationrate in manufacturing of 0.309, compared to alarge-firm innovation rate of 0.202. The relativeinnovative advantage of small and large firms wasfound to vary considerably across industries. Insome industries, such as computers and processcontrol instruments, small firms provide theengine of innovative activity. In other industries,such as pharmaceutical products and aircraft, largefirms generate most of the innovative activity.Knowledge regarding both the determinants andthe impact of technological change has beenlargely shaped by measurement.


Acs and Audretsch (1988, 1990) concludedthat some industries are more conducive to small-firm innovation while others foster the innovativeactivity of large corporations corresponds to thenotion of distinct technological regimes – theroutinized and entrepreneurial technologicalregimes.

Empirical evidence in support of the traditionalmodel of entry, which focuses on the role ofexcess profits as the major incentive to enter,has been ambiguous at best, leading Geroski(1991, p. 282) to conclude, “Right from thestart, scholars have had some trouble in recon-ciling the stories told about entry in standardtextbooks with the substance of what they havefound in their data. Very few have emerged fromtheir work feeling that they have answered halfas many questions as they have raised, much lessthat they have answered most of the interestingones.”

Perhaps one reason for this trouble is the inher-ently static model used to capture an inherentlydynamic process. Manfred Neumann (1993, pp.593–594) has criticized this traditional model ofentry, as found in the individual country studiescontained in Geroski and Schwalbach (1991),because they “are predicated on the adoption of abasically static framework. It is assumed thatstartups enter a given market where they are facingincumbents which naturally try to fend off entry.Since the impact of entry on the performance ofincumbents seems to be only slight, the questionarises whether the costs of entry are worthwhile,given the high rate of exit associated with entry.Geroski appears to be rather skeptical about that.I submit that adopting a static framework ismisleading . . . In fact, generally, an entrant canonly hope to succeed if he employs either a newtechnology or offers a new product, or both. Justimitating incumbents is almost certainly doomedto failure. If the process of entry is looked uponfrom this perspective the high correlation betweengross entry and exit reflects the inherent risksof innovating activities . . . Obviously it israther difficult to break loose from the inheritedmode of reasoning within the static framework.It is not without merit, to be sure, but it needsto be enlarged by putting it into a dynamicsetting.”

Still, one of the most startling results that have

emerged in empirical studies is that entry by firmsinto an industry is apparently not substantiallydeterred or even deterred at all in capital-intensiveindustries in which scale economies play animportant role (Audretsch, 1995). While studieshave generally produced considerable ambiguityconcerning the impact of scale economies andother measures traditionally thought to representa barrier to entry, Audretsch (1995) found con-clusive evidence linking the technological regimeto startup activity. New-firm startup activity tendsto be substantially more prevalent under the entre-preneurial regime, or where small enterprisesaccount for the bulk of the innovative activity,than under the routinized regime, or where thelarge incumbent enterprises account for most ofthe innovative activity. These findings are consis-tent with the view that differences in beliefs aboutthe expected value of new ideas are not constantacross industries but rather depend on the knowl-edge conditions inherent in the underlying tech-nological regime.

Geroski (1995) and Audretsch (1995) point outthat one of the major conclusions from studiesabout entry is that the process of entry does notend with entry itself. Rather, it what happens tonew firms subsequent to entering that sheds con-siderable light on industry dynamics. The earlystudies (Mansfield, 1962; Hall, 1987; Dunne,Roberts and Samuelson, 1989; Audretsch, 1991)established not only that the likelihood of a newentrant surviving is quite low, but that the likeli-hood of survival is positively related to firm sizeand age. More recently, a wave of studies haveconfirmed these findings for diverse countries,including Portugal (Mata et al., 1994; Mata, 1994),Germany (Wagner, 1994), and Canada (Baldwinand Gorecki, 1991; Baldwin, 1995; Baldwin andRafiquzzaman, 1995).

Audretsch (1991), Audretsch and Mahmood(1995) shifted the relevant question away fromWhy does the likelihood of survival vary system-atically across firms? to Why does the propensityfor firms to survive vary systematically acrossindustries? The answer to this question suggeststhat what had previously been considered to posea barrier to entry may, in fact, constitute not anentry barrier but rather a barrier to survival. Theanswer to this questions suggests that what hadpreviously been considered to pose a barrier to


entry may, in fact, constitute not an entry barrierbut rather a barrier to survival.

