recovery of a mexican cluster: devaluation bonanza or collective efficiency?

Recovery of a Mexican Cluster: Devaluation Bonanza

or Collective E�ciency?

ROBERTA RABELLOTTI *

University of Piemonte Orientale, Italy

Summary. Ð Mexico, as many other developing countries in Latin America and else-where, has been moving in the 1980s toward a liberalized trade regime after a longperiod of import substitution. This paper analyzes the impact of trade liberalization onthe cooperative behavior of shoe ®rms located in a cluster in Guadalajara. The empiricalevidence shows that cooperation has increased. It also suggests that cooperation posi-tively in¯uences ®rms' performance and together with a favorable market environmentcontributes to the cluster's recovery. The study is based on ®eldwork carried out inGuadalajara in 1996. Qualitative information was collected through in-depth interviewsand quanti®ed responses came from a questionnaire survey covering a sample of 63 shoemanufacturing enterprises. Ó 1999 Published by Elsevier Science Ltd. All rightsreserved.

1. INTRODUCTION

The literature on industrial districts inadvanced and less developed countries hasshown that clustering helps local enterprisesovercome growth constraints and compete indistant markets.1 In this paper we explore thequestion of whether ®rms derive some advan-tages from clustering to face turning points.2

More precisely, we investigate the ability ofclustered ®rms to shift gears in the face of thechallenges posed by trade liberalization. Ourresearch proposition is that (raising) coopera-tion among ®rms belonging to a cluster isessential for them to be able to compete in thenew market environment. This hypothesis isinvestigated in the Mexican footwear cluster ofGuadalajara.

Mexico, as many other developing countriesin Latin America and elsewhere, has beenmoving in the 1980s toward a liberalized traderegime after a long period of import-substitut-ing industrialization. The ®rst stage in importliberalization came in July 1985 when licenseswere eliminated for almost 3,600 items. In 1986,with Mexico's accession to the GATT, thecountry committed itself to eliminate all o�cialimport prices by the end of 1987. In December1987 the Economic Solidarity Pact was laun-ched bringing government, entrepreneurialorganizations and labor unions to an agreed

position to accelerate trade reforms. The resultwas a tari� structure with ®ve levels of 0-20%ad valorem taxes. With these tari� reductions,the Mexican trade reform was completed (TenKate, 1992).

This reform had a signi®cant impact onMexico's footwear sector. From June 1985 toDecember 1988 the value of domestic produc-tion covered by import licensing fell from 99.1to 0% and the average tari� decreased from46.8 to 18.1%. In response, imports grew from0.2 million pairs in 1987 to 107 million pairs in1991. This dramatic increase in imports wasaccompanied by a contraction in domesticproduction from 245.2 to 199.6 million pairs. Inthis new competition, Mexican shoe manufac-

World Development Vol. 27, No. 9, pp. 1571±1585, 1999Ó 1999 Published by Elsevier Science Ltd. All rights reserved

Printed in Great Britain0305-750X/99/$ - see front matter

PII: S0305-750X(99)00071-6www.elsevier.com/locate/worlddev

* This paper is part of an IDS research project directed

by Dr. Hubert Schmitz and ®nanced by the Department

for International Development, London. I am grateful

to participants in seminars held at the Institute of

Development Studies (Sussex) in April 1997 and at

Bocconi University (Milan) in September 1997 for their

helpful comments. Paolo Giudici helped me with statis-

tical analysis and Khalid Nadvi, Hubert Schmitz and an

anonymous referee made valuable comments on a

previous draft. Finally, I have a debt of gratitude to

the Camara de la Industria del Calzado and many

entrepreneurs in Guadalajara. The usual disclaimer

applies.

1571

turers were initially ill-equipped to competewith imports on price, quality and fashioncontent.

A recovery from the crisis in the shoe sectorcame in 1993 when tari�s on imports fromChina were increased. The footwear industryalso took advantage of the 1994 peso devalua-tion, boosting exports and increasing importprices. Finally, in 1995 the governmentincreased tari�s on imported shoes from therest of the world.

The return of (at least partial) importprotection, together with the peso devaluation,are commonly seen as the basis of recovery forthe Mexican shoe industry. Can this recoverybe entirely explained as a windfall gain fromchanges in the trade regime? Or are therestructural changes in the footwear industrywhich have enhanced the ability of domesticproducers to compete? Is the industry's recov-ery the outcome of a process of restructuring orthe result of devaluation and return to protec-tion?

This paper addresses these questions focusingon structural changes in vertical and horizontalrelationships between Mexican shoe ®rms andtheir suppliers, subcontractors, buyers,competitors and supporting institutions. Theserelationships are investigated in the footwearcluster of Guadalajara. Our objective is toassess if trade liberalization induces greatercooperation in vertical and horizontal linkages.Our empirical evidence shows that cooperationhas indeed increased. We also suggest thatcooperation positively in¯uences ®rms' perfor-mance and together with a favorable marketenvironment contributes to the cluster's recov-ery. Furthermore, within the cluster heteroge-neity is also increasing because only some of the®rms enter into cooperative actions.

