regional development theory

8/12/2019 Regional Development Theory

1/43

10.1177/0885412203254706ARTICLEJournal of Planning LiteratureCPL Bibliography 370CPL Bibliography

370

Regional Development Theory:Conceptual Foundations, Classic Works,and Recent Developments

Casey J. Dawkins

This annotated bibliographygives an overview of the theoret-ical literature on regional economic growth and examines itsconceptual foundations,majorcompeting paradigms,and re-centdevelopments. The overviewconcludeswitha discussionof the policy implications suggested by this body of theory.Throughout the review, three themes are emphasized: (1) thetheoretical predictions regarding the convergence or diver-gence of per capita incomes across regions over time, (2) the

assumptions regarding the importance of internal and exter-nal scale economies to regional economic growth, and (3) therole of space in shaping regional labor market outcomes.

Keywords: regional development theory; regional plan-ning; economic development; convergence

TABLE OF CONTENTSI. Introduction

A. What Is a Region?B. Conceptual Foundations of Regional

Economic Development Theory

1. The Interregional ConvergenceHypothesis

2. Location Theory and Regional Science3. External Economies4. Models of Spatial Competition5. Central Place Theory

C. Alternative Theories of Regional EconomicDevelopment

1. Theories of Regional EconomicConvergencea) Export Base Theoryb) Neoclassical Exogenous Growth Theor

CASEYJ. DAWKINSis an assistant professor of urban affairs anplanning at Virginia Polytechnic Institute and State UniversitHis doctoral dissertation at the Georgia Institute of Technologexamined theconnectionbetweenTieboutchoice and residential segregation by race.

Journal of Planning Literature,Vol.18,No.2(November2003)

DOI: 10.1177/0885412203254706Copyright 2003 by Sage Publications


2/43

2. Theories of Regional EconomicDivergencea) Cumulative Causation Theory

b) Growth Pole Theory3. Structuralist Theories

a) Stage/Sector Theoriesb) Profit/Product Cycle Theories

c) Industrial Restructuring Theoriesd) Flexible Specialization and NetworkTheory

e) Marxist Theory4. Political Institutions and Regional

Economic Developmenta) Growth Machine Theory

b) The New Institutional Economics5. Emerging Neoclassical Perspectives

a) Endogenous Growth Theoryb) The New Economic Geography

6. Conclusion: Toward an Integrated Theoryof Regional Economic Development

D. Theoretical Perspectives on the Role ofRegional Development Planning and Policy

E. ReferencesII. Annotated Bibliography

A. What Is a Region?B. Conceptual Foundations of Regional

Economic Development Theory1. The Interregional Convergence

Hypothesis2. Location Theory and Regional Science3. External Economies4. Models of Spatial Competition

5. Central Place TheoryC. Alternative Theories of Regional EconomicDevelopment1. Theories of Regional Economic

Convergencea) Export Base Theory

b) Neoclassical Exogenous Growth Theory2. Theories of Regional Economic

Divergencea) Cumulative Causation Theory

b) Growth Pole Theory3. Structuralist Theories

a) Stage/Sector Theories

b) Profit/Product Cycle Theoriesc) Industrial Restructuring Theoriesd) Flexible Specialization and Network

Theorye) Marxist Theory

4. Political Institutions and RegionalEconomic Development

a) Growth Machine Theory

b) The New Institutional Economics

5. Emerging Neoclassical Perspectives

a) Endogenous Growth Theory

b) The New Economic Geography

6. Conclusion: Toward an Integrated Theory

of Regional Economic Development

D. Theoretical Perspectives on the Role of

Regional Development Planning and PolicyE. Other Reviews of the Regional Development

Theory Literature

III. Acknowledgments

IV. Author Index

I. INTRODUCTION

How do regions grow? Why do some regionsgrow more rapidly than others? Why are differencesin levels of social welfare across regions so persistent?These central questionshave attractedtheattention of a

diverse group of scholars during the past fifty years.Topics that were initially of interest only to economistsand geographers are now being investigated by sociol-ogists, political scientists, and researchers from othersocial science disciplines. This growing interest inregional development studies is due in part to the rec-ognition that the processes driving innovation andnational economic growth are fundamentally spatial innature. In short, space matters.

This review and annotated bibliography give anoverview of the theoretical literature on regional eco-nomic growth. The reviewis intendedto serve as both asummary of the state of the field and an overview bibli-

ography for use in a graduate course on this topic.Given that this field of inquiry spans several works inseveral disciplines,no attempt wasmade to give a com-prehensive survey of all works in the field. Instead, Ireviewed seminal works and comprehensive over-views of the most important theoretical concepts.Empirical works were not examined unless they con-tributed substantially to theory development. Theannotated bibliography follows the same structure asthe introduction. Within the annotated bibliography,references are organized alphabetically by section, butnot every reference has a separate explanatory para-graph. The reference list at the end of the introduction

only includes works notdiscussed in theannotated bib-liography. To locate references in the annotated bibliog-raphy, the reader may refer either to the author index(section IV)or thesectionof theannotated bibliographythat corresponds to the section in the introduction.

The review places emphasis on three themes that arediscussed throughout the regional development litera-ture: (1) the theoretical predictions regarding the con-

132 Journal of Planning Literature
http://lastpage/


3/43


4/43

more likely to migrate to rural areas or commercial cen-ters that offer service and/or quality-of-life benefits.The implications of these changes for the definition of aregion are that regional boundaries become more diffi-cult to defineand maynotrepresent thespatial depend-encies between labor and employment centers.

Another limitation of the functional economic areaconcept and the related concept of a nodal economicregion is that local political boundaries rarely corre-spond to functional economic areas or other nodal defi-nitions, suggesting that there may rarely be a one-to-one correspondence between a particular regionalproblem and the tools that a planner may employ toresolve the problem. Similarly, since geographic or eco-logical boundaries rarely correspond to functional eco-nomic areas, planners may incorrectly estimate the eco-logical impacts of regional economic developmentusing the functional economic area definition.

Alternative definitions of regions have been pro-posedto account for theshortcomingsof the traditional

functional economic area approach. The United StatesCensus Bureau relies on estimates of commuting pat-terns to delineate metropolitan statistical areas that aresimilar to Fox and Kumars (1994) functional economicareas but correspond to administrative boundaries(counties) rather than actual commutingareas. Regionshave also been defined in terms of thedegree of internalhomogeneity with respect to some factor (Richardson1979).This approachhas been used bytheUnited StatesDepartment of Commerce to define regions across theUnitedStates basedonhomogeneousgroupingsof con-tiguous states (Hoover and Giarratani 1985). Using theexample of a region defined in terms of the size of a

labor market, a regional boundary would emerge inplaces where differences appear in the characteristicsofthe labor force. Using income as an example, one coulddefine low, medium, and high-income regions with rel-atively similar per capita incomes within regions anddifferent per capita incomes across regions. Similarly,regions may be identified on the basis of the sectoralspecialization of labor (e.g., manufacturing-basedregions versus service sector regions). Using thisapproach, regions are treated much like homogeneousnations and can be analyzed using modified methodsfrom international trade theory.

Planningregionsthat correspondto units of politi-cal or administrative control have also been proposed(Richardson 1979). The advantage of this approach isthat political and administrative boundaries directlycorrespond to the boundaries over which planners andpoliticians design and implement policies. Its disad-vantage is that economic and/or environmentalregions rarely conform to political boundaries. Thus, inthe case of an environmental region, a policy designed

for a particular political region may have spillovereffectsonadjacent environmentalregions.Boththenewinstitutional economics and the growth machine litera-ture rely on this approach to define regions.

Regions may also be defined in terms of naturalresource, ecosystem,or othergeographic boundaries. Afewauthors suggest an interesting approachto defining

regions in terms of the interdependencies between nat-ural resource systems and human populations.Markusen (1987) defines a region as a historicallyevolved, contiguous territorial society that possesses aphysical environment, a socioeconomic, political, andcultural milieu, and a spatial structure distinct fromother regions and from the other major territorial units,city and nation (pp. 16-17). This definition recognizesthat regions are historically determined entities thatemerge largely due to the interaction between humansand local natural resources. Althoughimprovements intransportation have removed many of the constraintsimposed by geography, the historical patterns ofregional formation still affect the evolution of modernregions. Cronon (1991) adopts a similar view and dis-cusses the emergence of the Chicago metropolitanregion as a historical pattern of interdependence

between economic forces and the Chicago River andLake Michigan.

For the purposes of this review, a region will bedefined as a spatiallycontiguous population (ofhuman

beings) that is bound either by historical necessity or bychoice to a particular geographic location. The depend-ence on location may arise from a shared attraction tolocal culture, local employment centers, local naturalresources, or other location-specific amenities.