What has become known as Gibrat’s Law, orthe assumption that growth rates are invariant tofirm size, has been subject to numerous empiricaltests. Studies linking firm size and age to growthhave also produced a number of stylized facts(Wagner, 1992). For small and new firms thereis substantial evidence suggesting that growthis negatively related to firm size and age(Hall, 1987; Wagner, 1992, 1994; Mata, 1993;Audretsch, 1995). However, for larger firms,particularly those having attained the minimumefficient scale (MES) level of output, the evidencesuggests that firm growth is unrelated to size andage.

An important finding of Audretsch (1991,1995) and Audretsch and Mahmood (1995) is thatalthough entry may still occur in industries char-acterized by a high degree of scale economies, thelikelihood of survival is considerably less. Peoplewill start new firms in an attempt to appropriatethe expected value of their new ideas, or poten-tial innovations, particularly under the entrepre-neurial regime. As entrepreneurs gain experiencein the market they learn in at least two ways. First,they discover whether they possess the right stuff,in terms of producing goods and offering servicesfor which sufficient demand exists, as well aswhether they can product that good more effi-ciently than their rivals. Second, they learnwhether they can adapt to market conditions aswell as to strategies engaged in by rival firms. Interms of the first type of learning, entrepreneurswho discover that they have a viable firm will tendto expand and ultimately survive. But what aboutthose entrepreneurs who discover that they areeither not efficient or not offering a product forwhich their is a viable demand? The answer is, Itdepends – on the extent of scale economies as wellas on conditions of demand. The consequences ofnot being able to grow will depend, to a largedegree, on the extent of scale economies. Thus,in markets with only negligible scale economies,firms have a considerably greater likelihood ofsurvival. However, where scale economies play animportant role the consequences of not growingare substantially more severe, as evidenced by alower likelihood of survival.

What emerges from the new evolutionary

theories and empirical evidence on the economicrole of new and small firms is that markets are inmotion, with a lot of new firms entering theindustry and a lot of firms exiting out of theindustry. But is this motion horizontal, in that thebulk of firms exiting are comprised of firms thathad entered relatively recently, or vertical, in thata significant share of the exiting firms had beenestablished incumbents that were displaced byyounger firms? In trying to shed some light on thisquestion, Audretsch (1995) proposes two differentmodels of the evolutionary process of industriesover time. Some industries can be best character-ized by the model of the conical revolving door,where new businesses enter, but where there is ahigh propensity to subsequently exit from themarket. Other industries may be better character-ized by the metaphor of the forest, where incum-bent establishments are displaced by new entrants.Which view is more applicable apparently dependson three major factors – the underlying techno-logical conditions, scale economies, and demand.Where scale economies play an important role, themodel of the revolving door seems to be moreapplicable. While the rather starting result dis-cussed above that the startup and entry of newbusinesses is apparently not deterred by thepresence of high scale economies, a process offirm selection analogous to a revolving doorensures that only those establishments successfulenough to grow will be able to survive beyondmore than a few years. Thus the bulk of newentrants that are not so successful ultimately exitwithin a few years subsequent to entry.

There is at least some evidence also suggestingthat the underlying technological regime influ-ences the process of firm selection and thereforethe type of firm with a higher propensity to exit.Under the entrepreneurial regime new entrantshave a greater likelihood of making an innovation.Thus, they are less likely to decide to exit fromthe industry, even in the face of negative profits.By contrast, under the routinized regime theincumbent businesses tend to have the innovativeadvantage, so that a higher portion of exitingbusinesses tend to be new entrants. Thus, themodel of the revolving door is more applicableunder technological conditions consistent with theroutinized regime, and the metaphor of the forest,where the new entrants displace the incumbents


– is more applicable to the entrepreneurialregime.

Why is the general shape of the firm-size dis-tribution not only strikingly similar across virtu-ally every industry – that is, skewed with only afew large enterprises and numerous small ones –but has persisted with tenacity not only acrossdeveloped countries but even over a long periodof time? The evolutionary view of the process ofindustry evolution is that new firms typically startat a very small scale of output. They are motivatedby the desire to appropriate the expected value ofnew economic knowledge. But, depending uponthe extent of scale economies in the industry, thefirm may not be able to remain viable indefinitelyat its startup size. Rather, if scale economies areanything other than negligible, the new firm islikely to have to grow to survival. The temporarysurvival of new firms is presumably supportedthrough the deployment of a strategy of compen-sating factor differentials that enables the firm todiscover whether or not it has a viable product.