The study is based on the ®ndings of ®eldwork carried out in Guadalajara, one of the twomain Mexican shoe clusters3, in August andSeptember 1996. Qualitative information wascollected through in-depth interviews with shoeentrepreneurs, suppliers, buyers, trade organi-zations and other key informants, and aquantitative survey conducted covering asample of 63 shoe manufacturing enterprises.4

Furthermore, the study also draws on primary®eld data from 30 ®rms collected in 1991, priorto the results of the liberalization programmaterializing5 (Rabellotti, 1997).

The following section discusses the impact oftrade policy reform in clusters. Then, in a briefhistorical overview of the Guadalajara cluster,

its main turning points and its recent growthrecord are presented. The changes that haveoccurred in backward, forward and horizontallinkages are discussed in Section 4. Section 5uses regression analysis to test the relationshipsbetween ®rm performance and cooperation andSection 6 investigates if size plays a role in thecooperation behavior of the sample ®rms. The®nal section summarizes the main ®ndings.

2. TRADE LIBERALIZATION ANDCLUSTERS

Trade policy has long been acknowledged ashaving a major in¯uence on the process ofindustrialization in developing countries. Thecase for trade policy reform is extensivelydiscussed in the economic literature.6 Here weneed only to recall that according to orthodoxneoclassical theory trade liberalization isexpected to produce both static and dynamicbene®ts. The static argument refers to thereduction of ine�ciencies arising from resourcemisallocation. The dynamic e�ciency gains areexpected to come from greater capacity utili-zation, greater specialization, enhanced tech-nological capabilities, greater learning bydoing, higher process and product innovation.A representative statement can be taken fromBalassa (1988, p. 45):

The carrot and stick of competition gives inducementfor technological change. For one thing, in creatingcompetition for domestic products in home markets,imports provide incentives for ®rms to improve theiroperations. For another thing, in response to compe-tition in foreign markets, exporting ®rms try to keepup with modern technology in order to maintain orimprove their market position.

Although this view became conventionalwisdom, the empirical foundations of sucharguments regarding the dynamic bene®ts ofliberalization are not very clear. Rodrick (1995)presents a survey of empirical studies at ®rm,industry and country level, emphasizing thatthere are problems of measurement and direc-tion of causality.7

On the contrary, a case in which tradeliberalization generates unambiguous positivedynamic bene®ts is in presence of excessdemand, when ®rms have no need to increasedemand for their product by improving it. Thisis what happened in the Mexican shoe industry.Before trade policy reform, for many decadesthe footwear manufacturers took advantage of

1572 WORLD DEVELOPMENT

a captive market where there was excessdemand; making money in the sector was easybecause every kind of product was sold, nomatter what its quality, design and cost. Tradeliberalization increased the incentive for intro-ducing product and process innovation,improving quality, increasing productivity andlowering costs (Rabellotti, 1997; Woodru�,1997).

In this paper, we test the hypothesis that inthe Guadalajara cluster trade policy reform hasalso had the e�ect of increasing cooperationamong shoe manufacturers, with their suppli-ers, subcontractors and buyers. How is thisincrease in cooperation related with thedynamic e�ects of liberalization, predicted byorthodox economic theory?

In the small but growing literature onindustrial clusters in developing countries,cooperation, together with external econo-mies,8 represent those collective e�ects whoseinterplay is supposed to bring about e�ciencygains for ®rms in the cluster and to increasetheir capability to innovate and grow (Rabel-lotti, 1997).

As regards trade liberalization and itsdynamic bene®ts, we saw that in presence ofexcess demand the opening up of the marketincreases the incentive for innovating, improv-ing quality, reducing costs. Furthermore, weknow that in a cluster cooperation among ®rmscan induce innovation and growth. Therefore,this paper provides an empirical test, for thecase of a Mexican cluster, of the hypothesisthat trade policy reform brings about morecooperation among ®rms and that this produ-ces dynamic e�ciency gains.

3. THE SHOE CLUSTER OFGUADALAJARA

(a) A brief historical overview

Guadalajara, capital of the state of Jalisco,located north-west of Mexico City, is thesecond largest city in the country9 and the thirdmost important industrial center after MexicoCity and Monterrey. The economy of Guadal-ajara is traditionally characterized by a highpresence of small ®rms in sectors such as food,textiles and shoes (Arias, 1985). Since the lastcentury, there has been a local tradition of shoeproduction in small workshops to satisfy localdemand. According to Arias (1992), in the1920s the footwear industry was the most

important industrial sector in Guadalajara:there were 34 plants producing shoes, onemaking lasts, one tannery and about 100 smallshoe workshops. From the 1930s, the footwearindustry began to develop rapidly helped by thepresence of a US enterprise, United ShoeMachinery, which lent machinery to local ®rmsand contributed in spreading technical know-how within the cluster. During the 1940s theshoe sector continued to expand, generatingalso a considerable increase in the number oflocal suppliers. At that time, Guadalajara wasalready well known in the rest of the countryfor being one of the main centers for shoeproduction.