B. Conceptual Foundations of Regional EconomicDevelopment Theory

Regional development theory emerged from severaldifferent intellectual traditions. Neoclassical trade the-ory and growth theory provide the conceptual basis forunderstanding whether regional economies will

become more similar or more differentiated over time.The spatial dimension of modern regional growth the-ory can be traced to several sources. Location theoristsprovide a framework for understanding the role oftransportation costs in regional growth and decline.The literature on external scale economies that beganwith Marshall ([1890] 1961) has been rediscovered bymore recent neoclassical theorists and those writing inthe flexible specialization tradition. Finally, ideas fromcentral place theory resurface throughout the regionaldevelopment literature, especially in the growth poleliterature andin manyrecent structuralist approaches.

http://lastpage/


5/43

1. THE INTERREGIONALCONVERGENCE HYPOTHESIS

Most early theories of regional economic growthwere aspatial extensions of neoclassical economic theo-ries of international trade and national economicgrowth. Together, these early neoclassical theories pre-dict that over time, differences in the price of labor andother factors across regions will diminish and tendtoward convergence. This prediction has generatedconsiderable controversy among theorists, particularlyin light of the apparent tendency toward internationaldivergence between the per capita incomes of industri-alized nations and less developed nations. Early theo-ries of regional economic development emerged out ofthis controversy and can be distinguished from oneanother in terms of differences in the theoretical predic-tions regarding interregional convergence or diver-gence in per capita incomes and factor pricesover time.Thesetheoreticalresponsesare examinedin moredetailin the next section. First, it is important to look at the

interregional convergence hypothesis.Neoclassical trade theorists draw on the Heckscher-

Ohlin-Samuelson (HOS) theorem to explain interna-tional factor price convergence using staticequilibriumtrade models. This well-known theory of internationaltrade begins with the following simplifying assump-tions (thisdiscussion drawsheavilyonSalvatore1998):

1. Two regions (1 and 2) trade two commodities (Aand B) using two factors of production.

2. Theproductionof A is labor-intensive, and thepro-duction of B is capital-intensive.

3. Bothregionsrelyonthesametechnologyinproduc-

tion and have the same production functions.4. There areconstantreturns to scale in theproduction

of Aand B.5. Both regions produce some of Aand some of B.6. Tastes are homogeneous across regions.7. Commodity and factor markets are perfectly

competitive.8. Factors are mobile within nations but not mobile

across nations.9. There are zero transportation costs.

10. Allresourcesare used upintheproduction of AandB.

11. Trade between 1 and 2 is balanced such that thevalue of regional exports is equal to the value of

regional imports.

With theseassumptions, Heckscher(1919) andOhlin(1933) demonstrate that a factor-abundant region willhave a comparative advantage in the production ofgoods that require the intensive use of that factor. Thisregion will then specialize in and export the factor-abundant good and import goods for which factors of

production are scarce. This result can be explained asfollows:

If theassumptionis made that themarkets forfactorsand commodities are perfectlycompetitive, the relativeabundanceof a factorin a givenregion canbe expressedin terms of the ratio of prices for the two factors. Laborabundance, for example, can be expressed in terms ofthe ratio of wages to interest rates. If labor is relativelymore plentiful in a given region, then this implies thatthe relative price of labor (wages) is lower, which fur-ther implies that the region will have a comparativeadvantage in the production of labor-intensive goods,

because the productionof labor-intensivegoods is rela-tivelycheaperfor that region. If regions specialize intheproductionof goods forwhichrelativefactors areabun-dant and export those goods, importing goods forwhichfactors arescarce, both regions gain fromspecial-ization and trade.

Samuelson (1953, 1949, 1948) elaborates on theHeckscher-Ohlin result to demonstrate how free trade

and/or factor mobility equalizes the relative and abso-lute long-run prices of factors of production amongregions involvedin trade.Assumethat region 1 special-izes in the production of A, the labor-intensive good,whereas region 2 specializes in the production of B, thecapital-intensive good. Once trade opens between thetwo nations and specialized production begins, the rel-ative price of labor in the labor-abundant region risesdue to relative increases in the demand for labor. Con-versely, the relative price of capital rises in the capital-abundant region due to relative increases in thedemand for capital. Even if capital and labor are immo-

bile, the aggregate effect of these market forces is to

equalize relative factor prices across regions. With fac-tor mobility, less trade is required to equalize relativefactor prices. Furthermore, with our assumptions ofperfectcompetition, homogeneous production technol-ogies,andconstant returns to scale, trade also equalizesthe absolute prices of labor and capital. In other words,real wages and real interest rates for similar types oflabor and capital will be the same in both regions fol-lowing trade and specialization (Salvatore 1998).

The HOS theorem is complementary to DavidRicardos theory of comparative advantage (1817) inthat the Heckscher-Ohlin model explains why compar-ative advantages exist (differences in initial factorendowments), whereas Ricardos theory only estab-lishes why comparative advantages may lead to spe-cialized production.The HOStheorem alsohas obviousimplications for regional trade and development. In itssimplest form, themodelsuggests that specialization infactor-abundant production combined with free inter-regional trade will result in equal per capita incomesacross regions for workers with similar skills. This

CPL Bibliography 370 135
http://lastpage/


6/43

hypothesis is a comparative statics version of the inter-regional convergence hypothesis.

Dynamic versions of the convergence hypothesisdraw on neoclassical growth theory, particularly themodels proposed by Solow (1956) and Swan (1956). Inneoclassical growththeory, there aretwodifferent typesof convergence. Conditional convergence refers to theconvergence toward a steady state growth rate result-ingin constant percapita incomes, consumption levels,and capital/labor ratios. This is termed conditional,

because savings rates, depreciation rates, and popula-tion growth rates are allowed to differ across countries.Therefore, conditional convergence need not necessar-ily result in equal per capita income levels across coun-tries.Absolute convergenceoccurs when growth modelparameters are equal for all countries, which in turnimplies that richer countries will grow slower thanpoorer countries, and per capita incomes will becomeequalized across countries over time as in the HOSmodel of international trade.

There are several reasons why it is important to dis-tinguish between the convergence hypothesis of theHOSmodel and the convergence hypotheses from neo-classical growth theories. First, neoclassical growthmodels are, by definition, dynamic models, so theirconvergence hypotheses refer to the convergence ingrowthratesratherthan thestaticconvergence of factorprices. Although both models predict the eventuallong-run convergence of per capita incomes acrossregions, the process that brings about convergence dif-fers between the neoclassical trade and growthmodels.Since most neoclassical growth models typicallyassume away trade by modeling growth within closed

economies, convergence occurs not through trade orfactor mobility but through diminishing returns to cap-ital investment. In neoclassical growth theory, regionswith less capital per unit of labor will tend to havehigher rates of return and higher initial growth ratesthan regions with high levelsof capital perworker (Barroand Sala-i-Martin 1999). Althougha regional version ofthe neoclassical exogenous growth model proposed byBorts and Stein (1964) considers interregional factormobility, most neoclassical growth models assume per-fect intraregional factormobility butassume zero inter-regional factor mobility. Finally, neoclassical growthmodels often allow for differences in production tech-nologies and/or savings rates across regions. If theseparameters are assumed to be exogenous, then regionswill only conditionally converge toward a steady stateconstant rate of growth. In neoclassical growth modelsthat allow for variability in growth parameters, thesteady state may differ across regions, but all regionseventually reach constant per capita income, consump-tion, and capital/labor ratio values.

The possibility of dynamic or static interregionalconvergence has obvious implications for regionaldevelopment theory: tradeand investmentwilleventu-ally lead to an equalization of wages acrossregions. It isimportant to note, however, that this does notnecessar-ily imply equalization of per capita incomes, since percapitaincomesdependon additional factors such as theskill level of the population and the percentage of thepopulation that is in the labor force. Thus, although theHOSmodel implies convergence in wages across coun-tries, it does not necessarily imply convergence in percapita incomes, a point that is often ignored by critics ofneoclassical trade theories. Also, since growth parame-ters may differ across countries, we may only observethe weaker form of conditional convergence over time,and per capita incomes may differ due to differencesacross regions in production technologies or savingsrates.

2. LOCATION THEORY AND REGIONAL SCIENCE

Most current theories of regional economic develop-ment can be viewed largely in terms of their criticismsand response to the convergence hypothesis and neo-classical economics more generally. Location theorywas developed as an early response to the ignorance ofspace in traditional economic analyses. Originallydeveloped by Alfred Weber (1929) and later extended

by Edgar Hoover (1937), Melvin Greenhut (1956), andWalterIsard(1956), locationtheory has focused primar-ily on developing formal mathematical models of theoptimal location of industry given the costs of trans-porting raw materials and final products. Simplystated, firms will tend to locate near markets when the

monetary weight (defined as the shipping costs permile times the physical weight of the item shipped) ofthe final product exceeds the monetary weight of theinputs required to produce that product. Conversely,firms will tend to locate near primary input sourceswhen the monetary weight of rawmaterials is large rel-ative to the weight of the final product. Firms may alsoweigh the relative production cost savings from partic-ular locationswith theincreased transportation costs tominimize the total costs of production andtransportation.