The empirical evidence supports such anevolutionary view of the role of new firms inmanufacturing, because the post-entry growth offirms that survive tends to be spurred by the extentto which there is a gap between the MES level ofoutput and the size of the firm. However, thelikelihood of any particular new firm survivingtends to decrease as this gap increases. Such newsuboptimal scale firms are apparently engaged inthe selection process. Only those firms offering aviable product that can be produced efficiently willgrow and ultimately approach or attain the MESlevel of output. The remainder will stagnate, anddepending upon the severity of the other selec-tion mechanism – the extent of scale economies –may ultimately be forced to exit out of theindustry. Thus, the persistence of an asymmetricfirm-size distribution biased towards small-scaleenterprise reflects the continuing process of theentry of new firms into industries and not neces-sarily the permanence of such small and sub-optimal enterprises over the long run. Althoughthe skewed size distribution of firms persists withremarkable stability over long periods of time, aconstant set of small and suboptimal scale firmsdoes not appear to be responsible for this skeweddistribution. Rather, by serving as agents ofchange, new firms provide an essential source of

new ideas and experimentation that otherwisewould remain untapped in the economy.

5. The public policy response

The policy response to this new view of theknowledge production function has been to shiftaway from targeting outputs to inputs. In partic-ular, this involves the creation and commercial-ization of knowledge. Examples include thepromotion of joint R&D programs, education andtraining programs, and policies to encouragepeople to start new firms. As Saxenian (1985,p. 102) points out, “Attracting high-tech hasbecome the only development game of the 1980s.”Justman (1995) and Justman and Teubel (1986)show how investment in infrastructure provide animportant source of growth.

The provision of venture and informal capitalto facilitate the creation and growth of new firmshas replaced concern about the market power ofexisting ones in policy debates (Hughes, 1997;Mason and Harrison, 1997). The lack of financecapital for new ventures has been blamed for theinability of Germany and France to shift economicactivity into new industries that generate high-wage employment. One of the most repeatedphrases on the pages of the business news over thelast few years has been “Put Bill Gates in Europeand it just wouldn’t have worked out.”1

Policy efforts to address the most pressingcontemporary economic problems have focused onenablement rather than constraint. Emphasis onenabling firms and individuals to create andcommercialize new knowledge is not restricted toany single country or set of countries. Laura Tyson(1994), former chair of the Council of EconomicAdvisors in the Clinton Administration, recentlyemphasized the importance of government policiesto promote entrepreneurship and new-firm startupsin the former Soviet Union. Similarly, as unem-ployment in Germany surpassed four million, andstood at nearly eleven percent of the labor force,it is not surprising the Chancellor Helmut Kohlwould undertake action to spur the creation of newjobs. What is more surprising is the main emphasisannounced by the Chancellor in the Initiatives forInvestment and Employment2 in 1996 on new andsmall firms. The first and main point of theChancellor’s Program consists of a commitment


to the “creation of new innovative firms.”3 Therationale underlying this policy approach by theChancellor is stated in the Program: “New jobs arecreated mainly in new firms and in small- andmedium-sized enterprises.”4

Audretsch and Feldman (1996) argue thatindustrial policies targeting the production andcommercialization of new economic knowledgewill have a greater impact on particular regionsand not diffuse rapidly across geographic space.They point out that knowledge spillovers are a keysource of new knowledge generating innovativeactivity, but due to the tacit nature of thatknowledge, knowledge flows tend to be geo-graphically bounded. Although the cost of trans-mitting information has become invariant todistance, the cost of transmitting knowledge, andespecially tacit knowledge, rises with distance. Bycreating regions of knowledge-based economicactivities, government policies can generate highlyconcentrated innovative clusters.

