Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
Economic democracy wikipedia , lookup
Production for use wikipedia , lookup
Ragnar Nurkse's balanced growth theory wikipedia , lookup
Business cycle wikipedia , lookup
Steady-state economy wikipedia , lookup
Chinese economic reform wikipedia , lookup
Protectionism wikipedia , lookup
Uneven and combined development wikipedia , lookup
ECONOMICS OF INNO" ATION: BlACK HOIJE OI~ I)OSl'flVJi: SUM? Ralph Young CSIRO Curporate Phllll1ing Office P~)per prepared for prcscnlatiol1 to 37th Annual (~onrercnce Ausralian Agricultural Economics Societ)' University uf Sydu(')' Februnry 1993 ECONOMICS ~)F INNOVATION: BLACK HOLE OR POSITIVE SUM? Ralph Young CSIRO Corporate Planning Office tfA civilisation which ca1lnot burst through its abstractions is doomed to :.;terility after a very limited period of progress" A.M. Whitehead l In recent decades, the dominant theory of macroeconomic growth has been the neoclassical model based on the work of Solow (1956, 1957) ,md Swan (1956). Technological change is acknowledged ·'lS the crucial driving force behind economic.. growth, but is treated as an exogenous vari.able. In consequence this -basic neoclassical model is quite limited in its ability to explain economic growth. 2 Sheehan (1992) suggests this constitutes "something of an admission of defeat for economic analysis". TIle view that technical change is endogenous rather than exogenous to economic growth is not new. Schumpeter J042 and Arrow 1962 hoth emphasised investment in innovation as central to economic growth. Schumpeter stressed the importance of economic incentives for entrepreneurs to introduce innovation, and the diffusion process whilst Arrow focussed on the spillover effects of increased knowledge due to "leming oy doing". The work of Schumpeter and Arrow fonn the acknowledged antecedents for lht' new economic growth models. Disatisfaction with· the haf)ic neoclassica.l model was also a stimulus for the evolutionary the'ory of innovatiun and growth which built on dw seminal work of Nelson and Winter 1974. 111is body of literature treats technological change and innovation as endogenous to the process of economic growth and seeks 10 eliminate the gap between the basic neoclassical model mld the empirical evidence of the microecononllc hterature pertaining to the economics of innovation (nasi 1988; Dosi et a/1988; Freeman 1990). Similarly the new economic growth models have grown out of dissatification with the perfoIDlance of the neoclassical model in explaining and predicting economic growth. l1lis work hat) been pioneered by Rom ~r (1986). Features of these new growth theories include endogenous teclmical change, investment in human capital andlor research and innovation as key sources of growth, and increasing returns. Despite the correspondence of their origins and the overlap of their subject matter, there is surprisingly little or no cross referencing between the economics of innovation literature and that of the new economic growth theories. Nevertheless, the l1lpidly expanding new econom.ic growth literature has been hailed by Sheehan 1992 as bringing "greater realism to growth theory" and being a contributor to "a fundamental 2 Quoted in Clark 1985. For recent reviews of the basic neoclassical model and its ltmitntions see BIE 1992, Lucas 1988 and Sheehml 1992. 2 change in. the way economists; and ultimately govemmentandbusiness leaders in Anglo-Saxon countries think about basic issues of growth and trade and the way they develop 'policies for industl)' and technology". TIle aim of this paper is to consider the processes of innovation spedfied futhe new t growth models and their policy implications, to assess these innovation processes in the context of the innovation economics literature and to look at possible future directions for research. The Processes of Innovation in the New Growtb Models The literature on the new growth theories has been extensively reviewed p!cently by BIE~ 1992, Sheehan 1992 and Verspagen 1992, I do not propose to replicate