Although location theory alone does not provide atheory of regional economic development, the explicitmodels of transportation costs have been highly influ-entialin later theoriesofeconomicgrowthand develop-ment,particularly the neweconomicgeography. WalterIsard (1956) eventually drewon concepts from locationtheory to develop the field now known as regional sci-ence, a branch of the social sciences that examines theimpactof space oneconomic decision making. Theana-lytic methodologies developed by Isard (1960) and

http://lastpage/


7/43

extended by Isard et al. (1998) have become standarditems in the regional planning professionals toolbox.

3. EXTERNAL ECONOMIES

One problemwithtraditional Weberian location the-ory is that the cost advantages of spatial proximity toinputs and markets are modeled purely in terms ofinternal transportation cost economies. As economistssince Marshall ([1890] 1961) have pointed out, indus-tries maycluster together forreasons unrelatedto inter-nal cost considerations. Instead, firms may cluster totake advantage of external economies that result fromclose proximity to a large number of other firms. Fol-lowing Hoover (1937), these external economies mayinclude (1) localization economies that result from thefirms in the same industry colocating in the same areaand (2) urbanization economies, which result from thecolocation of firms in different industries. Since theseexternal benefits tend to increase with the number andoutput of colocating firms, they are usually referred to

as external scale economies or agglomeration economies.Economists continue to disagree over the nature andcause of these external economies, but broadly speak-ing, knowledge spillovers, labor pooling, and econo-mies in the production of intermediate inputs have all

been cited as contributing factors. Because externalscale economies are characterized by both positiveexternality effects andincreasing returns to scale, tradi-tional competitive market models have tended toignore these effects.

4. MODELS OF SPATIAL COMPETITION

Another benefit of spatial proximity from a firms

perspective is the ability to charge higher prices to cus-tomers that are located within close proximity of agiven distribution point. This observation, firstexplored by Harold Hotelling (1929), has produced aconsiderable literature on the role of space as it affectsthepricing behaviorof firms. Theessence of Hotellingsargument is that spatial proximity gives firms marketpower, because nearby customers would be willing topay more for goods that can be consumed withoutincurring substantial transportation costs. In the sim-plest case with two firms competing along a straightline, monopolistic competition in space produces a ten-dency toward concentration with firms splitting themarket along the line segment. This optimum locationis not socially efficient, however, since customers ateither end of the line must incur higher transportationcosts. Works by Devletoglou (1965), Eaton and Lipsey(1978), and many others extend Hotellings originalmodel to incorporate thethreatof entry by competitors,demand elasticity, and competition along a plane.Theseextendedmodels demonstrate thatconcentration

is not always the equilibrium outcome and that thethreat of entry may or may not always drive profits tozero.

5. CENTRAL PLACE THEORY

An early attempt to bring some of these perspectivestogether in a more general theoryof the spatial locationof firms can be found in the work of Christaller (1933)and Losch (1954). Christaller first formulated central

place theory, as it came to be called, to describe the distri-bution of cities of different sizes within southern Ger-many. Losch expands on the initial ideas of Christallerand places them into an economic context, introducingtheideaofa demand cone into thehexagonalmarketareaframework developed by Christaller. The basic ideaelaborated by Losch is that the relative size of a firmsmarket area, defined as the territory over which it sellsits product, is determined by the combined influence ofscale economies and transportation costs to markets. Ifscale economies are strong relative to transportation

costs, all production will take place in a single plant. Iftransportation costs are large relative to scale econo-mies, firms will be scattered around the region. For anygiven market, free entry among firms drives profits tozero and causes all spaces to be occupied by equallyspaced firms with hexagonal market areas. However,due to differences in transportation costs, scale econo-mies, and demand fordifferentproducts, the size of theindividual hexagons will be different for different mar-kets.Central placesemerge in locations where marketareas fordifferentproducts overlap. As indicated in theprevious section, this process of monopolistic competi-tion in space produces a hierarchically structured sys-

tem of cities of different sizes and different levels ofproduct diversity.

Although location theory and central place theoryhave each contributed considerably to ourunderstand-ing of the spatial pattern of firms, their static perspec-tive and ignorance of many important dimensions ofregional economic growth, particularly labor migra-tion, has impaired their use as a general theory ofregional economic development. Theories of regionaleconomic development incorporate these concepts intomoreformal expressions of regionalgrowth dynamics.

C. Alternative Theories of Regional Economic Development

According to the interregional convergence hypoth-esis, interregional trade and regional investmentshould eventually lead to the equalization of wagesacross regions and the equalization of per capitaincomes across regions with equal labor participationrates, skill levels, and investment levels. The first twosets of theories examined in this section can be definedin terms of their stance on the interregional conver-

http://lastpage/


8/43

gencehypothesis discussedabove. Following thesetwotheoretical perspectives are several alternative perspec-tives that view regional growthanddecline as resultingfrom underlying structural changes in the organizationof industry and thepolitical-economic system. Thefinalsection examines two newperspectives from neoclassi-cal economics that incorporate many earlier criticisms intonew theoretical approaches. The final subsection exam-ines the common themes among all of these theories.

1. THEORIES OF REGIONALECONOMIC CONVERGENCE

a) Export Base TheoryAmong all theories discussed in this review, few

have been as influential as theexport base model devel-oped in the1950s by Charles Tiebout (1956a, 1956b) andDouglass North (1956, 1955). North (1955) argues thatregional growth in local political, economic, and socialinstitutions is largely determined by the regionsresponse to exogenous world demand. This responseproduces growth in both the economic base, or export

sector, and the residentiary, or nonbasic, sector,which exists only to serve thebasic sector. Furthermore,North points out that regions need not necessarilyindustrialize to grow, since a regions exports may con-sistof either manufactured goods, service-basedgoods,or agricultural goods.

As regions grow, their economy becomes morediversified,dueto increases in local productionto serveincreasing local per capita incomes and the emergenceof new industries serving export markets. Over time,regions will tend to lose their identity as regions(North 1955, 258). With the increasing diversity ofregional exportbases and the mobility of factors of pro-

duction,production will tendto disperse across regionsover time, and per capita incomes will tend towardinterregional convergence as in theHOS model of inter-national trade.

In a debate between Douglass North and CharlesTiebout in theJournal of Political Economy, the modernversion of thetheory came into fruition.Tiebout (1956a)argues that Norths model ignores the importance ofmany important supply-side factors that ultimatelyaffect a regions ability to support an emerging export

base. He also criticizes Norths article (1955) by point-ing to other instances when exports are not the soledeterminants of regional economic growth. For exam-ple, in regions with populations large enough to affectthe worldwide demand for exports, regional growth inper capita incomes may be affected by an increase inexport demand and may affect world demand forexports. This bidirectional causality implies that thereare significant feedback effects between regional percapita income growth and export market demand.North (1956) replies to Tiebouts criticisms by pointing

out that the model should be viewed as a long-runmodel of economic growth that may not always beapplicable in the short run when certain factors of pro-duction are fixed and immobile. In the long run, how-ever, the model still holds as an adequate account ofregional economic growth. In a final rejoinder (1956b),Tiebout reiterates that theexportbase concept is merelyan oversimplified version of more sophisticated gen-eral equilibrium national per capita income models.Furthermore, the stages theory of economic growthcriticized by North is not necessarily wrong, accordingto Tiebout. Instead, it is only applicable for a more lim-ited number of cases.

b) Neoclassical Exogenous Growth Theory

The mainstream neoclassical economic view ofregional economic growth draws heavily on the litera-ture of national economic growth developed by Roy F.Harrod (1939) and Evsey D. Domar (1946). In contrastto the demand-side approach of export base theory,neoclassical growth theory models regional growthusing supply-side models of investment in regionalproductive capacity. Early versions of this theory areoften referred to asexogenousgrowth theory, becausesavings rates, population growth rates,and technologi-cal progress parameters are all determined outside themodel. The models developed by Solow (1956) andSwan (1956) have been the most influential in moderngrowth theory, primarily due to the more general formof the regional production function, which allows forsubstitutability among production inputs in accor-dance with production functions that assume constantreturns to scale and a positive elasticity of substitution

among inputs (Barro and Sala-i-Martin 1999). Thesefeatures generate predictions ofconditional convergenceof growthratesover time acrosscountriesand thelevel-ingoff of percapitaincomes withincountries. If growthparameter values are the same across countries, thenneoclassical exogenous growth theory also predictsabsolute convergence in per capita incomes, as dis-cussed in the previous section.

Borts and Stein (1964) modify the neoclassicalgrowth model for the regional context by allowing foropen regional economieswith net exogenous labor andcapital inflows. Barro and Sala-i-Martin (1999) arguethat interregionalconvergence is morelikely thaninter-

national convergence, because factors of production aremore highly mobile across regions. Furthermore,abso-lute convergencein per capita incomes across regionswithin a country is more likely due to the homogeneityof savings rates, depreciation rates, population growthrates, and production functions within countries.