As long as the major policy issue wasrestricting large, oligopolistic firms in commandof considerable market power, a federal or nationallocus of control was appropriate. This is becausethe benefits and costs derived from that marketpower are asymmetric between the local regionwhere the firm is located and the national market,where the firm sells its product. Not only was pro-duction concentrated in one or just several regions,but the workers along with the ancillary suppliersalso tended to be located in the same regions.These workers as well as the community at largeshare the fruits accruing from monopoly power.Systematic empirical evidence (Weiss, 1966)shows that wages are positively related to thedegree of market power held by a firm, even aftercontrolling for the degree of unionization. Higherprofits resulting from market power are shared bylabor. Workers and firms in the region have thesame interest.

As Olson (1982) shows, relatively small coali-tions of economic agents benefiting from somecollective action tend to prevail over a large groupof dispersed economic agents each incurring asmall cost from tat action. The costs or organizingand influencing policy are relatively low for thesmall coalition enjoying the benefits but largefor the group of dispersed economic agents.Government policies to control large oligopolistic

firms with substantial market power were notlikely to be successful if implemented on the locallevel. Rather, as Olson (1982) predicts, a regionallocus of policy towards business tends to resultin the capture of policy by the coalition of localinterests benefiting from that policy. Only byshifting the locus of policy away from the regionto the national level can the capture of policy byspecial interest groups be minimized. This isbecause the negative effects of market power inthe form of higher prices are spread throughoutthe national market while the benefits accruingfrom that power are locally concentrated.

Starting in the Carter Administration in the late1970s and continuing into the Administrations ofpresidents Reagan, Bush and Clinton, antitrust hasbeen de-emphasized and a twenty year wave ofderegulation has led to a downsizing and evenclosure of a number of the former regulatoryagencies.

Many economists interpret the downsizing ofthe federal agencies charged with the regulationof business as the eclipse of government inter-vention. But to interpret the retreat of the federalgovernment as the end of public intervention isto confuse the downsizing of government with ashifting of the locus of government policy awayfrom the federal to the local level. The last decadehas seen the emergence of a set of enabling policyinitiatives at the local level. This new type ofindustrial policy is decentralized and regional innature. As Sternberg (1996) emphasizes in hisreview of successful technology policies in thefour leading technological countries, the mostimportant industrial policies in the last decadeshave been local not national. They have occurredin locations such as Research Triangle (Link,1995), Austin, Texas and Cambridge (U.K.).Sternberg (1996) shows how the success of anumber of different high-technology clustersspanning the four most technologically advancedcountries is the direct result of enabling policiesundertaken at the regional level.

Eisnger asks the question, “Do American StatesDo Industrial Policy?” in a 1990 article publishedin the British Journal of Political Science. Loweryand Gray (1990) confirm Eisinger’s affirmativeanswer by analyzing the impact of state industrialpolicy in the United States. They develop a newdata set on gross state product and a new measure


of state industrial policy activism. Their resultssuggest that the implementation of industrialpolicy at the state level tends to promote growth.For example, Feller (1997, p. 289) points out that“in theory and implementation, state technologydevelopment programs – as in Texas, Ohio, NewYork, New Jersey, and Pennsylvania – may beviewed as bands on a wide spectrum from basicresearch to product development, with the endsreflecting quite divergent state strategies . . .” TheAdvanced Research Program in Texas hasprovided support for basic research and thestrengthening of the university infrastructure,which played a central role in recruiting MCC andSematech and developing a high-tech clusteraround Austin. The Thomas Edison Centers inOhio, the Advanced Technology Centers in NewJersey, and the Centers for Advanced Technologyat Case Western Reserve University, RutgersUniversity and the University of Rochester havesupported generic, precompetitive research. Thissupport has generally provided diversified tech-nology development involving a mix of activitiesencompassing generic research, applied research,and manufacturing modernization through a broadspectrum of industrial collaborators spanningtechnology-intensive multinational corporations,regional manufactures and new-firm startups.

This shift in the locus of policy is the result oftwo factors. First, because the source of compar-ative advantage is knowledge, which tends to belocalized in regional clusters, public policy requiresan understanding of region-specific characteristicsand idiosyncrasies. As Sternberg (1996) con-cludes, regional strengths provide the major sourceof innovative clusters. The second factor is thatthe motivation underlying government policy isnow growth and the creation of (high-paying) jobs,largely through the creation of new firms. Thesenew firms are typically small and pose no oligop-olistic threat in national or international markets.There are no external costs imposed on consumersin the national economy in the form of higherprices as in the case of a large oligopolisticcorporation in possession of market power. Thereis no reason that the promotion of local economiesimposes a cost on consumers in the nationaleconomy, so that localized industrial policy isjustified and does not result in any particular lossincurred by agents outside of the region.