their work~ but will draw on it to focus on the processes of innovation specified in the new growtll theories. A feature of the new growth models is that innovation and growth are endogen0us. nlis iInplies that these olodels face the problem of accommodating increasing returns. Followmg Sheehan 1992, two basic approaches have been adopted. • treat increasing returns as extemaJities; • treat jncreasing returns as monopoJi<;tic competition. TIlese are Imp0r1ant and novd features of the new growth models and are associated with the public good aspect: of technjc~il knowledge and the appropriability of innovation Ie the private ~ood a~pect~. The different approaches to incorporating extemabilities ,md increasing returns to scale are summarised in tabular [oml by Verspagen 1992 and are shown in TahJe 1. In the early papers hy Romer (1986) and Lucas (J988)~ competitive markets are assumed and increasing returns are based on externalities in production associated with knowledge accumul~ilion (learning by doing) mId hmmm capital accumulation via increasing ski111evels. Romer 1990 acknowledges t.hat the leanling by doing formulation ha~ the advantage of making the rate of accumulation of nonrival knowledge endogenous but "is unsatisfaclOfY because it takes the strict proportionality between knowledge and physical capital or knowledge ,md education as an unexplained and exogenously given feature of the technology". This fannulation thus ignores the appropriability issue and "rules out the possibility t.hat finns ma.1ce intentional investments in research and development". Later Inodcl~ building on the seminal efforts of Romer 1986 and Lucas 1988 have attempted to address this issue. Most notable are the papers by Romer 1990 and Grossman and Hclpman 1989, 1990, 1991. It is on these that we focus our attention, 3 In introducing monopoly powe.r, twoaItemative approaches are adopted ... one involves innovative products whlchadd to thee:dsting variety; the other involves inuovation products which are .ofhlgher quality (Sheehan 1992). Romer 1990 specifies three premises to justify his introduction of monopoly power. • • teclmological change - improvement in tbe instructions for mixing together raw materials - lies at the heart of economic growth • technological change arises in large part because of intentional actions taken by people who respond to market incentives ie endogenous technological cbange • instructions for working with raw materials are inherently different from otber econorruc goods. Once the (fIXed) cost of creating a new set of instructions has been incurred, the instructions can be repeatedly llsed without additional cost. To incorporate these premises in his model, Romer specifies two components of knowledge: human capItal, H, which is flval, and teclmological knowledge, A, which is non-rival and can grow without bound; three sectors : a research sector which produces designs for new durables, an intennediate goods sector which uses the designs to produce durable goods and a final goods sector which uses theintennediate goods to produce final consunler goods; and four inputs : capital (K), labour (L), human capital (It) and an index of the level of tedmology (A). K and L are fIXed. In tbe research sect.or, the accumulation in the stock of designs evolves according to: where: 8 .- a constant HA thl' quantity of human capital employed in research ie HA + Hy ~ Ii and By = the amount of H and devoted to production. = On this basis, the productivity of research vades linearly with the growth in technological knowledge and the marginal productivity of HA continues to grow in proportlO11 to A. According to Romer, the cru.cial feature of this specification is that "knowledge enters into production in two distinct way", A new design enables the production of a new good Ulat can be used to produce output. A new design also increase.') the total stock of knowledge and thereby increases the productivity of u human capital in the research sector • 'nle intennediate good\) sector is monopolistic with one finn i for each durable good j t using a fixed number of units of foregone consumption n to produce each unit of durable good. 