Cluster policy in developing countries:The ugly, the bad, and the not-so-bad

Gilles DurantonUniversity of Toronto

March 2007

Motivation

• Pervasiveness of ‘cluster initiatives’ in developedcountries

• Many attempts in developing countries

• Objective:

Are cluster initiatives a good idea in developingcountries?

2

Main message

• Industrial policy is regional policy

• But doing ‘activist’ industrial policy locally is not a goodidea...

• ... especially in developing countries

• However, letting places specialise more should providebenefits

"Don’t actively pursue clusters but don’t prevent fromhappening (and possibly respond to the needs of existingones)"

3

Background 1: Specialisation / Localisation ofindustries

• Cities and regions do specialise

• Industries have a tendency to agglomerate (Ellison andGlaeser, 1997; Duranton and Overman, 2005)

– Overwhelming evidence of developed economies

– More sketchy for developing countries (Henderson,2005)

• There are benefits to agglomeration / localisation but wedon’t know how they percolate (Rosenthal and Strange,2004)

4

Background 2: Regional disparities

• Regional disparities are pervasive

– In developed countries

– Much larger in developing countries (Aten andHeston, 2005; Milanovic, 2004)

• Rising in many countries

5

From here:

• Two temptations

– Use industrial policy to reduce regional disparities

– Do industrial policy locally

• One thing to keep in mind

– Industrial policy will have regional implications

6

Cluster initiatives

• Originating from Porter (1990; 1998; 2000a; 2000b)

• Thousands of initiative worldwide including hundredsin developing countries (Sölvell, Lindqvist, and Ketels,2003; Ketels, Lindqvist, and Sölvell, 2006)

• Porter’s definition: "A cluster is a geographicallyproximate group of interconnected companies andassociated institutions in a particular field, linked bycommonalities and complementarities."

7

Why cluster initiatives?

• The short answer: "Improve competitiveness"

• Porter’s longer answer: The ’diamond’

8

Porter’s diamond

graphically depicted in a diamond; the diamond metaphor has become common in referring to thetheory (Figure 2).8 In a recent two-part article, I explore and statistically test the sequential processby which the diamond must upgrade if an economy is to advance (Porter, 1998b). Parallel improve-ments in the sophistication of company operations and strategies and the quality of the diamondprovide the microeconomic foundations of economic development.

A few elements of this framework deserve highlighting because they are important to under-standing the role of clusters in competition. Factor inputs range from tangible assets such as physi-cal infrastructure to information, the legal system, and university research institutes that all firmsdraw on in competition. To increase productivity, factor inputs must improve in efficiency, quality,and (ultimately) specialization to particular cluster areas. Specialized factors, especially thoseintegral to innovation and upgrading (e.g., a specialized university research institute), not only arenecessary to attain high levels of productivity but also tend to be less tradable or available fromelsewhere.

The context for firm strategy and rivalry refers to the rules, incentives, and norms governing thetype and intensity of local rivalry. Economies with low productivity are characterized by little localrivalry. Most competition, if present at all, comes from imports. Local rivalry, if occurring at all,involves imitation. Price is the sole competitive variable, and firms hold down wages to compete inlocal and foreign markets. Competition involves minimal investment.

Moving to an advanced economy requires that vigorous local rivalry develop. Rivalry must shiftfrom low wages to low total cost, and this requires upgrading the efficiency of manufacturing andservice delivery. Ultimately, rivalry also must evolve from cost to include differentiation. Competi-tion must shift from imitation to innovation and from low investment to high investment in not onlyphysical assets but also intangibles (e.g., skills, technology). Clusters, as will be evident, play anintegral role in these transitions.

20 ECONOMIC DEVELOPMENT QUARTERLY / February 2000

Parallel improvements inthe sophistication ofcompany operations andstrategies and the qualityof the diamond providethe microeconomicfoundations of economicdevelopment.