6. Conclusions

While traditional theories suggest that entrepre-neurship will retard economic growth, newtheories suggest exactly the opposite – thatentrepreneurship will stimulate and generategrowth. The reason for these theoretical discrep-ancies lies in the context of the underlying theory.In the traditional theory, new knowledge plays norole; rather, static efficiency, determined largelyby the ability to exhaust scale economies dictatesgrowth. By contrast, the new theories are dynamicin nature and emphasize the role that knowledgeplays. Because knowledge is inherently uncertain,asymmetric and associated with high costs oftransactions, divergences emerge concerning theexpected value of new ideas. Economic agentstherefore have an incentive to leave an incumbentfirm and start a new firm in an attempt to com-mercialize the perceived value of their knowledge.Entrepreneurship is the vehicle by which (the mostradical) new ideas are sometimes implemented.

While this policy emphasis on small and newfirms as engines of dynamic efficiency may seemstartling after decades at looking to the corporategiants to bestow efficiency, it is anything but new.Before the country was even half a century old,Alexis de Tocqueville, in 1935, reported, “Whatastonishes me in the United States is not so muchthe marvellous grandeur of some undertakings asthe innumerable multitude of small ones.”

Notes

* An earlier version of this paper was presented at the SBAOffice of Advocacy Conference “The Invisible Part of theIceberg: Research Issues in Industrial Organization and SmallBusinesses”, January 21, 2000, Washington, DC. I am gratefulto the suggestions of Fred Tarpley, Ken Simonson and otherconference participants.1 “Where’s the Venture Capital?” Newsweek, 31 October,1994, p. 44. Similar sentiment was expressed by JoschkaFischer, parliamentary leader of the Green Party in Germany,who laments, “A company like Microsoft would never havea chance in Germany” (“Those German Banks and TheirIndustrial Treasures,” The Economist, 21 January, 1994,77–78).2 This was announced as the Akrionsprogramm fuerInvestitionen und Arbeitsplaetze (“Soziale Einschnitte undSteuerreform sollen Wirtschaftswachstum anregen:Bundesregierung beschliesst Aktionsprogramm fuerInvestiitionen und Arbeitsplaetze,” Der Tagesspiegel, 31January, p. 1).3 The original text of the Aktionsprogramm states, “Offensive


für unternehmerische Selbständigkeit und innovations-fähigkeit” (“Ein Kraftakt zu Rettung des StandortsDeutschland,” Frankfurter Allgemeine, 31 January, 1996,p. 11).4 “Ein Kraftakt zu Rettung des Standorts Deutschland,”Frankfurter Allgemeine, 31 January, 1996, p. 11. The originaltext reads, “Neue Arbeitsplätze entstehen zumeist in neuge-gründeten Unternehmen und im Mittelstand.”

References

Acs, Zoltan J. and David B. Audretsch, 1988, ‘Innovation inLarge and Small Firms: An Empirical Analysis’, AmericanEconomic Review 78(4), 678–690.

Acs, Zoltan J. and David B. Audretsch, 1990, Innovation andSmall Firms, Cambridge: MIT Press.

Arrow, Kenneth J., 1962, ‘Economic Welfare and theAllocation of Resources for Invention’, in R. R. Nelson(ed.), The Rate and Direction of Inventive Activity,Princeton: Princeton University Press.

Audretsch, David B., 1991, ‘New Firm Survival and theTechnological Regime’, Review of Economics andStatistics 73(3), 441–450.

Audretsch, David B., 1995, Innovation and Industry Evolution,Cambridge: MIT Press.

Audretsch, David B. and Talat Mahmood, 1995, ‘New-FirmSurvival: New Results Using a Hazard Function’, Reviewof Economics and Statistics 77(1), 97–103.

Audretsch, David B. and Maryann P. Feldman, 1996, ‘R&DSpillovers and the Geography of Innovation andProduction’, American Economic Review 86(3), 630–640.