'I11e initial expenditure on the new design is a sunk cost, and hence the monopolistic firm chooses the level of output (x) to maximise its revenue minus variable cost at every date. The flow of rental income is p(x) times x and it turns uut that the monopoly price is a simple mark up over marginal cost. 4 In addition there are two non-convexitiesinthemodel.' One ism fmal out}?ut "" the existence of monopolistic c;ompetitioltin .theintennediate sectorrneansthat the price of durable goods to the consumer goods sector will 'be higher than incompetition~ while in theresearchsectof, investment in l~arcb will ignore the impact ofthe:;todt of designs, A, on the 'productivity of the research sector (Sheehan 1992). Romer also shows that in the equilibrium steady state, growth occurs ata rate equal to the rate of endogenous innovation, a rate which is less than the social (command economy) solution, implying a role forgovenunent to support the accumulation ofA. ParatleIing the analysis of growth ba.~ed on increasing variety of products, has been the ancuysis of quality improvement in products. The key contributors are Aghi.on and Howitt 1990 and Grossman 2nd Helpman 1991. A major feature underlying the model developed by Grossman and Helpman is the concept of a qu"liry ladder of a fu(-d range of consumer goods. The highestq!1ality product enables the producer t~ capture monopoly profits, R&D activities achieve quality increments and hence with each new innovation builrlingon the previous one, the productivity/quality of the intennediate/consumer goods is always higher for the next innovation. 111e research teclmology is similar to the product variety case with technological spilluvers to all films from prior research and the capturing of monopoly profits by quality leaders. luere is also a negalive externality in the model - labelled "creative destruction" Of "business stealing". it arises because a new innovator destroys the monopoly rents for the previous iIUlovator. In these quality models. technological advances ;rre essemjalJy stochastic, with research succes~ bein:' dependent on the runount of human capital. As a result, this specificmion cnptures the uncenainty \\'hich characterises the innovation proces~ lU tht! real world (Verspagen 1992). In summary the achievement of endogenous technological cbangeis based on the assumption of a distinction between appropriable and non-appropriable effects in the production of itmovation~ which in tU01 is required to provide incentives to produce innovation in the presence of extemabilittes associated with the non-appropriable effects of innovation. To. achieve appropriabiJity some degree of monopoly power is a">sumed. Policy Implications Sheehan 1992~ following his review of the new growth models, idcnlifies four main policy related themes which are "often quite at variance with received doctrine It. For a discussion ot the reJevance of nonconvexlties for growth~ se~ Romer 1990a Romer concludes that ttnonrival goods exist, that they are important for 3 aggregate growth, and thrt they create nonconvexities that matter for aggregate level analysis", 5 These are: • Suboptimal market solution - the nlaiket growth rate is typically less tbanthe social optimum, but in the case of modelsincolpOratmgnegative externalities (Aghion and Howitt 1990, and Grossman and Helpmaa1991),ilie direction·of the, net effect is ambiguous". • policy action will increase growth ie interventions targeted at ale source .ufthe deviation from the social optimum will increase growth. Examples include subsidies to R&D, subsidies to investment in human capital and subsidies to innovation. • Generating compamlive advantage .. when growth is driven by innovation or other extenlalities which are geographically concentrated, mitial conditions can generate major long tem.1 differences between countries in comparative advantage and growth potential. • Low growth traps - an implication of many of