In the 1950s the industry experienced a boomwith continuous expansion in the domesticmarket and some occasional export initiatives,mainly to the United States. In 1959, the localCamara del Calzado created a National Foot-wear Trade Fair in Guadalajara, contributingto the sector's growth. Growth acceleratedduring the 1960s and the 1970s, thanks to risingdomestic purchasing power, fast populationincrease and protection of the domestic market.

During the 1980s, the industry began tosu�er from the decline of domestic purchasingpower, but the main negative shock came fromthe opening up of the market in 1988 whentari�s on imports were cut and all importlicenses eliminated (Dominguez-Villalobos andGrossman, 1992). Imports, which accountedfor less than 0.1% of the domestic market in1987, increased substantially and by 1991accounted for nearly a third of domestic sales.The ¯ood of imports caused a profound crisisin the cluster: many ®rms went bankrupt,others drastically reduced production.

Following a series of events there haverecently been some signs of recovery fromimport liberalization. In 1993 the Mexicangovernment placed anti-dumping tari�s (vary-ing from 160 to 1000%, according to the type ofshoes) on imports from China. In 1994 the pesowas devalued. Finally in 1995 the tari� on shoesimported from the rest of the world was putback at 35% for a period of ®ve years. Thiscombination of events seems to have providedthe breathing space and time for Mexican shoe®rms to restructure.

(b) Performance and economic structure of thecluster

This section analyzes the most recent years,with a particular emphasis on changes since the

RECOVERY OF A MEXICAN CLUSTER 1573

turning point, that is, the 1988 liberalization ofthe domestic market. Because some of thechanges are not unique to Guadalajara, wepresent some empirical evidence on the Mexi-can shoe industry in general as well as somespeci®c data on the cluster under analysis.

In Table 1 we ®nd con®rmation of the majorturning point that occurred in 1988, indicatedby a huge increase in imports during 1988±92and by a decrease in the total number of pairsproduced. This table also shows a substantialincrease in exports in the 1990s, with theexception of 1993 and 1994 which were themost severe years of crisis for the industry andfor the Mexican economy as a whole. Exportsagain accelerated after the devaluation in 1994.On the import side, there was a continuousdecline since 1993 when tari�s on Chinese shoeswent up.

While the impact of the 1988 turning point isvery evident from the data presented, the morerecent recovery, facilitated by the increase intari�s and the peso devaluation, is less clearlydetectable from aggregated data. There is anincrease in exports, but the decrease in outputseems to continue, although at a decreasingrate. This may be due to time lags between thechange in external circumstances and the reac-tion of local producers. Furthermore, thedecreasing number of pairs could be compen-sated by an increase in value added.

As regards the structure of the Mexicanfootwear industry, according to the 1994Industrial Census the sector is composed of4,981 enterprises with a total employment of80,971 people. Of this total 82% of ®rms aresmall, employing fewer than 15 persons. In

Guadalajara, according to estimates by thelocal trade association, the number of footwear®rms in 1993 was about 1,100 with 25,000-oddemployees, producing 27% of all footwearmade in Mexico (ITESM, 1995). Unfortunatelymore recent data are not available, butaccording to the director of the local Camaradel Calzado there has been a substantialdecrease in the number of ®rms over 1993±95(interview with Ricardo Orta, September 1996).Other informants con®rmed the decliningnumber of ®rms, stressing that small and micro®rms as a size group were most severely hit bythe crisis. Considering only the members of theCamara, in Guadalajara the member ®rmswent down from more than 500 to 315.

The rest of the paper concentrates onGuadalajara and analyzes changes in back-ward, forward and horizontal linkages as aconsequence of the turning point.

4. LIBERALIZATION AND COLLECTIVEEFFICIENCY

This and the next two sections are based onthe ®eldwork carried out in Guadalajara.Information was collected through in-depthnetwork case studies, interviews with keyinformants and a questionnaire survey coveringa sample of 63 shoe ®rms.

For the network case studies, six enterprisesof di�erent size were selected (two with lessthan 20 employees, two with more than 20 andless than 70 and two with more than 70). Foreach size class, with the assistance of the localtrade association,10 one enterprise known for

Table 1. The Mexican footwear industry, 1970±96 a

Years Total n° of pairsproduced (millions)

Exports (US$ml.) Export year growthrate 1990±96 (%)

Imports (US$ml.) Import year growthrate 1990±96 (%)

1970 n.a. 3.4 15.51975 n.a. 1.1 20.81980 n.a. 31.0 62.01985 232.6 27.3 15.71988 245.2 68.2 54.31990 208.5 77.4 127.81991 199.6 126.5 63.4 189.9 48.61992 193.3 160.6 26.9 213.5 12.41993 173.3 150.3 ÿ6.4 172.7 ÿ19.11994 172.4 156.4 4.0 145.2 ÿ15.91995 167.0 201.9 34.8 87.4 ÿ39.81996 n.a. 355.3 68.5 74.0 ÿ15.3

a Source: CANAICAL (1996).


being dynamic and innovative and the othermore conservative were picked. Each of theenterprises selected was also asked for thenames of their main suppliers and buyers, whowere there interviewed to get their perspectiveon the vertical relationship.