Williamson (1965) modifies the Borts and Stein(1964) argument somewhat by suggesting several rea-

http://lastpage/


9/43


10/43

spatial context. Perrouxs (1950) space as force viewof spatial interaction, which defines space as a type ofnetwork that is held together by centripetal forces, hasformed the basis of most growth pole theories.Although this view of space is not unlike that which isadvocated by those in the flexible specialization/net-work theory tradition, the two theoretical perspectiveshave largely developed in isolation from one another.

In Perrouxs (1950) original formulation, a growthpole referred to linkages between firms and industries.Propulsive firms are those that are large relative toother firms and generate induced growth throughinterindustry linkages as the industry expands its out-put. Hirschman (1958) argues similarly in his discus-sion of backward and forward linkages between firms.Boudeville (1966) is credited for placing Perrouxs for-mulation into geographic space. For Boudeville, agrowth pole is defined in terms of the presence of pro-pulsive firms and industries that generate sustainedregional growth through linkages with other firms in a

region.Hirschman (1958) discusses how polarized develop-ment may benefit both the growing region and the sur-rounding hinterland. Like Myrdals spread andbackwash effects, Hirschman argues that growthin adeveloped regionproducesfavorabletrickling-downeffects within a lagging region as the lagging regionsgoods are purchased and labor hired by the developedregion. Growth may also produce unfavorable polar-ization effects resulting from competition and trade

barriers erected by the developed region. Despite thesesimilarities, Hirschman rejects Myrdals cumulativecausationapproach as overlybleakdueto thefact that it

hides the emergence of strong forces making for aturning point oncethemovementtowardsNorth-Southpolarization within a country has proceeded for sometime(p.187, n. 5a). In theend,Hirschmanhas faith thattrickle-down effects will outweigh polarization effectsdue to increased pressure to enact economic policies tocombat the latter.

A related perspective is Friedmanns (1966) center-periphery model, which includes elements of Myrdalstheory of unbalanced regional growth and export basetheory. Like export base theory, Friedmann recognizesthat growth may be externally induced. He also pointsto the impact of interregional labor migration on theconvergence of incomes across regions. However,Friedmann departs from traditional export-based theo-ries of economicgrowthby pointing outthat localpolit-ical and economic entrepreneurship and leadershipmay affect the translation of export demand intogrowth in the nonbasic residentiary sector. The qualityof local leadership is in turn affected by the regionsdevelopment history. Friedmann also points out that

regions may vary in the extent to which supply con-straints limit a regions ability to respond to increaseddemand for exports. Finally, large urban areas have theinitial advantage in the competition for new growth

because of the decreasing cost benefits of urbanizationeconomies. All these factors tend to work to the advan-tage of core regions, which are incumbents in the eco-nomic development game. Outside of the core, regionsare differentiated by their relative degree of regionaleconomic autonomy. Resource frontiers are undevel-oped regions whose primary draw is the plentiful sup-ply of untapped natural resources. Downward-transi-tional areas are rural areas trapped in a stage ofstructural poverty, primarily due to their structuraldependence on adjacent core regions.

Growth pole theory was largely abandoned in the1980s dueto growing dissatisfaction with theperceivedlackof coherence betweentraditional notionsof growthpoles and empirical reality. Many growth pole policieswere shown to fail in their intended objectives of induc-

ing new economic growth in lagging regions. Othercriticisms also emerged, such as the inappropriate useof input-output analyses to examine the spatial interac-tions between firms, the difficulties of translatingPerrouxs original abstract formulation into useful the-ories of regional economic development, the lack ofemphasis on the process of structural change withingrowth poles over time, the weak behavioral basis ofthetheory, andthelack of explanationwithinthe theoryabout why some growth poles tend to grow faster thanothers (Darwent 1969; Higgins 1983; Thomas 1972;Hermansen 1972).

3. STRUCTURALIST THEORIES

Another body of theoryexamines regional economicdevelopment as a process of structural adjustment bothwithin and outside the region. Rather than viewregional economic growth in terms of the factors push-ing regional economies toward or away from someequilibrium rate or distribution of growth, these theo-rists view economic growth as a path-dependentevolu-tion through various stages of economic maturity.

a) Stage/Sector Theories

Early perspectives in the structuralist traditioninclude several different stage theories of regionaleconomic growth. Since many of these theories alsoinclude a focus on sectoral change, some are alsoreferred to as sector theories (Perloff et al.1960). Hoo-verand Fisher (1949) present an early theoryof sectoralchange through various stages of regional growth. Inthe early stages of regional growth, agricultural pro-duction predominates and the economy is largely self-sufficient. As transportationimproves, producers begin

http://lastpage/


11/43

to specialize and engage in outside trade with otherregions. As diminishing returns begin to occur in theproduction of the regions primary extractive and agri-cultural industries, the region enters a phase of indus-trialization. At themost advanced stage,the region spe-cializes in export production. In this theory, theprogression from self-sufficiency to export producer islargely seen in terms of the internal changes in the divi-sion of labor that produce economic specialization.

Other early stage theories developed to explainnational economic growth have also been used toexplain regional economic growth and development.Schumpeter (1934), like Hoover and Fisher, sees eco-nomic development as occurring from within theregion. Regional economic change can be viewed as aprogression through long waves of growth and declinethat are distinguished from one another through thedifferences in the nature of the innovations that charac-terize each period. New innovations emerge through aprocess of creative destruction, where old ideas are

constantly replaced by new ones. Rostow (1977, 1956)providesa related view inhisdescriptionof a take-offperiod, where a rapid revolution in the means of pro-duction leads to a relatively long-term wave of sus-tained growth.

Thompson (1968) presents another stage theory ofurban/regional economic growth. According toThompson, urban areas grow by progressing from anearly stage where the local economy is largely equatedwith a single large industry or firm through variousphases of export-led growth until the export of services

becomes the major function in the final stage. At somepointduring this development process,a ratchet effect

occurs, where growth patterns become locked intoplace and future contraction becomes unlikely. Thomp-son gives several possible explanations for this effect:

1. In diverse urban economies, small firms have mul-tiple local linkages that are difficult to reproduce ifthe firms relocated to another region.

2. Cities with larger populations are more successfulin garnishing political spoils from state andnational government entities.

3. Per capita public service costs are significantlylower due to economies of scale.

4. A large local population base is valuable as a labor

resource and a potential consumer market forlocally oriented industries.5. Large areas give birth to more local industries than

small areas, which in turn increases the probabilityof local innovations.

Thompson concludes by pointing out that disecono-mies of scale associated with congestion and bureau-cratic costs may set in to counter the ratchet effect.

Pred (1977) discusses how the structure of informa-tion flows between economic agents affects the eco-nomic development of city systems. A central premiseof the work is that spatial biases in the flow of infor-mationtend to give incumbenturban centers an advan-tage in economic growth. Furthermore, the flow ofinformation across the landscape occurs primarilyamong the system of large metropolitan centers, thusreinforcing the stability of the system of cities. Predrelies on this basic idea to account for the historicald ev el op me nt o f u rb an a re as . D ur in g t hepretelegraphic period of urbanization, when urbancenters emerged primarily to facilitate trade, thespatial

bias was most pronounced due to the importance offace-to-face communication within cities and amonglarge trading centers. Dueto spatial biases, trading cen-tersestablished in the pretelegraphic centers weremorelikely to become the sites of initial industrialization.Once established, multiplier effects gave these initialindustrial centers a cumulative advantage in economic

growth as innovations in production technologies dif-fused among local factory owners. During thepostindustrial period, multilocational corporationsemerged to transform cities. Unlike growth pole theo-r is t s, P re d c on c lu d es t ha t t he i nt er - a n dintraorganizational linkages between the administra-tive hubs found in large metropolitan areas are self-reinforcing and not likely to result in a spread ofgrowth into lower-level urban centers or surroundinghinterlands (p. 122).

Henderson (1974) provides an interesting look athow changes in industry structure lead to different citysize distributions. Essentially, Henderson argues that

the relationship between the utility of any given cityresidentand city size can be represented by an invertedU, which captures external economies of scale on theleft side anddiseconomiesof scale ontheright.Thelink

between this argument and industry structure is thetwist that whereas diseconomies are predominantlyassociatedwithpopulation size,external economies areindustry specific. Thus, it only makes sense to groupfirms that share thesame external economies withinthesame city. This suggests that the optimal size of anygiven city over time will depend on its role, which is afunction of the industry structure that dominates the

city.b) Profit/Product Cycle Theories

Vernons (1966) product cycle approach viewsregionaldevelopment andchangein termsof theevolu-tion of regional industry structures required to sellexport goods. Due to lowprice elasticity of demand fornewproducts, an innovating firm cares less about smallinitial cost differences between regions than about

http://lastpage/


12/43

future cost considerations. Furthermore, in the earlystagesof a products life, locational proximity to suppli-ersandresearchanddevelopment firms is importanttofacilitate the flexible incorporation of product changesand process innovations. Thus, large urban areas will

be preferred locations for firms producing new imma-ture products. As the product matures and becomesmore standardized, the need for flexibility diminishes,and the need to focus on economies of scale increases.Once production has been standardized, the firm canemploy cheap low-skilled labor, so underdevelopedregions become preferred locations. Weinstein et al.(1985) rely on this perspective to explain patterns ofregional development in the United States.