the models is thaI a country which is specialised in industries which are deficient in relevant respects nmy be trapped in a low growth situation under free trade. (see also BIE 1992). Romer 1990 makes the point in relation to input subsidies that Hwhen the decision t.o invest in physical capital i~ uncoupled from the decision to invest in research, the effects of a subSIdy to physical capital are quite different from the effects of a reduction in the mtrket interest rate". Romer concludes that "Although all the research is embodieo in capital goods, a subsidy to physical capital accumulation may be a very pOlir substitute for direct subsidies that increase the incentive to undertake re~earch." The re~earch sector in the Romer 1990 model is characterised by increasing retuOlS. such that doubling of both human capital and the stock of knowl.edge leads to an increase in the marg.inal product of human capit J.1 and a more than proportional increase in the amount of human cnpiull devoted t.o the research sector - see Figure 1. Romer considers this result to be consistent with trends observed in the real world. Romer 1990 also points to the possibility of stagnation occurring if H is too Illw because of a binding nonnegativity constraint on the amount of human capital in t.he research sector {HAl. In such a situation growth does not take place - see Figure J. Romer explains this phenomenon in tenns of all feasible rates of growth for A being too small relative to the discount rale to justify the sacrifice in current output necessary for growth to take place, and suggests that this result offers an explanation for the wide variation in growth rates observed among countries. The proposed solution for such a country is to engage in trade with other countries which has a large amount of human capital. The stimulus to growth in India and China resulting from the (partial) deregulation of trade and investment may provide support for lhis view. 4 At an aggregate economy level, it is not clear that business stealing wiU reduce consumer surplus since each suc....~edjnb inllov~tion must represent an improvement over the previous one. At an industry level, however, this may not be the e,LS<! as Grossman and HeJpman acknowledge. 6 In the contex,tof subsidising researr -: .R01l1er 1990 comments that two ,reasons can be e4pCCtedto account fQc i4~~ffi.cient hmnancapital devoted to x-esearcb. Oneisthat research has positive extemal effe:C1S hut because these take the {onn -of nonexcludable benefits, they are not .reQectW in the market price of the research, output. Thesecond~asQn is that research produces an input to a sector which engages in monopoly pricing which "forces a wedge between tbe .marginal social product of an input used in thlssectorand its market compensation".. Both of these effects cause human capital to be undercompensated whk,h in tum will cause the supply to be "too low", Romer proposes that a .govemmeot which cannot affect the allocation of human capital between different sectors should adopt a second-best policy of subsidisil1g the production of human capital. I\·ticroeconomic ASfJt.~lS of the Innovation Process Auempts in the more recent nlodels of the new growth literature to develop a microeconomic framework to incorporate various aspects of the innovation process undoubtedly achieve a greater degree of realism. Inevitably, a significant level of abstraction is also involved. In this section, the aim is to review the main features of the innovation process identified in the recent innovation economics literature and to assess whetllcr any significant gaps still remain to be addressed by the new growth models. Features of the innovation process which ate repeatedly strersed in the innovation economics Iherature are the complex ity of the innovation process, and the significant differences between sectors (see for exc:unple Dosi 1988, Schererlt)92. Pavitt 1984, Nelson and \Vinter 1974). 