As regards the sample survey, 63 enterpriseswere randomly taken from the registry ofmembers of the local entrepreneurial associa-tion.11 The sample is strati®ed by size: 20enterprises employ 70 or more employees, 21between 21 and 69 employees and the remain-ing 22 employ 20 or fewer employees.

In designing the questionnaire, the initialobjective was to observe changes in coopera-tion over time.12 A pre-test however, revealedthis to be infeasible in Guadalajara. We,therefore, collected information about theextent of cooperative behavior after tradeliberalization. This is compared with the evi-dence of cooperation today with cooperationbefore the turning point using qualitative datafrom the 1997 survey and ®ndings from aprevious survey carried out in 1991 (Rabellotti,1997).

(a) Backward linkages with suppliers

In Guadalajara, clustering induced theconcentration of a critical mass of suppliers.Shoe producers can therefore buy most of theirinputs locally, saving on transportation costs.But, the long closure of the domestic markethas not encouraged the growth of a competitiveindustry of suppliers, hampering price, qualityand service. From the previous empiricalinvestigation it appeared that before liberal-ization most of the relationships between shoemanufacturers and suppliers were market links,based purely on price with very little coopera-tion (Rabellotti, 1997).

Since the opening of the domestic market andthe increase in imports of shoes and compo-

nents, the relationships between footwearproducers and suppliers have been improvingand becoming more collaborative. As can beseen from Table 2, the share of the sample ®rmswhich after liberalization cooperate intensivelywith their leather and sole or heel suppliers ishigh for each of the di�erent speci®ed forms ofcollaboration.

The information collected during thenetwork case studies con®rmed that themanufacturer-supplier relationship is undergo-ing a profound process of change. Suppliers,who were also hit by liberalization,13 reacted tothe crisis increasing their attention to quality,variety of products and fashion content. Forthe ®rst time, after many years of copying fromEuropean and North American journals,suppliers have regularly begun to visit interna-tional trade fairs. According to most of thesuppliers and manufacturers interviewed, thereis now much more information locally avail-able, which circulates within the cluster moreintensively than before liberalization.

Notwithstanding these positive changes, theempirical investigation suggests that the localsupply of competitive raw materials andcomponents is still inadequate to satisfy theneeds of the footwear industry, because onlysome of the existing ®rms have adapted theirsupply to the requirements of the open market.Many suppliers, especially very small ones,have as yet introduced few improvements totheir products.

According to our survey, 68 and 76% of thesample ®rms maintained the same leather andsole suppliers respectively since the liberaliza-tion. The large majority of the ®rms inter-viewed declared that in case of problems withtheir suppliers they always try to solve them.This high stability could have two explanations:lack of choice or strong cooperation andcommitment. It is hard to say which of the twoexplanations prevails.

Table 2. Post-liberalization cooperation with suppliers a;b

Leather suppliers (%) Sole or heel suppliers (%)

Information exchange 54.0 54.0Negotiation of payment and delivery conditions 55.6 41.9Joint product development 44.4 50.0Quality improvement 46.0 52.4Respect of delivering time 58.7 66.7

a % of ®rms which cooperate a lot with their suppliers. The total number of ®rms is 63.b Source: author's survey.


Quality components and raw materials arescarce and often not delivered on time. On thisparticular point, according to the surveyresults, many ®rms cooperate with theirsuppliers over delivery schedules. Nevertheless,many ®rms interviewed for the network casestudies complain about late deliveries. In otherwords, there is more cooperation on thisaspect but this has not solved all deliveryproblems. The suppliers themselves agree thatthis is a crucial issue, because they oftenreceive orders at the very last moment, andthus are unable to satisfy demand on time.Furthermore, delivery time becomes a reallyserious problem when suppliers are required todevelop ad hoc inputs.

From what has been said so far, it appearsthat there has been an improvement in supplier-manufacturer relations but also that suchimprovement has not been universal. Somecategories of shoe ®rms obtain better servicesfrom suppliers than others. The most privilegedare export-oriented enterprises. Suppliers preferto work with them because they pay withoutdelay, sometimes in advance.

Although export-oriented ®rms have moreprivileged relationships with their suppliers andgenerally do not complain about service andquality, many of them stress that to meet thedelivery and quality conditions ®xed by buyersthey have to supervise their suppliers veryclosely. Firms are still learning how to work inan integrated system based on a commonlanguage, which facilitates understanding andreduces costs of transactions.

A major problem is lack of standardizationor, in other words, lack of a common system ofmeasurement adopted by the majority of theMexican manufacturers and suppliers. In thisrespect, the footwear entrepreneurial associa-tions together with the main suppliers haverecently promoted a joint initiative to agree acommon system of standards.