Taylor (1986) provides several criticisms of thisapproach:

1. Themodel provides an ambiguous treatment of theinternationalization of production and the owner-ship dimensions of internationalization.

2. The model is at odds with more conventional incre-mental views ofproductinventionandinnovation.3. The model ignores product differentiation.4. Theassumption of shifts to low-cost labor locations

in the final stage of the product cycle implicitlyassumes that cheaplabor is theprimary cost consid-eration, an assumption that is not always true.

5. The model assumes a homogeneous geographicplane on which firms compete.

6. Market cycles may not always be consistent withinternational product cycles.

Markusens (1985)profit cycle theoryis one modifica-tion of theproduct cycle approachthat responds to sev-

eral of Taylors (1986) criticisms by incorporating afocus on industry structure at various stages of a prod-ucts history. According to Markusen, sectoral changewithinregionscorrespondstooneof five profit cyclesthat are determined by the structure of competition atvarious stages of product development. Initially, sec-toral development precedesfrom a periodof zero profittoward a period of superprofits, where initial innova-tors earn monopolyprofits.Thesector then entersa nor-mal profit stage as new firms enter the market. Even-tually, the market becomes saturated, and destabilizingfactors set in. During this phase, firms either tend

toward oligopolistic formsof organizationto gainaddi-tionalprofitsor thefirmsentera stage ofdecline,as sub-stitute or imported products take over the market. Thefinal negative profit stage is one of sectoral declineand disinvestment.

Each stage in Markusens (1985) profit cycle is char-acterized by unique spatial relationships. In the initialstages of a products life, the location of firms is largely

determined by historical accident or by the physicallocation of the innovation. Entrants into the marketmaybe drawn to thelocation of theinitial innovationorto regions whose resources are favorable to the indus-try. During the superprofit stage, industries colocateto benefit from knowledge spillovers and a localizedskilled labor force. Eventually, firms grow in size,diminish in number, and become increasingly orientedtoward the location of consumer markets. If the indus-try tends toward an oligopolistic structure, firms willtend to concentrate to take advantage of market powerresulting from proximity to consumer markets andlocation-oriented political supports. During later peri-ods, oligopolies seeking to minimize labor costs mayrelocate to escape unionization. If firms enter a finalstage of decline, the spatial tendency will be one ofdivestiture and gradual abandonment of location-specific facilities.

c) Industrial Restructuring Theories

Several new empirical realities began to emerge inthe late 1970s and early 1980s that led to the emergenceof new structural explanations of regional growth anddevelopment. Among these trends have been thedecline of manufacturing and the emergence of the ser-vice sector in the industrialized world, the increasinginternational mobility of capital and labor flows, andthe growing interregional disparities in labor condi-tions across gender and ethnic lines. These and otherstudies in theliterature suggest that fundamental shiftsin the organization of industry and labor have resultedina deskillingof thelaborforce(Harrison 1985),a rel-ative decline in the proportion of workers earning mid-

dle-income wages (Leigh 1994), and a spatial stratifica-tion of the workforce (Massey 1984).

One response among regional development theo-rists wasto look forexplanations forthesetrendswithinthe changes that were occurring in industrial organiza-tion. The industrial restructuring perspective exam-ines how structural changes in the organization ofindustry have affected regional capital and labormarkets.

Several studies in this tradition point to the interna-tionalization and mobility of capitalist production andits effect on workers. According to Sassen (1988), direct

foreign investment has disrupted traditional laborstructures. In the developing world, frequent layoffsresulting from insecure manufacturing jobs have cre-ated a large supplyof female migrant workers, many ofwhom were previously employed in the nonwagehousehold sector. In the United States, the internation-alization of capital flows resulted in the disinvestmentin many U.S. industries. This disinvestment in national

http://lastpage/


13/43

productive capacity has in turn resulted in the destruc-tion of socialand communityties inmany regions of theUnited States (Bluestone and Harrison 1982).

Massey and Meegan (1982) examine of the geogra-phyof employmentdecline andhowfirmsusejob elim-ination as a corporate strategy. The authors begin byexamining three different corporate strategies that

typically lead to job loss. Intensification strategies seek toimprove labor productivity without substantial newinvest-ments. Investment and technological change strategiesresult in changes to productive technology.Rationaliza-tion strategiesare those focused onthesimple reductionof labor capacity. Since each of these strategiesmay leadto differences in the number of plant closures andinvestments in new capacity, the authors argue thatregions are affected in differentways depending on thenature of the strategy leading to employment decline.

Another trend has been the transformation from amanufacturing-based to a service-based economyamong advanced industrialized nations. Noyelle and

Stanback (1983) attribute the rise of the service sector tothe increased geographic size of markets, innovationsin transportation technology, the increased importanceof public and nonprofit sectors, and the rise of the mul-tinational corporation (p. 3). Their empirical studyfocuses on the structural changes resulting from theseforces acting on U.S. metropolitan areas. Regardingregional development, the authors argue that the trans-formation to a service-based economy has fosteredincreased centralization among corporate activitiesaccompanied by the decentralization of many low-skilled white-collar jobs. Theauthors also pointout thatmany olderregions withintheSnowbelt have managed

to stave off decline following the loss of manufacturingjobs with concomitant growth in corporate activitiesand government-sector jobs.

Storper and Walker (1984) explore the importance oflabor to the location decisions of industries. Labor,unlike other factors of production, is inherently hetero-geneous across space due to differences in culture,social institutions, and production requirements acrossthat same space. Furthermore, unlike other commodi-ties, labor is not purchased outright. Instead, it is

boughtand sold subject to uncertainexpectationsaboutfuture performance and reproducibility. Firms respond

to andtake advantageof thespatial heterogeneityof thelabor force as a way to exert control over their workers.For example, firms can easily escape unionization byrelocating to another region. Similarly, firmscan exploitthe spatial relationships between workers within indi-vidual plants to reduce the tendency toward workersolidarity. Workers, on the other hand, may also takeadvantage of the mutual dependency between the firm

and the worker and successfully capture concessions iftheir skills are sufficiently scarce outside the region.

Danson (1982) argues that the emergence of adualist industrial structure has been a primary causeof the stratification and segmentation of the labor mar-ket. The author begins by discussing the emergence ofthe recent period of monopoly capitalism, where

large core firms embedded in oligopolies have come todominate smaller competitively structured peripheryfirms. The labor market is in turn differentiated basedon skill level and is largely confined to distinct seg-ments based oninternal linkagesto specific firms, tradegroups, or industries. Jobs within different skill strataor industry segments often are also highly differenti-ated in terms of worker benefits and wages. Theremainder of thearticle discusseshowthe emergence ofperipheral firms in many inner cities and regions hasled to the relative decline of those locations relative tothe locations housing core firms.

d) Flexible Specialization and Network TheoryAnother theoretical response to these recent changes

inthestructure of industry hasbeenthedevelopment ofa new theoretical approach that focuses on the patternsof interrelationships found in new industrial districts.Piore and Sabel (1984) discuss how increasing socialunrest, floating exchange rates, oil shocks, the interna-tional debt crisis, the saturation of industrial markets,andthediversificationof consumer demands have pro-duced a new form of production designed to perma-nently respondto changethroughinnovation. Thisnewflexible specialization is based on the use of flexiblelabor and capital that can easily be tailored to the needs

of changing markets. Firms engaged in flexible special-ization are bound together through highly localizednetworks where knowledge and information areshared. These networks are bound by trust rather thanhierarchical authority relationships found in verticallyintegrated forms of organization. Piore and Sabel pointto the Marshallian industrial districts of Italy as oneexample of flexible specialization in action.

Saxenian (1994) attributes Silicon Valleys success inthehigh-tech industry to itsadoptionof a network formof industrial structure. She observes that althoughRoute 128 in Bostonand Silicon Valley, California, were

both high-tech centers in electronics during the 1970s,onlySiliconValley emergedfroma regionalrecessioninthe 1980s to become an international leader in the high-tech industry. Her work examines the causes of thesedivergent growth trajectories and concludes that thedifferences have been primarily attributable to differ-ences in the nature of industry structure within the tworegions. Silicon Valleys industry was defined by dense

http://lastpage/


14/43


15/43


16/43


17/43

tors. Similarly, large vertically integrated firms mayemerge to monitor wage labor if contractswith externalsuppliers are not recognized.