11 is usually taken as read that t.edmoJogical innovation represents a cruCial factor in the dynamics of economic growth. Four ba~ic modes of technological advance are identified by Dosi 1988a: • fonnalised and economically expensive processes of search eg R&D laboratories • illfomlal processes of diffusion eg publicati()ns, techmcal associations • learrung by doing and learning by using ~ adoption of innovation developed in other industries and embodied in capital equipment and intennediate inputs. Dosi 1988a summarises the innovation process in the foHowing teons ..... t. the process of innovation in Westem economies embodies complex and varying baJances between public and proprietary fonns of knowledge, and different combinations between notional oppo.rtu nj ties of innovation, finn based capabilities to reap these opportunities and (.~onomic jncentive~ to do so (related to appropriahility mechansisJns. market conditions, rehttive prices, broat!~r socio-economic conditions such as industrial relations) .............. technology..specific forms of dynamic increasing returns t~nd to ~llock-in" the. pro~ses of technical change into particular trajectories, entailing a mutual ff ,m Jrcement (a positive teedback) between it certain pattern of 7 learning and .a pattern ofalloccuionQf resources into lnnovati\'e .activities where learning has aXceady occurred in the past"• Particular aspeCts of the innovation process are highlighted by Dosi 1988 as "stylised facts"~ • illnovationinvolves a fundaxnenta! element of uncertainty • increasing reliance of major new technological opponunities on advances in scientific knowledge • increasing complexjty of research and innovation militates in favour of fonnal organisations rather than individual innovators • a significant amount of innovations and improvements originate from learning by doing and learning by using • technical ch:wge is a cumulative actIvIty, and the probability of making a technological advance is ~l function of the tedmoIogical levels already achieved. TIle mtroduction of stochastic product improvement.f) by Agitjan and Howitt 1990 and Grossman and Helpman 1991 represents a step towards dealing with lhe uncertainty issue. Howeyer, Dosi 1988a claims that. innovation search is characterised by strong uncenamty, which is more th~m the imperfe.ct inlonnation of economic analysis. The 501u11on prcp<,sed by Gold I QRO, and Winter 1986 is to employ the concept of bounded rationality 1.0 the adaptive and ieanling behaviour of economlC agents subject to severe Iimitions on the capacity of finns to make ex ollie assessments of future changes in the environment. Many of w'! other features of the immvc.uion process identified in the innovation economic literature including the endogenous nature of the proces:) (see Allen 1988. Clark and Juma 1988); the cumulative and path dependent nHture of technical change based on technological paradigms and traject.ories (see Arthur 1988, Freeman 198B; Dosi 1988. 1988a; Nelson 198~); the interaction of demand ,md supply forces in the selection process (see Nelson lQi'S~. Dosi 1988; Kline and Rosenberg 1986), and the distinction betw . . l~n public and private al)pects of knowled!te and the associated externalities and fh.:.:d tor monopoly power which from the ba.~is of increasing retums (see Allen 1988; Art.hUf 1988, Freeman 1990, Dos! 1988); appear to have been incorporated into the Jatest versions of the new growth models (see forex,unple Romer 19(0). The most trenchant criticism of the orthodox view relates to the dynrumc tUld evolutionary nature of the innovation process which is better characterised by disequilibrium and profit seeking rather than equilibrium and profit maximising. Nelson and Winter 1974 observe that neochL'isical theory ha') an "explicit or implicit commitment to the assumptions of faultless maximisation and equilibrium. Very few of the studies {In a survey b} Kennedy and ~nlirwaIlI ......concemed with the proces~cs of technical ch,Ulge employ these assumptions. Several.... implicitly deny diem. However if equilibrium is not Ubed to "close" the modcl~ some ;dtemative is required. Just a") the Jn2.croeconomic growth theories