Going back to the relationships betweenshoe ®rms and their suppliers, a furtherdi�erentiation derives from size. From theempirical investigation it appears that small®rms usually have pure market relationshipswith dynamic, good-quality suppliers. Tobargain on delivery and payment conditions,some small ®rms prefer to buy componentsand raw materials from traditional, low-qual-ity suppliers.

To conclude, in Guadalajara two main e�ectsof liberalization on the supplier-manufacturersystem can be stressed:

Ð®rst, there is an increase in cooperation:suppliers and shoe manufacturers have be-gun to think and act as an integrated systemto compete in the open, highly competitivemarket;Ðsecond, this evolution concerns only someof the enterprises, generating an increasingheterogeneity by size and market within thecluster.

(b) Backward linkages with subcontractors

In Mexican footwear clusters division oflabor is generally low, certainly much lowerthan in some Italian or Brazilian clustersinvestigated in recent studies (Schmitz, 1995;Rabellotti, 1995). According to our samplesurvey, 60% of the ®rms do not externalize anystage of the production process, 19% external-ize upper stitching, 18% sole cutting, 14% handstitching and 12% heel covering. The majorityof ®rms which externalize some of theirproduction have more than 70 employees.

A comparison with the results obtained inour previous investigation (Rabellotti, 1997)suggests that division of labor has not increasedconsiderably since liberalization, but, coopera-tion with subcontractors has increased inseveral areas. While previously we did not ®ndany sign of cooperation with subcontractors,according to the recent survey after tradeliberalization 80% of subcontracting ®rms inthe sample cooperate in quality control, 60% ininformation exchange and 52% in negotiationof payment and delivery conditions.

Moreover, there are some other signs of aslow evolution toward a greater division oflabor in the cluster. For example, some ®rmshave begun to specialize only in upper stitching,working for large enterprises, which haverecently increased their export production anddecentralized part of their orders for thedomestic market to local subcontractors.

(c) Forward linkages in the domestic market

Before liberalization Mexican shoe producersmainly sold their products directly to retailers.Because of their geographical concentrationthey acted as a coalition, imposing their qualityconditions onto a very fragmented market14

(Woodru�, 1997). In the protected domesticmarket, the local trade associations played acrucial role in gathering and di�using infor-mation about retailers' behavior,15 reinforcingthe dominant position of the Mexican manu-


facturers toward retailers. In his study on thestructure of contracts in the Mexican footwearindustry, Woodru� (1997) concludes that theinstitutional structure of the market in theclosed economy gave the manufacturers theability both to set the low quality standards forthe industry and to judge whether those stan-dards were met, reducing their incentives tomake innovation in product quality and todevelop a more sophisticated commercialstrategy. Moreover, the existence of weakrelationships between shoe manufacturers andretailers, characterized by very little coopera-tion, is also a clear result of our previousempirical investigation (Rabellotti, 1997).

In the post-liberalization environment, thepower of the manufacturers' coalition wasweakened by providing retailers with the optionof procuring product from foreign producers.This had an initial negative impact on theindustry, with large increase in imports and theloss of ``morality'' of retailers, who began topay late and return unsatisfactory productswith more frequency. Woodru� writes that``Trade liberalization also had the e�ect ofreplacing the quality standard set by themanufacturers' coalition with one determinedby world markets'' (Woodru�, 1997, p. 19).

The ®ndings of our survey show that sincetrade liberalization there is some cooperationbetween shoe ®rms and domestic buyers ininformation exchange, quality control, negoti-ation of payment and delivery conditions andsetting of product speci®cations (Table 3).Furthermore, relationships are rather stable,given that 67% of the sample ®rms trade withthe same buyers.

Trade liberalization also stimulates anincreasing concentration in the domesticmarket: wholesalers, large retail chains anddepartment stores are growing at the expense of

small individual retailers. The large majority of®rms interviewed for the network case studiessaid that after liberalization they increased theirshare of sales to wholesalers, large retail chainsand department stores. The main reasons citedfor their preference for large customers oversmall individual retailers were prompter andmore reliable payments and larger size oforders.

In the Mexican domestic market, retailchains are probably the most innovativebuyers. From our empirical investigation itappears that retail chains have recentlyimproved their relationships with shoe manu-facturers, increasing information exchange andfrequently providing suggestions on compo-nents and raw materials and where to buythem. Some retail chains even provide to theirsuppliers components and raw materials neededfor their products.

A new way of shoe distribution, increasinglypopular in the Mexican market, is cataloguesales. The ®rst catalogue distributor wasestablished successfully a few years ago inMexico City. The success of this ®rst initiativesoon became an example for others, adoptingthe same selling system: the distributors createa catalogue and then sell shoes to nonprofes-sional door-to-door sellers. Catalogue salesrequire a strong capability to plan productionand stocks. Distributors create cataloguesincluding a large number of di�erent patterns,styles and colors and receive very fragmentedorders, often consisting of a few pairs. In orderto reduce the fragmentation of orders, somecatalogue distributors have recently begun too�er training courses to some of their salesvendors.