5. EMERGING NEOCLASSICAL PERSPECTIVES

This overview of the theoretical literature onregional economic growth concludes with a discussion

of two new perspectives that attempt to address earliercriticisms of the neoclassical exogenous growth andtradetheories:endogenousgrowth theory and the neweco-nomic geography.

a) Endogenous Growth Theory

The new endogenous growth theories modifyassumptions of theexogenousgrowthmodelsto gener-atea range of economicpredictions, some ofwhichtendtoward economic divergence across regions. However,endogenous growth theory stays true to the neoclassi-cal traditionof general equilibrium modeling.The rootsof endogenous growth theory can be traced to early

work by Ramsey (1928), Cass (1965), Koopmans (1965),and Schumpeter (1947, 1934). Models by Cass (1965)and Koopmans (1965) adopt the utility function pro-posed by Ramsey to incorporate a savings rate that isdetermined by household choice, a feature that makessavings rates endogenous to the growth model. Undercertain conditions, the Ramsey-Cass-Koopmans modelpredicts conditional convergence. If the savings raterises with the capital/labor ratio, then the model pre-dicts a slower speed of convergence than the Solow(1956) and Swan (1956) model (Barro and Sala-i-Martin1999).

Other variants of the endogenous growth theories

make technological change and innovation endoge-nous to the model. Schumpeter (1947) was the first topoint out that the process of innovation is largely a raceformonopoly control over thestreamof rents from newinnovations, which are essentially public goods onceintroduced. Arrows (1962) learning-by-doingframework is within the Schumpeterian tradition byarguing that firms can gain monopoly power over newknowledge through experience in internal production.Innovations are modeled as declining costs that arefunctions of a firms previous investments. If a firm caninternalize these costs, they can gain a competitiveadvantage.

Romer (1986) relies on Arrows learning-by-doingframework to incorporate technical change as anendogenous parameter within a competitive equilib-rium model of economic growth. Romers model is

based on the crucial assumption that knowledge exhib-its increasing marginal productivity characteristics. Inother words, the production of consumption goods ismodeled with a production function that includes the

stock of knowledge and other inputs. This productionfunction assumes increasing returns to scale in the pro-duction of consumption goods, but decreasing returnsto scale in the production of new knowledge, a featurethat ensures mathematical tractability. In Romersmodel, per capita output may be persistently slowerover time insome countries than others; thus, themodeldeparts from the standard neoclassical exogenousgrowth model by predicting divergence in regionalgrowth rates. Furthermore, the equilibrium outcome isnot necessarily pareto optimal.

Other than an early regional growth model withagglomeration economies and endogenous technicalchange proposed by Harry Richardson (1973), modelsof endogenous growthhave only recentlybegunto con-sider the role of space and geography in shaping pat-terns of regionalgrowthand decline.Nijkamp andPoot(1998) extendthe Romer-Arrow frameworkto allowforspatial considerations such as factor mobility, the spa-tial diffusion of innovations, and interregional trade.

The authors demonstrate that when these spatial inter-actions across regions are incorporated into a regionalendogenous growth model, the empirical implicationsof themodel areindeterminate.Depending onthespec-ification of the model, absolute convergence, condi-tional convergence, and divergence are all theoreticalpossibilities.

Other recent theoretical models have extended theendogenous growth framework to account for theimpacts of infrastructure investment on regional pro-ductivity. Barro(1990) developsan endogenous growthmodelthat incorporates tax-financed public services.Inthis model, there is a nonlinear relationship between

public investment andprivate output. Increases in gov-ernment spending raise the marginal productivity ofcapital and labor, assuming a small to moderate ratio ofgovernment spending to total output. If this ratio

becomes too large, the distortional effects of taxationpredominate and lead to a declining growth rate.Aschauer (2000) extends Barros framework to investi-gate the relationship between changes in public capitalinvestment at the U.S. state level and statewide eco-nomic growth.

A special issue of theAnnals of Regional Science (1998)surveys recent developments in regional endogenousgrowth theory. The model discussed by Rosser in thisissue is unique, because it focuses on the impact ofinfrastructure on private sector coordination ratherthan complementarities between public and privatecapital investment. Drawing on interacting particlesys-tems theory, Rosser demonstrates that infrastructureinvestments, especially communications, logistical net-works, and transportation, allow private sector agentsto coordinate activities across space. However, the

http://lastpage/


18/43


19/43


20/43


21/43

captured by all regions in the national economy. Withthese learning-based externalities, there may also bepositive feedbacks that result in cumulative benefitsover time. Stiglitz (1989) eloquently describes thiscumulative process:

Assumethat thereare twogroups within thepopulation:

innovators and inventors. Inventors generate new ideas;innovators turn them into profitable businesses. Innova-tors search among inventors for new ideas. The moreinventors there are, the more it pays to be an innovator;andthe more innovators thereare, thegreaterthe returnsto invention. (P. 199)

Thus, regions with a large initial supplyof inventorswill tend to attract innovators, which in turn attractsmore inventors. This process closely resembles thedivergent cumulative process of regional developmentdescribed by Myrdal (1957).

A final characteristic of learning by doing that is

important for regional development is hysteresis.Here, the extent of learning by doing in any givenperiod may be determined by historical factors such aswars, plagues, or depressions that have permanenteffects on a regionsability to generate newknowledge.Similarly, a regions ability to learn by doing may bedetermined by its previous history of learning bydoing. For example, Richardson (1979) points out thatlearning maybe more likelyin large cities because citiestend to have a disproportionate share of innovation-adopting elites, defined as an entrepreneurial classthat is more receptive to new ideas and innovations.

The learning-by-doing phenomenon creates severaldilemmas for regional economic development plan-ners. First, even if knowledge diffused instantaneouslythroughout society once created, it still may beunderprovided, because it would not be economicallyrational for anyprivate inventor to investin thecreationof new knowledge if they immediately lost all gainsfrom knowledge creation. This is the market failureproblem created by the public good nature ofknowledge.

Even without the spatial problems discussed above,to increasetheprobabilityof knowledgegeneration, theplanner must develop a patentingsystemor other formof knowledge-generating incentive. Sincepatents effec-tively grant monopoly power to the inventor for adefined period of time, the planner is faced with anoptimization problem when deciding whether to issueincentives such as patents. He or she must choose aknowledge-generating institution that maximizes theexpected benefits from new knowledge, taking intoaccount the market distortions resulting from monop-

oly pricing by the innovating firm (Carlton and Perloff2000).

When the spatial dimension of knowledge genera-tionis introduced, theplanners decisionbecomesmorecomplex,because now, the choice of policy interventionhas implications for both national efficiency and inter-regional equity. With learning by doing, regional

growth is more likely to produce interregional percapita income divergence over time. Thus, even whenthe public good dilemma of knowledge generation has

been resolved in a region like Silicon Valley, the local-ized nature of knowledge diffusion suggests that the

benefits of the new knowledge may not extend beyondSilicon Valley. If there are increasing returns to scale innew knowledge generation, the emergence of a SiliconValley may also draw high-skilled labor away fromother regions in the United States, thus further exacer-

bating interregional inequities in economic growthalong with potential inefficiencies associated with theunderproduction of new knowledge in declining

regions.

Given the nature of this problem, a regional plannerin a declining region may proceed in one of three ways.First, he or she may acknowledge the market imperfec-tions associated with the public good characteristics ofknowledgeand work tocreate local institutions, such asfinancial incentives for innovations, business incuba-tors, or other local incentiveprograms, that increase theprobability of local innovation without creatinganticompetitive local monopolies. If successful, thismay restore the competitiveness of lagging regions andrestore trends toward per capita income convergence,

thus simultaneously accomplishing equity and effi-ciency objectives by diverting growth away from moredeveloped regions toward undeveloped regions.

A problem with this approach from Stiglitzs (1989)perspective is that the nonmarket institutions created

by less developed regions to resolve market failurestend to be less successful in accomplishing this objec-tive than the nonmarket institutions in more developedregions. Forexample, thediffusion of innovationsoftenrequires the presence of mature financial, managerial,and marketing firms. Even if firms have an incentive tocreatenew ideas in lagging regions, they mayhave littleaccess to the financial capital or the marketingresources required to bring their ideas to market. Simi-larly, the probability of interaction and knowledgesharing among local firms may be determined by thenumber of opportunitiesforlocalinteraction. Inregionswith declining downtowns and few venues for eating,drinking, andsocializing, workers maynot interact fre-quently enough to cultivate substantial local knowl-edge spillovers.

http://lastpage/


22/43

Another problem is that if financial incentivesoffered by local planners encourage newfirms to locatein an underdeveloped region, when they would havelocated in a more advanced region, then the underde-veloped region may be inefficiently supporting anindustry in a location where it is suboptimal for thatindustry to conduct business. For example, an automo-

bile manufacturing firm may be attracted to a region bygenerous property tax subsidies, only to discover uponlocating in that region that the regions naturalresources and labor skill level are poorly suited forautomobile production. This may, in turn, reduce theprobability of economic success for that firm, whichwould have negative long-term regional consequencesdue to job losses.