Audretsch, David B. and Paula E. Stephan, 1996, ‘Company-Scientist Locational Links: The Case of Biotechnology’,American Economic Review 86(3), 641–652.

Audretsch, David B. and Roy Thruik, 1999, Innovation,Industry Evolution, and Employment, Cambridge:Cambridge University Press.

Baldwin, John R., 1995, The Dynamics of IndustrialCompetition, Cambridge: Cambridge University Press.

Baldwin, John R. and Paul K. Gorecki, 1991, ‘Entry, Exit, andProduction Growth’, in P. Geroski and J. Schwalbach(eds.), Entry and Market Contestability: An InternationalComparison, Oxford: Basil Blackwell.

Baldwin, John R. and M. Rafiquzzaman, 1995, ‘Selectionversus Evolutionary Adaptation: Learning and Post-EntryPerformance’, International Journal of IndustrialOrganization 13(4), 501–523.

Brown, Chalres, James Hamilton and James Medoff, 1990,Employers Large and Small, Cambridge: CambridgeUniversity Press.

Carrol, Paul, 1993, Big Blues: The Unmaking of IBM, NewYork: Crown Publishers.

Caves, Richard E., 1998, ‘Industrial Organization and NewFindings on the Turnover and Mobility of Firms’, Journalof Economic Literature 36(4), 1947–1982.

Chandler, Alfred D. Jr., 1977, The Visible Hand: TheManagerial Revolution in American Business, Cambridge:Harvard University Press.

Coase, R. H., 1937, ‘The Nature of the Firm’, Economica 4(4),386–405.

Cringley, Robert X., 1993, Accidental Empires, New York:Harper Busienss.

Dunne, T., M. J. Roberts and L. Samuelson, 1989, ‘TheGrowth and Failure of U.S. Manufacturing Plants’,Quarterly Journal of Economics 104, 671–698.

Feller, Irwin, 1997, ‘Federal and State Government Roles inScience and Technology’, Economic DevelopmentQuarterly 11(4), 283–296.

Galbraith, John Kenneth, 1956, American Capitalism, Boston:Houghton Mifflin.

Geroski, Paul A., 1991, ‘Some Data-Driven Reflections on theEntry Process’, in Paul Georski and Joachin Schwalbach(eds.), Entry and Market Contestability: An InternationalComparison, Oxford: Basil Blackwell.

Geroski, Paul A., 1995, ‘What Do We Know About Entry’,International Journal of Industrial Organization (SpecialIssue on The Post Entry Performance of Firms, D. B.Audretsch and J. Mata, eds.), 13(4).

Geroski, Paul A. and Joachim Schwalbach (eds.), 1991, Entryand Market Contestability: An International Comparison,Oxford: Basil Blackwell.

Griliches, Zvi, 1979, ‘Issues in Assessing the Contribution ofR&D to Productivity Growth’, Bell Journal of Economics10, 92–116.

Hall, Bronwyn H., 1987, ‘The Relationship Between Firm Sizeand Firm Growth in the U.S. Manufacturing Sector’,Journal of Industrial Economics 35, 583–605.

Hannan, Michael T. and John Freeman, 1989, OrganizationalEcology, Cambridge, MA: Harvard University Press.

Hirschman, Albert O., 1970, Exit, Voice, and Loyalty,Cambridge: Harvard University Press.

Hughes, Alan, 1997, ‘Finance for SMEs: A U.K. Perspective’,Small Business Economics 9(2), 151–166.

Ijiri, Yuji and Herbert A. Simon, 1977, Skew Distributions andSizes of Business Firms, Amsterdam: North Holland.

Jovanovic, Boyan, 1982, ‘Selection and Evolution ofIndustry’, Econometrica 50(2), 649–670.

Justman, M., 1995, ‘Infrastructure, Growth and the TwoDimensions of Industrial Policy’, Review of EconomicStudies 62(1), 131–157.

Justman, M. and M. Teubal, 1986, ‘Innovation Policy in anOpen Economy: A Normative Framework to Strategic andTactical Issues’, Research Policy 15, 121–138.

Knight, Frank H., 1921, Risk, Uncertainty and Profit, NewYork: Houghton Mifflin.

Link Al, 1995, A Generosity of Spirit, Durham, NC: DukeUniversity Press.

Mansfield, Edwin, 1962, ‘Entry, Gibrat’s Law, Innovation, andthe Growth of Firms’, American Economic Review 52(5),1023–1051.

Mata, Jose, 1994, ‘Firm Growth During Infancy’, SmallBusiness Economics 6(1), 27–40.

Mata, Jose and Pedro Portugal, 1994, ‘Life Duration of newFirms’, Journal of Industrial Economics 27(3), 227–246.

Mata, Jose, Pedro Portugal and Paulo Guimaraes, 1995, ‘TheSurvival of New Plants: Start-Up Conditions and Post-Entry Evolution’, International Journal of IndustrialOrganization 13(4), 459–482.


Moore, John H., 1992, ‘Measuring Soviet Economic Growth:Old Problems and New Complications’, Journal ofInstitutional and Theoretical Economics 148(1), 72–92.

Nelson, Richard R. and Sidney G. Winter, 1982, AnEvolutionary Theory of Economic Change, Cambridge:Harvard University Press.

Neumann, Manfred, 1993, ‘Review of Entry and MarketContestability: An International Comparison’, InternationalJournal of Industrial Organization 11(4), 593–594.

Palfreman, Jon and Doron Swade, 1991, The Dream Machine:Exploring the Computer Age, London: BBC Books.

Pratten, C. F., 1971, Economies of Scale in ManufacturingIndustry, Cambridge: Cambridge University Press.

Saxenian, Annalee, 1990, ‘Regional Networks and theResurgence of Silicon Valley’, California ManagementReview 33(1), 89–111.

Scherer, F. M., 1976, ‘Industrial Structure, Scale Economies,and Worker Alienation’, in Robert T. Masson and P. DavidQualls (eds.), Essays on Industrial Organization in Honorof Joe S. Bain, Cambridge: Ballinger, pp. 105–122.

Scherer, F. M., 1991, ‘Changing Perspectives on the Firm SizeProblem’, in Z. J. Acs and D. B. Audretsch (eds.),Innovation and Technological Change. An InternationalComparison, Ann Arbor: University of Michigan Press.

Schumpeter, Joseph A., 1911, Theorie der wirtschaftlichenEntwicklung. Eine Untersuchung über Unternehmer-gewinn, Kapital, Kredit, Zins und den Konjunkturzyklus,Berlin: Duncker und Humblot.

Schumpeter, Joseph A., 1942, Capitalism, Socialism andDemocracy, New York: Harper and Row.

Simon, Herbert A. and Charles P. Bonini, 1958, ‘The SizeDistribution of Business Firms’, American EconomicReview 48(4), 607–617.

Sternberg, Rolf, 1996, ‘Technology Policies and the Growthof Regions’, Small Business Economics 8(2), 75-86.

Sutton, John, 1997, ‘Gibrat’s Legacy’, Journal of EconomicLiterature 35(1), 40–59.

Tyson, Laura d’Andrea, Tea Petrin and Halsey Rogers, 1994,‘Promoting Entrepreneurship in Eastern Europe’, SmallBusiness Economics 6(3), 165–184.

Wagner, Joachim, 1992, ‘Firm Size, Firm Growth, andPersistence of Chance: Testing Gibrat’s Law withEstablishment Data from Lower Saxony, 1978–1989’,Small Business Economics 4(2), 125–131.

Wagner, Joachim, 1994, ‘Small Firm Entry in ManufacturingIndustries: Lower Saxony, 1979–1989’, Small BusinessEconomics 6(3), 211–224.

Weiss, Leonard W., 1979, ‘The Structure-PerformanceParadigm and Antitrust’, University of Pennsylvania LawReview 127, 1104–1140.

Williamson, Oliver E., 1968, ‘Economies as an AntitrustDefense: The Welfare Tradeoffs’, American EconomicReview 58(1), 18–36.

Williamson, Oliver E., 1975, Markets and Hierarchies:Antitrust Analysis and Implications, New York: The FreePress.

Winter, Sidney G., 1984, ‘Schumpeterian Competition inAlternative Technological Regimes’, Journal of EconomicBehavior and Organization 5, 287–320.


Documents

Research Issues Relating to Structure, Competition, and Performance of Small Technology-Based Firms