From what has been said so far, it appearsthat the distribution system prompted byliberalization is increasingly characterized by a

Table 3. Post-liberalization cooperation with buyers a;b

Domestic market (%) International market (%)

Information exchange 67.2 (N� 61) 80.0 (N� 25)Negotiation of payment and delivery conditions 60.7 (N� 61) 60.0 (N� 25)Technological assistance 18.0 (N� 61) 28.0 (N� 25)Quality control 63.9 (N� 61) 80.0 (N� 25)Setting of product speci®cations 42.6 (N� 61) 60.0 (N� 25)Production organization 13.6 (N� 59) 20.0 (N� 25)

a % of ®rms which cooperate a lot with their suppliers. In parenthesis, the total number of respondents for eachquestion.b Source: author's survey.


clear division between production andcommercialization. Shoe producers are awarethat selling to wholesalers or large retail chainsmay create a dependency on them, but at thesame time they know that in the new, highlycompetitive market it is di�cult to sustain anindependent sale strategy. According to manyenterprises interviewed, the establishment ofstable and collaborative relationships withbuyers is the most e�ective way to limit theirdependence on the distribution system. At thesame time, most wholesalers and retailersinterviewed know that to sustain competitionwith adequate quality standards and fast andreliable delivering times they need to establishstable and collaborative relationships with theirsuppliers.

(d) Forward linkages in the export market

Before devaluation, the manufacturers ofJalisco state (of which Guadalajara is thecapital) exported an average of ®ve millionpairs of shoes per year. In 1995 exports doubledto 10 million pairs. Approximately 40 of Jalis-co's 315 shoe factories, which are members ofthe Camara, now export with some regularity,as compared to only ®ve in 1990.16

The export market is dominated by USbrokers, who after the devaluation discoveredin Mexico some enterprises able to producegood quality shoes at competitive price.Exporting ®rms are generally of large size: inour sample 59% of total exporting enterpriseshave more than 70 employees, 30% havebetween 21 and 69 employees and only 11%have less than 20 employees.

The exporting ®rms interviewed in Guadal-ajara are generally happy with their experiencewith brokers. Usually US buyers decide prod-uct speci®cations and then help the Mexicans toproduce them. Manufacturers and brokersoften collaborate in identifying, and sometimesdirectly purchasing, components and rawmaterials.

The exporters interviewed in Guadalajara allagree that they are learning a lot from thebrokers, particularly regarding organization ofproduction, quality control and technology.Table 3 shows that exporting ®rms cooperateintensively with international buyers in infor-mation exchange, quality control, negotiationof delivery and payment conditions and settingof product speci®cations. Other advantages ofthe export market mentioned by the enterprisesinterviewed are the large size of orders and the

relative standardization of products, allowingeconomies of scale.

Apart from exports to the US market, thecluster had a few isolated experiences ofexporting to Latin American countries, oftenthanks to direct contacts established during thetrade fair in Guadalajara. According to theDirector of the Camara del Calzado, LatinAmerica is a promising market, particularly forsmall and medium ®rms in Guadalajara, sincequality and speed matter less.

To conclude, we can say that regular exportsare only experienced by a restricted number ofenterprises in GuadalajaraÐthose with thecapacity to supply large orders. Export manu-facturers have to meet demanding price, qualityand delivery standards. This requires rapidaccess to suitable inputs, quick response capa-bilities on the part of suppliers and therefore awell-functioning and highly integrated localfootwear system. Exports thus generate positiveexternalities to other ®rms located in the clus-ter.

(e) Horizontal linkages with other shoeenterprises

Our interviews suggest that during the yearsof most intense crisis horizontal cooperationwas particularly low: ®rms were too involved intheir day-to-day survival, and were not able toestablish cooperative links and invest in jointprojects. In those years groups of ®rms, whichhad previously regularly exchanged informa-tion, machines or sometimes orders, ceased doso. In addition, the crisis changed profoundlythe relationships at cluster level, as many ®rmswent bankrupt and others reduced or trans-formed their activities. Since the recovery,however, footwear enterprises have begun tobuild new networks. As can be seen in Table 4,this process is still at an early stage and onlyhorizontal cooperation aimed at informationexchange takes place among a signi®cantnumber of ®rms. Other forms of horizontalcooperation within the cluster are quite low.

While horizontal cooperation did not emergeas a signi®cant phenomenon in quantitativeterms, our empirical investigation providedinteresting information about recent as well asearlier initiatives which were not a�ected by thecrisis.

Among the latter are the agrupamientosindustriales, which began in 1983 and formallyended in 1987.17 The agrupamientos werepromoted by National Financiera (Na®n), the


main Mexican development bank, using amethodology originally developed by UNIDO.The groups of ®rms, after a period of training,set up joint projects such as exchanges ofmachines, workers, orders, information orcommon purchasing of inputs. Over 1983±87Na®n promoted the creation of seven groups offootwear producers, the majority of which wenton to meet regularly after the end of the project.Furthermore, some other collective initiativeswere stimulated as side e�ects of the program,namely the creation of a credit union and of atechnological institute for the footwear sector.

