American Political Science Review Vol.94,No.2 importance of physical proximity in many aspects of may be,however,the relevance of increasing returns economic life,agglomeration effects are widespread. processes to the economics of trade is now widely That is,initial centers of economic activity may act like accepted.7 a magnet and influence the locational decisions and Economists also have applied increasing returns investments of other economic actors.Established arguments to economic change more broadly.The firms attract suppliers,skilled labor,specialized finan- most prominent development in recent discussions of cial and legal services,and appropriate physical infra- economic growth centers on "endogenous growth" structure.The concentration of these factors may in theory (Romer 1986,1990).Economists in the 1980s turn make the particular location attractive to other became puzzled by growth rates (notably in developed firms that produce similar goods.So do social net- countries after World War II)far greater than could be works,which allow for easy exchange of information explained by measured increases in inputs of capital and expertise.Increasing returns arguments help ex- and labor.Romer and others argue that increasing plain the prevalence of pockets of specialized eco- returns associated with economic applications of nomic activity,from Silicon Valley to the high-end knowledge help account for the anomaly.Unlike cap- textile manufacturers of northern Italy.Krugman ital and labor,many aspects of knowledge are nonri- (1991,80)concludes:"If there is one single area of val-their use in one firm does not prevent their use in economics in which path dependence is unmistakable, another.A single gain in knowledge can be applied in it is in economic geography-the location of production many settings and can lead to dramatic improvements in space.The long shadow cast by history over location in productivity.Economic growth generates the posi- is apparent at all scales,from the smallest to the tive feedback that defines increasing returns processes. largest-from the cluster of costume jewelry firms in A somewhat different analysis of growth based on Providence to the concentration of 60 million people in increasing returns emphasizes the importance of the Northeast Corridor." complementarities (Milgrom and Roberts 1990).Var- These claims closely parallel recent analyses of in- ious economic activities (e.g.,in information technol- ternational trade,an area in which arguments about ogy)are complementary to other related activities increasing returns have gained wide acceptance.Re- Improvements in a core activity can spill over by searchers began by focusing on economic trends that improving related parts of the economy(lowering costs appeared anomalous from the perspective of tradi- or increasing productivity).These improvements in tional trade theory-most notably,the explosion of turn may increase the attractiveness of the core activity. intraindustry trade after World War II (Krugman Economists are now applying increasing returns ar- 1996).If comparative advantage results from"natural" guments to a wide range of important economic phe- features of different countries,then one would expect nomena,but Douglass North's application to issues of most trade to occur between quite different countries, institutional emergence and change is perhaps most such as North-South trade of manufactured goods for important for students of politics.North (1990a,95) raw materials.Yet,most trade is North-North,includ- argues that all the features identified by Arthur in ing extensive exchanges within particular industries. investigations of increasing returns in technology can This pattern suggests a puzzle:Why have broadly be applied to institutions.In contexts of complex social similar countries developed highly specialized niche interdependence,new institutions often entail high comparative advantages? fixed or start-up costs,and they involve considerable Increasing returns provide an answer.Knowledge- learning effects,coordination effects,and adaptive ex- intensive sectors are prone to positive feedback.Coun- pectations.Established institutions generate powerful tries that gain a lead in a particular field,for whatever inducements that reinforce their own stability and reason,are likely to consolidate that lead over time. further development. The result is a high degree of specialization.Even North emphasizes that not just single institutions are countries with similar initial endowments develop di- subject to increasing returns.Institutional arrange- vergent areas of economic strength.Comparative ad- ments induce complementary organizational forms, vantage is not simply given,it is often created through which in turn may generate new complementary insti- a sequence of events over time. tutions.For social scientists interested in paths of It is worth noting that this research on trade has development,the key issue is often what North calls been used to derive some controversial policy implica- "the interdependent web of an institutional matrix." tions.If first-mover advantages are significant,then This matrix,he emphasizes,"produces massive increas- free trade may not be an optimal policy for a country ing returns"(North 1990a,95).Path dependent pro- whose trade partners are willing to subsidize emerging cesses will often be most powerful not at the level of sectors.Under certain (restricted)conditions,a policy individual organizations or institutions but at a more of picking winners may make economic sense (Krug- macro level that involves complementary configura- man 1996;Tyson 1993).There remains considerable tions of organizations and institutions (Hall and Sosk- dispute about the significance of such opportunities for ice 2000:Katznelson 1997). strategic intervention.Krugman,for instance,main- tains that they will appear relatively infrequently,not so much because path dependence is rare,but because 7 As Krugman (1996,109-10)notes,in the American Economic Association's classification system for journal articles,one now finds governments will not be able to identify winners ex "models of trade with increasing returns and imperfect competition" ante.Whatever the appropriate policy implications alongside the category for "conventional trade models." 255