Another approach is to assume the role of a nationalpolicymaker and devise institutions that increase thegeographic spread of knowledge generation across allregions, thereby reducing the localization of knowl-edge generation. For example, a national policymaker

may publish information about local innovations in anoutlet that is accessible to firms in other regions. Theproblem with this approach is that it would likelydestroy the very incentive that leads to the creation ofknowledgeinthefirstplace: thepotential for monopolyprofits from new knowledge. If the strategy of geo-graphic dispersion of knowledge is chosen, it wouldonly be effective if such dispersal had no effect on themonopoly profits earned by the generators of newknowledge.

A final approachis to redistribute some of thewealthgains from rapidly growing regions experiencing newknowledge generation to other regions. This approach

is tantamount to intervening directly on equity ratherthan efficiency grounds. Bolton (1992) offers one of themostinteresting perspectiveson thisissue. The purposeof his article is to challenge an earlier article by LouisWinnick (1966), which criticized national policiesaimed at improving placesratherthan thepeople livingwithin the places. Winnick argued that such place-spe-cific redistributive policies are inappropriate, becausethey tend to result in the subsidized survival of high-cost firms in particular locations, tend to ascribe to anentire region the average characteristics of that regionwithout focusing on the internal heterogeneity of thepopulation, and tend to provide too much support tolocation-bound governments that cannot migrate toescape regional malaise. Bolton challenges Winnicksargument by pointing to specific instances when place-specific policies may be necessary for economic effi-ciency. This wouldbe thecase if places provide a pub-lic good valued by society. Regarding the nature of thispublicgood, Boltonpoints to theoption value of places,the pure existence value of places, and the correlation

between donor preferences for redistribution to recipi-ents in specific locations. For example, if entrepreneursin Silicon Valley see an option value, pure existencevalue, or have specificdonor preferencesfor redistribu-tionin regions outside of Silicon Valley, thenredistribu-tion on equity grounds may be required for efficiencyreasons (Bolton 1992).

The important point here is that although efficiencyand equity are often intertwined when we consider thespatial dimension of market failures resulting fromknowledge generation, the traditional approach toregional market intervention (i.e., Intervene on effi-ciency grounds and equity goals will be achieved auto-matically.) need not always be optimal. In fact, it may

be the case that direct interventionon equity grounds ismore likely to accomplish both efficiency and equityobjectives than is intervention on efficiency groundsalone, especially if the institutions in poverty-strickenregions are ill-equipped to resolve market failures andregional growth processes fail to conform to the stan-

dard neoclassical models.To conclude, the convergence-divergence debate isno longer simply an academic debate when viewed inlight of policy issues related to efficiency and equity. Ifone accepts the convergence hypothesis, then one canassume that lagging regions will tend to grow fasterand approach standards of living in developed regionsover time, and inequities will be resolved in the longrunsimplyby improving thefunctioning of themarket.If, on the other hand, there are substantial marketimperfections in regional trade and knowledge diffu-sion, as suggested by Stiglitz (1989), then market ineffi-ciencies will result in interregional inequities. The

appropriate strategy for improving interregional effi-ciencyand/orequitydepends on thenature of theorig-inal source of divergence and the benefits and costs ofdiverting the path of growth in the other direction.Either way, theory has much to say about theconsequences.

E. References

Carlton, Dennis W., and Jeffrey M. Perloff. 2000.Modern industrialorganization. Reading, MA: Addison-Wesley.

Domar, Evsey. 1946. Capital expansion, rate of growth, and employ-ment.Econometrica14, 2: 137-47.

Greenhut,Melvin L.1956.Plantlocationintheoryandinpractice. ChapelHill: University of North Carolina Press.

Harrod, RoyF. 1939.An essay in dynamic theory. Economic Journal 49,193: 14-33.

Isard, Walter. 1960. Methods of regional analysis: An introduction toregional science. Cambridge, MA: MIT Press.

Kondratieff, Nikolai D. 1935. Thelong waves in economiclife. Reviewof Economic Statistics17, 6: 105-15.

Ricardo, David. 1817. On the principles of political economy and taxation.London: John Murray.

http://lastpage/


23/43

Winnick,Louis. 1966. Placeprosperityversus peopleprosperity: Wel-fare considerations in the geographic redistribution of economicactivity. InEssays in urban land economics in honor of the sixty-fifthbirthday of Leo Grebler, Real Estate Research Program, University ofCalifornia at Los Angeles. Los Angeles: Real Estate ResearchProgram.

II. ANNOTATED BIBLIOGRAPHY

A. What Is a Region?

001. Fox, Karl A., and T. Krishna Kumar. 1994. The functionaleconomicarea: Delineation and implications for economicanalysis and policy. InUrban-regional economics, social sys-tem accounts, and eco-behavioral science: Selected writings ofKarl A. Fox. James R. Prescott, Paul van Moeskeke, andJatiK. Sengupta, eds. Ames: Iowa State University Press.

In this chapter, the authors discuss their functional eco-nomic area concept. Within a functional economic area,the majority of households commute to jobs within theregion, and the majority of goods sold are consumedwithin the region. Functional economic areas, much like

the metropolitan statistical areas identified by the U.S.Census Bureau, are defined by the spatial extent of house-hold commuting areas (Richardson 1978). The UnitedStates Department of Commerce Bureau of EconomicAnalysis (BEA) currently relies on a version of thisapproach to define BEAeconomic areas within the UnitedStates.

002. Hoover, Edgar M., and Frank Giarratani. 1985.Introduc-tion to regional economics. 3d ed. New York: Knopf.

This classic text gives a nice overview of the conceptualfoundations of regional economic development theoryandurban land usetheory. Chapter9 of this book examinesseveral approaches to defining the concept of a region.

This entire text can be accessed online at the followingURL: http://www.rri.wvu.edu/WebBook/Giarratani/main.htm.

B. Conceptual Foundations of RegionalEconomic Development Theory

1. THE INTERREGIONALCONVERGENCE HYPOTHESIS

003. Balassa, Bela. 1961. The theory of economic integration.Homewood, IL: Irwin.

In this book, Balassa examines the integration of nationaleconomies with particular intereston European countries.Chapter 4 of this book provides a nice overview of the

Heckscher-Ohlin theorem and its implications for Euro-peanintegration.He alsoexamines therole of factor mobil-ity in the theory of economic integration.

004. Heckscher, Eli F. 1919. The effect of foreign trade on thedistribution of income.Ekonomisk Tidskrift21: 497-512.

005. Ohlin, Bertil. 1933. Interregional and international trade.Cambridge, MA: Harvard University Press.

Together, these two works establish the well-knownHeckscher-Ohlin theorem, which essentially states thatgiven two countries, two export goods, and two factors ofproduction,a factor-abundantcountrywill havea compar-ative advantage in the production of goods that requirethat factor. Therefore, the country will specialize in, andexport that good, importing goods for which productionfactors are scarce.

006.Salvatore,Dominick. 1998. International economics.6thed.Upper Saddle River, NJ: Prentice Hall.

This international economics text provides an excellentoverview and analysis of the Heckscher-Ohlin-Samuelsontheorem and the implications of this models assumptionsfor international trade and economic growth.

007. Samuelson, Paul A. 1953. Prices of factors and goods ingeneral equilibrium. Review of Economic Studies 21, 1 (Octo-

ber): 1-20.

008.. 1949. Internationalfactor-price equalization onceagain.Economic Journal59, 234: 181-97.

009.. 1948. International trade and the equalization offactor prices.Economic Journal58, 230: 163-84.

These three articles extend the Heckscher-Ohlin theoremto demonstrate the now-famous result: free trade and/orthe mobility of goods serves to equalize the relative andabsolute prices of factors of production across those coun-tries engaged in trade in the long run.

2. LOCATION THEORY AND REGIONAL SCIENCE

010.Alonso, William.1975. LocationTheory. InRegionalpolicy:Readings in theory and applications, John Friedmann andWilliam Alonso, eds. Cambridge, MA: MIT Press.

Thisarticlegives an overviewof thefundamental conceptsfrom location theory initially developed by Alfred Weber(1929) and subsequently expanded by Edward Hoover(1937),Walter Isard (1956),andMelvinGreenhut (1956).By1975, the foundations of location theory, as it is reviewed

by Alonso, had become well established. Additional com-plexities examined by Alonso include the influence ofassemblycosts anddistributioncosts, therole of transship-mentpoints, andthe implicationsof multiplerawmaterialsources and irregularly shaped transportation networksand market areas on optimal plant location. Alonso con-cludes the article by pointing out the limitations of thelocation theory approach, including its static orientation,itsignorance ofexternalscale economies, andthetendency

of early theories to treat demand as exogenous.

011. Isard, Walter. 1956. Location and space-economy. Cam-bridge, MA: MIT Press.

A comprehensive exposition of traditional location theoryas originally developed by Alfred Weber (1929) and EdgarHoover (1937).Theworkalso incorporatesthe market areaconcept developed by Christaller (1933) and Losch (1954)as well as the relationship between location theory and

http://lastpage/


24/43

international trade theory. Chapter 10 provides a mathe-matical formulation of Isards general theory of location.