Moving on to more recent experiences ofcooperation, we observed several group initia-tives. An interesting example is a group thatbrings together 80 suppliers of the largestMexican catalogue distributor and aims toencourage cooperation in product developmentand joint purchase of components and rawmaterials. A planned project between thedistributor and Na®n seeks to facilitate accessto credit for ®rms belonging to the group.

Another group is composed of the mainexporting ®rms located in Guadalajara. Thisgroup comes together for informal exchangesof information on technological aspects, todiscuss availability and quality of componentsand raw materials and to exchange machineryand technicians.

To conclude, although the experience ofcooperation described above can not begeneralized, there are an increasing number ofcases of inter®rm collaboration in the Guadal-ajara cluster. It is, however, too early to tellwhether this represents a shift toward a morecooperative culture.

(f) Institutional linkages

During the crisis of 1992±94 the three localbranches of the trade association concentratedmost of their e�orts on lobbying the Federal

Government to support the footwear industry.Finally, in May 1995 a document for thedevelopment of the sector was approved,including as its main measure the increase intari�s to 35% for a period of ®ve years. This actis generally considered as the beginning of therecovery for the industry.

In Guadalajara, the crisis hit several institu-tions aimed at supporting footwear enterprises:®rst, the Instituto Tecnologico del Calzado, aninstitution set up with the collaboration of theCamara and specialized in training and tech-nological assistance, was closed due to lack ofdemand of its services; second, the credit unionsu�ered from a very high rate of insolvency,caused by a huge increase in interest rates afterthe devaluation.

As regards the Camara del Calzado, duringthe crisis it engaged only in defensive activities.With the beginning of the recovery it hasstarted again to organize and promote the localtrade fair. With an intense promotional activityin the domestic and international markets, thenumber of exhibiting ®rms increased from 70in 1995 to more than 250 in 1996. There wasalso a signi®cant increase in the number offoreign buyers who visited the fair. Particularlylarge was the participation of buyers fromLatin America. Furthermore, the Camaraholds training courses to prepare small ®rmswhich seek to participate for the ®rst time inthe trade fair. For promotion abroad, theCamara also organizes joint participation ofgroups of local ®rms to the main internationalexhibitions, promotes new market studies andprovides an information service on foreignbuyers.

In the ®eld of training and technologicalassistance, the Camara is involved in theestablishment of a local branch of Ciatec inGuadalajara. Ciatec was created in Leon byConacyt, the Mexican national science andtechnological council, and is an institution with

Table 4. Post-liberalization cooperation with other local shoe enterprises (% of ®rms) a

A lot (%) A little (%) Never (%) Total number of ®rms

Information exchange 22.2 44.4 33.3 63Joint orders 9.5 12.7 76.2 63Joint product development 7.9 11.1 81.0 63Machinery lending 6.3 17.5 76.2 63Joint sale 7.9 12.7 79.4 63Joint training 3.2 6.5 88.9 62Joint purchase 7.9 12.7 79.4 63

a Source: author's survey.


a long tradition in supporting the footwearsector. In Guadalajara, it will supply services,previously o�ered by the Instituto Tecnologico,making available to local enterprises its highlyspecialized know-how, accumulated in manyyears of activities in Leon.18

Given these new institutional activities, theresponse of local footwear enterprises appearsto be quite good: according to the results of oursurvey, 45% of the sample ®rms said that theyuse the services supplied by the Camara morefrequently than before liberalization.

From what we have said so far, it emergesthat in Guadalajara institutional support hasbegun to ¯ourish again, particularly with afocus toward the promotion of the cluster inthe international market. Local enterprisesappreciate these e�orts and seem to be willingto strengthen horizontal cooperation throughinstitutions.

5. IS FIRM PERFORMANCEINFLUENCED BY COOPERATION?

In the previous section we presented adescriptive analysis of the changes occurred invertical and horizontal linkages within thefootwear cluster of Guadalajara after tradeliberalization. Our focus was on cooperativebehavior both vertical and horizontal. In whatfollows, with ordinary least squares (OLS)regression analysis, we assess if cooperation isrelated with performance using the resultsobtained from the questionnaire survey cover-ing a sample of 63 enterprises. Furthermore, wetest the relationship between size and ®rmperformance.