Increasing Returns,Path Dependence,and the Study of Politics June 2000 This argument provides the core to North's sweeping Margolis (1995)distinguish remediable and nonreme- reinterpretation of economic history.The central puz- diable path dependence.The latter occurs if there are zle motivating North's inquiry is the limited conver- no feasible improvements in the path,either now or in gence of economic performance across countries over the past.Nonremediable path dependence "stipulates time.Neoclassical theory suggests that laggard coun- that intertemporal effects propagate error"(p.207). tries should readily adopt the practices of high per- With hindsight,we wish that some other alternative formers,which would induce fairly rapid convergence, had been chosen.Yet,Liebowitz and Margolis ques- but this does not happen.According to North,path tion whether this type of path dependence has pro- dependent development of institutional matrices ex- found implications.If we acted as best we could with plains the anomaly of continued divergence in eco- the information available at the time,then the mistake nomic performance.Once in place,institutions are was unavoidable,and we cannot reasonably describe hard to change,and they have a tremendous effect on the outcome as inefficient.Liebowitz and Margolis the possibilities for generating sustained economic argue that the only kind of path dependence with growth.Individuals and organizations adapt to existing major ramifications is path dependence that is poten- institutions.If the institutional matrix creates incen- tially remediable. tives for piracy,North observes,then people will invest Is their dismissal of nonremediable path dependence in becoming good pirates.When institutions fail to convincing?As Williamson (1993)notes,for policy provide incentives to be economically productive,there purposes remediability is likely to be an appropriate is unlikely to be much economic growth. standard.Recognizing the existence of path depen- For political scientists,North's insight is crucial for dence may not help policymakers much if they do not two reasons.First,he highlights the parallels between know how to identify it ex ante.s But this objection characteristics of technology and certain characteristics loses its force if our purpose is instead to understand- of social interactions.In this context,it is worth noting perhaps ex post-why aspects of societies move in that Arthur's arguments about technology are not particular directions and the consequences of such really about the technology itself but about the char- movements.And,of course,it is precisely these ques- acteristics of a technology in interaction with certain tions about causality that are the central preoccupation qualities of related social activity.Second,North rightly of most social scientists. emphasizes that institutional development is subject to The second part of the Liebowitz and Margolis increasing returns.Indeed,it is the role of path depen- analysis is the claim that remediable path dependence dence in explaining patterns of institutional emer- is rare.Their argument is straightforward.If one of two gence,persistence,and change that may be of greatest options is superior in the long run but not in the short significance for political scientists. run,then market arrangements will generally assure The dialogue surrounding increasing returns in eco- the adoption of the superior path.The ability of private nomics is the impassioned discourse of an emerging actors to capture the returns from long-term invest- paradigm.Economists talk of "new"growth theory, ments prevents bad choices.Institutions of property new"trade theory,and so on-all based on arguments rights,provisions for patents,and extensive capital involving increasing returns.Yet,despite the preva- markets mean that options with low short-run pay-offs lence of such arguments and the intellectual excite- will nonetheless receive the support they deserve ment associated with them,there are excellent reasons Economic actors,in short,calculate in the shadow of to believe that the range of application should be at the future and are,thus unlikely to indulge in myopic, least as wide in politics as in economics.To understand short-term maximizing behavior at their own long-term why,it is helpful to consider the major objections to experse. increasing returns arguments that have recently sur- This argument has considerable merit,10 but how faced in economics.The discussion will clarify the much merit depends on the strength of these mecha- sources of path dependence and identify social mech- nisms for overcoming short-term thinking or free- anisms that might offset such processes.This clarifica- riding.Although Liebowitz and Margolis are more tion provides a useful analytical bridge to an investiga- than a little complacent about the capacity of these tion of path dependent processes in politics various market mechanisms,it is wise to leave those In a forceful critique,Liebowitz and Margolis (1995) issues to economists.Two objections,however,are raise some tough questions about the literature on critically important.First,arguments about the far- increasing returns.Two aspects of their argument are sightedness of markets seem to apply to only some relevant here.They emphasize that only "remediable" types of path dependence in the economy.The Liebow- path dependence is really of theoretical significance, itz and Margolis critique focuses on the decisions of and market mechanisms ensure that remediable path firms to invest in particular technologies or products. dependence is rare.I will take up each argument in In most of the illustrations discussed earlier (e.g., turn.s Following Williamson (1993),Liebowitz and spatial agglomorations,trade specialization,endoge- As noted before,precisely for this reason Krugman and others s Note that the Liebowtiz and Margolis critique depends on both question those who make broad claims about the implications of parts of the argument being truc.The significance of path depen- increasing returns arguments for trade policy. dence for social scientists can be sustained if either the relevance of 10 Indeed,Arthur(1994,28,fn 11)explicitly recognizes this possibil- nonremediable path dependence or the prevalence of remediable ity,although as far as I know he does not systematically pursue the path dependence can be sustained. implications. 256