Figure 2: Sources of Locational Competitive Advantage

9

The problems with Porter’s diamond

• Apparent complexity but structurally simple

• It is not a theory

• It does not make the case for cluster policies

10

The simplest cluster model

Assume

• Uncompensated external effects in production leading tolocal increasing returns

• Congestion costs

=> Real wages are bell-shaped as a function of clusteremployment

11

Z

ZN

Net income

Population

12

Lessons from the simplest cluster model

The two main inefficiencies associated with clusters

• The cluster co-ordination failure

– Clusters tend to be ‘too big’

– => Restrict the size of existing clusters

– => Create new ones

• Uncompensated externalities in production

– External effects at the root ofagglomeration/clustering also make productioninefficient in clusters

– Need to fix inefficiencies in production

13

Hence cluster policies should do two things

• Solve the co-ordination failure and impose optimalcluster size(but what is it?)

• Fix inefficiencies in production(but we don’t even know which market failures reallymatter)

• Both sets of policies are needed

14

• These recommendations seem very hard to implement inpractice

• Considering a more sophisticated cluster modelcomplicates the policy recommendation further

– Imperfect mobility of workers

– Imperfect mobility of knowledge

– Market failures on the cost side (congestion, etc)

– Cluster dynamics and industry mobility

– Political economy

15

• The magnitude of localisation economies does not justifythe large investments recommended by clusterproponents

16

Conclusions

• Do industrial policy (maybe) but do it nationally

• Keep in mind that this will have regional implications

• Not spending large amounts on cluster initiatives doesnot mean that specialisation / localisation should bediscouraged

• Not doing cluster policies does not mean that localgovernments should do nothing

17

For more on cluster initiatives:

Duranton, Gilles. 2007. California Dreamin’: The feeble casefor cluster policies. Chapter part of a forthcoming CEPRspecial report on clusters.

Available at:

http://individual.utoronto.ca/gilles/Papers/Cluster.pdf

18

References

Aten, Bettina and Alan Heston. 2005. Regional output differences ininternational perspective. In Ravi Kanbur and Anthony J. Venables(eds.) Spatial Inequality and Development. New York: OxfordUniversity Press.

Duranton, Gilles and Henry G. Overman. 2005. Testing for localisationusing micro-geographic data. Review of Economic Studies72(4):1077–1106.

Ellison, Glenn and Edward L. Glaeser. 1997. Geographic concentrationin US manufacturing industries: A dartboard approach. Journal ofPolitical Economy 105(5):889–927.

Henderson, J. Vernon. 2005. Urbanization and growth. In PhilippeAghion and Steven N. Durlauf (eds.) Handbook of Economic Growth,volume 1B. Amsterdam: North-Holland, 1543–1591.

18-1

Ketels, Christian, Göran Lindqvist, and Örjan Sölvell. 2006. Clusterinitiatives in developing countries and transition economies. Stockholm:Center for Strategy and Competitiveness.

Milanovic, Branko. 2004. Half a world: Regional inequalities in fivegreat federations. Processed, World Bank and Carnegie Endowmentfor International Peace.

Porter, Michael E. 1990. The Competitive Advantage of Nations. NewYork: Free Press.

Porter, Michael E. 1998. Clusters and the new economics ofcompetition. Harvard Business Review 76(6):77–91.

Porter, Michael E. 2000a. Location, competition, and economicdevelopment: Local clusters in a global economy. EconomicDevelopment Quarterly 14(1):15–34.

18-2

Porter, Michael E. 2000b. Locations, clusters, and company strategy. InGordon L. Clark, Maryann P. Feldman, and Meric S. Gertler (eds.)The Oxford Handbook of Economic Geography. New York: OxfordUniversity Press.

Rosenthal, Stuart S. and William C. Strange. 2004. Evidence on thenature and sources of agglomeration economies. In VernonHenderson and Jacques-François Thisse (eds.) Handbook of Regionaland Urban Economics, volume 4. Amsterdam: North-Holland,2119–2171.

Sölvell, Örjan, Göran Lindqvist, and Christian Ketels. 2003. The ClusterInitiative Greenbook. Gotheburg: Competitiveness Institute(TCI)/Vinnova.

18-3