are evolving to encompass a more 8 sophlsticatedmlcroeconomicfnmlewoIk~the evolutionary theories of growthseern also to be evolving from the descriptive to a Itu;~:refQnnal modeUingbas;stbuilding on tbeearly simulation work ofNelsQn 'and Wi~(ter 1974.. One example is the use of the principle of economic sel~on in a moddlirtg framewotkby Silverberg 1988. Future Directions A number of fundamental problems for economicrutaIysis are raised by the new growth modrls and Sheehan indicates that lht!se wIll foml an important basis for future resea rch. TIlese are issues which have also been identified intbe innovation economics iterature referred to earlier. I • Increasing retulTI}' the nonconvexit.ics associat,ed with increasing returns means that "standard competitive ru.alysis will not apply" (Sheehan 1992). Sheehan 1992 and HIE 1992 refer to evidence indicating that many modem industries are characterised by large economies of scale which arise from "massive sunk cost investment" in plant and equipment. in R&D, and in education and training, from cumulafive learning by doing benefits llld from complemen.t:nrities and cooperative linkages. Sheehan argue~ that ~,lese elements are neitherextemal nor exploited in a Inonopolistic situal n. but are key elements of competitive strategy. lbe issues of increasing returns and market power are likely to receive continuing attention ill future research. • Multiple Equi •• oria: one example, the Jow gro\'.· to. Sheehan 1992 identifies four distinct SOUCt growth m{)de1~ and suggests that this issue wi] model unalys is • ExpectHuons: mo~t of the new growth models are bas(!d on perfect foresight equHibda. rille a-;sumption of mtional expect ions on the part of economic agenL~ has been criticised in the evolutionary theory literature. Versp:\gen JtJ92 ohserves that the implicit assumption of f'lltionai technological expe .ltion~ in the ne\v growth theories combined with the assumption of optimising hell a ,'iour gives most of these models strong equilibrium characteristics. EmpiricaJ research is needed to <k:tennine whether the lauer approach. or the disequilibrium process characterised by cvolutioozU)' theories is more useful. 'ready heen referred Iple eqiHbria in the new .;, centre of future growth Indeed. there is wide agreement (Romer 1990a; BIE 1992, Sheehan 1992, Verspagen 1992) that funher research is required not only to make the new growth models more useful tor policy guidance but also to give Ulem empirical validation. In the context of innovation strategy for AustJatia. DIE 1992 concludes that it is foo early to draw positive conclusions. and that more research is required before the models can be accepted as provid.ing an adequate exphmatioJl of the growth process. It seems lih.cJy that t.he contributors to the new growth theories would ,not djsagf'~t! with such a view. Nevertheless, to the extent thHt there is agreement that the new growth models achieve a greater degree of realism. wld it can be assumed that further modelling 9 work is likely to confinn the view that govf.!nunent bas ,a positive role to play, the question arises as to what fonn this role Inay take. .In a recent analysis of the Australian innovation system, Gregory questions the demand for adjustment aimed at increasing productivity and economic gro\\'th. The demand for adjustment is focussed on increasing private sector R&D, .t:~ geting .government research more. towards national priorities, increas'vl ~ competition in the research community and expanding the education system. These measures might have plausibly come straight from a proponent of tbe new growth theories. Gregory's concern is based on a number of factors: • 1110se parts of manufacturing involved is substantial R&D expenditure are shrinking as a share of tomI employment; • The paradox that a country which has never lmd a comparative advantage in high technology manufacturing is discussing solutions to balance of payments dificulties in tenns of R&D expenditure to stiInulate high technology manufacturing; and • 'llle concentration