012. Isard, Walter, Iwan J. Azis, Matthew P. Drennan, RonaldE. Miller, Sidney Saltzman, and Erik Thorbecke. 1998.

Methods of interregionaland regionalanalysis. Brookfield,VT:Ashgate.

This book is Walter Isards sequel to the seminal regional

science text,Methods of regional analysis: An introduction toregional science(1960). The text presents a summary of thestandard methodological approaches to analyzingregional economies. Included in the discussion are tradi-tional techniques suchas comparativecostanalysis,input-output analysis, linear and nonlinear programming, andgravity and spatial interaction models in addition to morerecent methods of analysis including spatial econometrics,computable general equilibrium models, and spatialmicro-simulation.

013.Weber, Alfred.1929.Theory of the locationof industries. Chi-cago: University of Chicago Press.

Thebeginnings of locationtheorycan betracedto this clas-

sic by Alfred Weber. In essence, this work is an attempt toincorporate transportation costs intothe theory of thefirm.In Webers theory, transportation costs are influenced byweight and distance. He assumes that demand is givenand that firms seek to minimize the costs of transportingraw materials and final goods to the market. If naturalresources and raw materials areconcentratedin particularlocations, andthe weightof theraw material is heavy rela-tive to the weight of the final product, the firms produc-tion activities are weight losing. These firms will locatenearinput sources,while weight-gainingfirms willtendto locate near consumer markets. This simple idea has

become the foundationof modernlocationtheory. Chapter5 of Webers book also contains an early discussion of

agglomeration economies.

3. EXTERNAL ECONOMIES

014. Hoover, Edgar M. 1937.Location theory and the shoe andleather industry. Cambridge, MA: Harvard UniversityPress.

In chapter 6 of this classic work, Edgar Hoover developshis well-known typology of agglomeration economies inresponse to the oversimplification of this concept inWebers (1929)work.Hoover argues that thebenefits fromagglomeration include (1) large-scale economies resultingfrom traditional economies of scale; (2)localization econo-mies that result from the firms in the same industry

colocating in the same area; and (3)urbanization economies,which result from the colocation of firms in differentindustries.

015. Marshall, Alfred. [1890] 1961.Principles of economics: Anintroductory volume. 9th ed. Reprint, London: Macmillan.

In this classic text, Alfred Marshall became the first tomake the distinction between internaland externalscaleeconomies. Internal scale economies are generated inter-nally by individual firms that either expand their internalscale of operations to minimize average production costs(traditional economies of scale). External scale economiesrefer to declines in firm-level averagecosts thatresult fromincreases in industry-wide output within a given region.

Examplesof externalscale economies includethespilloverbenefits that result from a specialized local labor force, alarge diverse supply of input providers, and informationsharing among firms within a region. The metaphor of theMarshallian industrial district, fueled by these externalscale economies, is a keyconcept in most recent theoriesofregional economic development. It is this emphasis onlocal external scale economies that also differentiatesmany theories of regional economic development fromtheories of national economic development.

4. MODELS OF SPATIAL COMPETITION

016.Devletoglou, Nicos E. 1965. Adissenting viewof duopoly

and spatial competition.Economica32, 126: 140-60.This article extends Hotellings (1929) spatial pricingapproach by introducing realism into the model. Theextensions examined by Devletoglou include consumerdemand uncertainty and elasticity, the threat of entry bycompetitors, and an additional spatial dimension to con-sumer market areas. (In the Devletoglou model, consum-ers are distributed along a plane rather than along a linesegment as in the Hotelling model.) With these additionalassumptions, Devletoglou demonstrates that under cer-tain conditions, firms will tend toward dispersion ratherthan concentration, as in the Hotelling model. In theDevletoglou model, the threat of entry drives all profits tozero.

017. Eaton, B. Curtis, andRichard G. Lipsey. 1978. Freedom ofentry and the existence of pure profit.Economic Journal88,351: 455-69.

This article extends the model by Devletoglou (1965) todemonstrate how a spatial equilibrium distribution offirms need not necessarily driveprofits to zero. Accordingto Eaton and Lipsey, there is no incentive for new firms toenter, because for newfirms, all costs arevariable, whereasfor existing firms, all fixed costs are sunk and therefore donot bear on current production decisions.

018. Hotelling, Harold. 1929. Stability in competition. Eco-nomic Journal39, 153: 41-57.

Thisclassic of the industrial organization literature has ledto the emergence of a large literature that examines howspace affects the pricing behavior of firms. Although notalways identified as part of locationtheory perse, theliter-ature on spatial competition that begins with Hotelling

http://lastpage/


25/43


26/43


27/43


28/43

sisontherole of innovationinmaintainingthespatial core-peripherypattern and discusses the role of social relationsin legitimizing the innovations adopted by the core.Friedmann also points out that while the core-peripherypattern is relatively stable, the tensions in the patterns ofauthority-dependency relationships may eventually pro-duce undesirable social outcomes.

043.

. 1966.Regional development policy: A case study ofVenezuela. Cambridge, MA: MIT Press.

Friedmanns center-periphery model is largely based onGunnar Myrdals (1957) theory of unbalanced regionalgrowth, but it also incorporates elements of export basetheory and a discussion of the role of local leadership indetermining a regions growth trajectory.

044. Hermansen, Tormod. 1972. Development poles and de-velopment centres in national and regional development.In Growth poles and growth centres in regional planning,Antoni Kuklinski, ed. Paris: Mouton.

This chapter gives a thorough and critical review of thediverse growth pole literature. It also makes useful link-ages between growth pole theory and related works incumulative causation theory andcentral place theory. Thiswork,in addition tothe 1969workby Darwent,offers earlycriticisms of a theoretical perspective that was at thetime awell-accepted paradigm among regional developmenttheorists.

045. Higgins, Benjamin. 1983. From growth poles to systemsof interaction in space.Growth and Change14, 4: 1-13.

In this article, Higgins reexamines the growth pole litera-ture in light of its criticisms that emerged in the 1970s andthe reemergence of the termin the early 1980s. Among themajor criticisms discussedby Higginsarethe lackof coher-

ence between traditional notions of growth poles andempirical reality, the inappropriate use of input-outputanalyses to examine the spatial interactions among firms,and the difficulties of translating Perrouxs (1950) originalabstract formulation into useful theories of regional eco-nomic development.

046. Hirschman, Albert O. 1958. The strategy for economic devel-opment. New Haven, CT: Yale University Press.

The author offers an alternative to the prevailing balancedgrowth paradigm,also referred to as thetheory of the bigpush(p. 51),whichessentially states thateconomic devel-opment requires the simultaneous development of a largenumber of newindustries to support the linkagesrequired

for steady growth. Hirschman argues instead for a theoryof economic development based on the concept of back-ward andforward linkages among firms.Fora given firm,

backwardlinkages referto the inputsof other firms used inthe production process, whereas forward linkages refer tothe output of the firm that is then used by other local pro-ducersas an intermediate good forthe production of otherproducts (p. 100).

047. Lasuen, Jose Ramon. 1972. On growth poles. In Growthcenters in regional economic development, Niles M. Hansen,ed. New York: Free Press.

This chapter is a reinterpretation of Perrouxs (1950) initialpolarization concept in light of more recent policiesdesigned to promote the development of growth poles.Lasuen argues that due to recent transportation innova-tions and inter- and intrafirm organizational changes,

growth and development haveresulted in less geographicpolarization than in the past (p. 34). Lasuen recommendsthatdeveloping regions adopt policies to foster these link-ages, pointing to the case of Japan as an example. Thischapter points to other possible links between the growthpole literature and the network/flexible specialization lit-erature that would eventually emerge in the 1980s.

048.Perroux, Francois. 1950. Economicspace: Theory and ap-plications.Quarterly Journal of Economics64, 1: 89-104.

This is theseminal work in thegrowth pole/growthcentertradition. Although Perrouxs initial formulation of thetheory only treats space in the abstract, his ideas about the

spatial relationships between economic agents were even-tually developed by Boudeville (1966) and others into aspatial account of economic growth through polarizingforces among interdependent firms and industries in par-ticular locations. Perroux defines three ways to definespace:space as defined bya plan, space defined asa sumofforces, and space as a homogeneous aggregate (p. 94). Hisarticle explores each conceptualization of space by exam-ining the spatial dimension of monetary flows andnational sovereignty. The concept of the growth pole hascome to adopt the space as force view of spatial interac-tion, which defines space as a type of network that is heldtogetherby centripetal forces. Although this view of spaceis not unlike that which is advocated by those in the flexi-

ble specialization/network theory tradition, the two theo-retical perspectives have largely developed in isolationfrom one another.

049. Thomas, Morgan. 1972. The regional problem, structuralchange, and growth poletheory. In Growth poles and growthcentres in regional planning, Antoni Kuklinski, ed. Paris:Mouton.

This chapter seeks to address the failure of growth pole-inducedpolicyprescriptions toresult ina significant rever-sal of unbalanced growth trends. The failures of growthpole theory appear

regional development theory

Documents