of R&D within a few finns and industries suggesth R&D policy will not have sufficient leverage on the economy as a whole. Greg(uy concludes that t1there ha5 been a significant overselling of 'he economic benefits of a govenuuent led science and technology push", and tbat it may be best to think of the National Innovation System in tenns of micro refol1ns, thnt is doing things better rather than thinking of thl~ syMem as an instrument of mncro policy that will significmnly cluUlge the structure of l.hc Aust.ralian economy.'; Recent eventh in the Australian economy including the very rapid growth in the export of ETM's and the emergence of high value-ndded manufactunng fiml'> which uccounted for $8.3BilIjon in exp0l1s in 1992 nnd .an.~ growing at 13% compound real per annum (McKinsey & Co. 1992) suggest that there lllay be more prospects for a demand puH than ,Ul S&T push in relation to Rl~: D and innovation. Rather t.han wait for events to unfold in the traditional "shetn be right" idiom. there flU), be merit in giving some urgefl\..) and priority to the fChcarch needed for the empirical iesting of the new growth models in an Australian context. 10 REFERENCES Aghion P and P Howitt, 1990, A Model of Growth through Creative Destruction,. NBER Working Paper No 3223, Cambridge, Mass. Allen P M, 1988, Evolution, Innovation and Economics, in G Dosi et aI (edsl, Technical Change and Economic Theot)'. Pinter, New York. ArroVw K J. 1962. The Economic Implications of Leanling by Doing, Rev Econ Studies. 29. BIE. 1992. Recent Developments in the 111eory of Economjc Growth: PoHcy Implications. Occ Paper 11 AGPS, Canberra. t Clark N. 1985, The ,'oliticaI Economy of Science and Technoiogy, Dlack-well, Oxford. Clark N and G JUll1a. I ~88 Evolut.ionary TIleories in Economic l11Ought, III G Dosi et al (eds), Technical Change and Economic Theory, Pinter, New York. Dasi G, 1988, The Nature of the Innovative Process, in G Dasi et al (eds), Technical Chang.t~ and Ikonomic "'Cheory, Pinter. New York. Dosi G. l<)&~a. Sources. Procedures and Microeconomic Effects of Innovation, J (:ton Lit. XXVI, 1120-71. DoS} G, C l:ireeman, R Nelson, G Silverberg and L Soete (cds). 19R8. Technical Change and Economic Theory. Pinter, New York. f recman C, 1990, Introduction, in C f';reeman (ed). The Economics of Innovatiotl Elgar, AJdcrshot xj-xx. 4 t Gold H, 1980. On the Adoption of TechnoJoglc,.U Innovatj{)n~ in IndustJ)·; Superficial Models and Complex Decision Processes, Omega, 8:5, 505-14. Gregory RG, The Aust.rahan Innov;uian System, in R Nelson (cd). National Innovation Systems - A Comparative Study, OUP, in press. Grossm.m G and E Helpman, 1989, Growth and Welfare in a Small Open Economy, NBER Working Paper No 2970, Cambridge, Mass. Grossman G and E Hclpman. 1990, Comparative Advantage and Long Run Growth, Am Beon Rev. 80:4, 796~815. Grossman G and E Helpm:ll1, 199J, Quality Ladders in the 11leory 01 Growth, Rev EcQn Studies, 58. 11 Kline S J and N Rosenberg, 1986, Au Overview of Innovation, in R Landau and N Rosenberg (eds), The Positi-!e SUfll Strategy - Harnessing Teclmology for Economic, Growth, National Academy Press, \Vasrullgton DC" 275-305. Lucas R E, 1988, On the Mechanics of Economic Development, Journal of lvlonetary, Economics, 22, 3-42. McKinsey & Co, 1992, TIle Challenge of Leadership: Australia's Emerging High Value-Added Manufacturir g Exporters, AMC, Melbourne. Montigny P, 1991, From Technological Advance to Economic Progress, OECD Observer, 170,9-12. Nelson R R, 1988, Institutions Supporting Teclmical Change in the United States, in G Dosj et al (cds), Technical Change and EcDllomic rnleory, Pinter, New York. a Nelson R R and S Winter, 1974, Neoclassical vs Evolutionary 111cOl-ies of Economic Growth: Critique and Prospectus, Beon J, 8S{, :)05. Pavitt K, 1984, Sectoral Patterns of Technical Change: Towards a Taxonomy and a 111eory, Research Policy, 13:6,343-73. Romer P M~ 1990, Endogenous Technical Change, J Pol Econ, 98:5 pt 2, S71-S 102. Romer P M, 1990a, Arc Nonconvexit.ies hnport(Ult for Understanding Growth, Amer Econ Rev, 80:2, 97-101. Romer P M, 1986, Increasing Returns and Long-Run Growth, J Pol Beon, 94;5, 1.00237. Scherer F M, 1992, Schumpcter and Plausible Capitalism, J Beon Lit, 30:3, 1416-33. Schumpeter J At 1942, Capitalism, Socialism i.Uld Democracy, McGraw Hill, New York. Sheehau P J, 1992, Economic 'I11eory and Economic Strategy: The New Growth 'Models, paper presented to the 21st Conference of Econcomists, Univ of Melbourne. Silverberg 0, 1988, Modelling Economic Dymunics and Technical Change: Mathematical Approaches t.o Self-Orgmlisation mld Evolution. in G Dosi et al (cds), Technical Change rulCl Economic'111eory, Pinter, New York. Solow R M, 1957, Technical Progress ,md the Aggregate Production Function, Rev Beon and Stats, 39, 312-20. Solow R M, 1956, A Contribution to the ll1COIY of Economic Growth, Quart J Beon, 70,65-94. 12 Swan T W, 1956, Economic Growth and Capital Accumu1ation,E~on Record, 32. Verspagen B~ 1992, Endogenous Innovation in Neoclassical Growth Models: A Survey, Journal of Macroeconomics, 14:4,631-62. Winter S G, Comments on Arrow and Lucas, Journal of Business, 59:4 pt 2, 427-34. '1'Al1t .. E 1 .\Iorlds i\ Schematlt" J)('srd,Jtion of till' Em/ogf'Tli:;atiotl (If T('duw/ogil al Chnng(' .'1 .Vrw Nro·C/as .. ictl/ Grot dh ProdurtlOn of T!I1)l' Elf Sf,,'tOrs i· motil'l· Rrfi'rt'lICf' Hotlwr : jt1~(}l ntH" ('()n~unlt l'ntt/H1 r good [.lI('il<;, 1.\1('<1 .. lHSR)-I one ('onSH mi'r good (1 H'SH1-1t two consumer goods research; interAghion nnd mediate Howitt (1990) goods; con0;\1 mer good Grossman and H<.>lpman (lHf.lH) and (WOO) rt'search & inh'rnwdi;:ttc goods, (,Ollsumer good HI no- l'nlrf'U' procP'" mnovutiOlI thr()u~h knowl.·{lg(· :wcl1mlllation t "" dllTl tl Ii : : &(1 - ('0"(' provements in ' lut';" ints for interl'Y't'niatf" ':00 Is intertemporal impr{)V( ments (positive); htl rest'arch & con- improvements in qualaty of g(n suntcr good eonsum('r good HOO1('r (1900) research & In- fJ.' i in con'iumf'r good sector: x(;)" tIlr) via utility function knowledge SPl!Jovcrs research & intcrme ate sector (positivp) intertempornl improv. mf'nts. consumer 51 plus (positive); busi ness stealing effect (negative) termediate goods; con- sumer addition of n(>w lnter-medlatt' goods (Ethlt'r pro- ,\ .,. N(A ... fI ... n) in consum!'r good research & interme L" ate sector (positive) duction [unction) II"L" ~oods rlll .. "u).unn not "('par lIt·" It\ il .. !tOllhl hf' Inft·rpn·tf'tI :1\ knowledge spillovers sector: %(')'-'-. dl \vHh x(il :> n "ElIIIII'S 111 from average humnn capital (positive) ness steaHng effect (negative) d +. [f duction runction) Pi ~';ut'tivfty stimulus frmn average human capital (rr"itlve) in consumer good 1o;('ctor: o rt) J;"; productivity stimulus L(('en) ",n onf' (·(lulhtn.·d !oil'f lor·· "= = T growth in stock of ktlowlcdJ I = investment in research L = conventional input eg labou i ith finn = in consum('r good s{'('tor. F(Ji) x{i) GrO'isman nnd II(-)pman (}99l) knowlc·dge spillovers (p()~ilive) .n ('OI.'illmf'r good ~('f'tor; F(A. II. tI) fi = N(A", /I". Votaliml /.',(T,. 1". i.-1 T') Ii addition of new ifltermC'diale t!oorls (Ethit'r pro- gnod ... ('('Inf human (> .. pitat uc('umlllation II through learning by == ou doing (',(l) = ExH!('fW/itil'.S of itwountion J>lrurture III (,otl'iUnH'f ')(lllnth·d from nhov(' irlll()(" nwdr/ arltm nwdrl strurtur(' human ratiltal Ilccllmulation II Ihmllgl. s:lVin~ stochastic (poisson \ im- /':{frrf\' of ftl1wnlliot) 0 ror i H - humnn capital u - there invested in skill enhunc A techll ()logy level = see above c = production costs i intennediate good = y =size of innovation = = = n number of designslblucprinu A productivity parameter H human capital x = intennediate good = = = g fixed range of consumer goo J.1 n parameter j index for highest position on n =number of designs A productivity parameter = H ~.; human cnpital x = intenncdiare good Sflurce: Verspagcn J992 FIGURE 1 o .. ~ .. ",,, .... Ap H ... . AD .. - G/\T! hUlllan -Growth rate and aOlount of human capital in research as a function of total capital (fOI B = 1). Source: Romer 1990