The independent variables included in theregression model are indicators of horizontal,19

backward and forward cooperation20 in thedomestic market and two dummy variables forlarge and small size. The response variable is a

performance indicator obtained with principalcomponent analysis (Table 5).21

Concerning cooperation, the main results ofthe regression analysis, presented in Table 6,are as follows: horizontal cooperation throughthe entrepreneurial association (C31) and withother local shoe ®rms (C32) and backwardcooperation with leather and sole suppliers(LOGD4143) are positively and signi®cantlyrelated with performance. Among our cooper-ation indicators, forward cooperation withbuyers in the domestic market (LOGF61) is theonly variable not statistically signi®cantlyrelated to performance. Furthermore, thoughnot reported here, the computed values of thestandardized beta coe�cients22 suggest thathorizontal cooperation through the entrepre-neurial association is the largest of any regres-sors, followed by backward cooperation withleather and sole suppliers and horizontalcooperation with other local shoe ®rms.

As can be seen from Table 6, besides coop-erative behavior size also in¯uences ®rmperformance. The dummy variable for smallsize (Ds) is in fact negatively and statisticallysigni®cantly related with the dependent vari-able. In the next section we will see that coop-erative behavior depends also on size, furthercon®rming the hypothesis of the existence of aconsiderable internal heterogeneity within thecluster.

Finally, given that in the questionnaireperformance was originally measured byseveral indicators, the robustness of theregression model is also tested substituting thedependent variable FAC1 with the seven orig-inal variables.23 The results con®rm therobustness of the original model for at leastfour of the performance variables: production(B21), sales (B23), pro®ts (B24) and employees(B28) trends as dependent variables.24 They arepositively and signi®cantly associated withhorizontal cooperation with other shoe ®rms

Table 5. Variables included in the performance equation for OLS estimation

Variables

FAC1 Dependent variable obtained from principal component analysis on performance indicatorsC31 Horizontal cooperation through the entrepreneurial associationC32 Horizontal cooperation with other local shoe ®rmsLOGD4143 Log of backward cooperation with leather and sole suppliersLOGF61 Log of forward cooperation with buyers in the domestic marketDs Dummy for small size (20 or less employees)Dl Dummy for large size (70 or more employees)


(C32) and through the entrepreneurial associ-ation (C31), with backward cooperation withleather and sole or heel suppliers (LOGD4143)and with the dummy for small size (Ds).

To conclude, the hypothesis that since tradeliberalization shoe ®rms' performance has beenpositively in¯uenced by cooperation with other®rms is con®rmed by regression analysis on arandom sample strati®ed by size, covering 63enterprises located in Guadalajara. Horizontalcooperation with other local shoe ®rms andthrough the entrepreneurial association as wellas vertical cooperation with input supplierscontributes signi®cantly to sample ®rms' goodperformance.

6. A TEST OF SIZE HETEROGENEITYWITHIN THE CLUSTER

The industrial district literature has givenlittle attention to internal di�erentiation by sizeor by performance within clusters, but it hasinstead favored the di�usion of an idea ofhomogeneity and unity which rarely exists. Inour previous work on two Mexican footwearclusters and two Italian districts the enormousinternal heterogeneity came out clearly as a by-product of research (Rabellotti, 1997).Furthermore, internal heterogeneity is the focusof a recent comparative paper on footwearclusters in Brazil, Italy and Mexico (Rabellottiand Schmitz, 1999).

The present analysis of the structural changesoriginated by liberalization in the Guadalajara

shoe cluster con®rms the existence of aconsiderable internal heterogeneity. From thequalitative empirical evidence presented inSection 4 the following forms of internalheterogeneity within the cluster can beemphasized:

Ð®rst, heterogeneity is generated by thesuppliers' practices: some shoe manufactur-ers cooperate with them to develop ad hocinputs tailored to their requirements, othershave to buy what is available in the market;Ðsecond, heterogeneity is generated by theuse of di�erent marketing channels: exportsvs. domestic sales and within the domesticmarket department stores, large retailchains, catalogue sellers or small individualretailers;Ð®nally, there is also heterogeneity by size,con®rmed by Table 7, showing that aftertrade liberalization large and medium ®rmshave performed better than small ones.

The relationship between internal heteroge-neity by size and ®rms' performance is testedwith OLS in the model presented in Table 6. Inthis equation the e�ect of the dummy variablesfor large and small size on performance isadditive; in other words, it increases ordecreases performance by a ®xed amount,regardless of the other predictors. An alterna-tive model allows the e�ect of size to be di�er-ent for di�erent combinations of the otherpredictors, namely of cooperation indicators.The new equation to be tested is obtained by

Table 6. OLS estimation of performance equation a

Variable Beta Coe�cient Std. Error T-Statistic

Constant ÿ0.304 0.234 ÿ1.301C31 0.281�� 0.105 2.688C32 0.538� 0.286 1.881LOGD4143 1.571�� 0.715 2.196LOGF61 0.282 0.324 0.871Ds ÿ0.494� 0.262 ÿ1.883Dl 0.117 0.262 0.447

R-squared 0.422Adjusted R-squared 0.355S.E. of regression 0.802F-statistic 6.237P-value (F-statistic) 0.000

a Dependent variable is FAC1; number of observations: 63.� 10% level of signi®cance.�� 5% level of signi®cance.�� 1% level of signi®cance.


de®ning a new set of independent variables thatare the product of the dummy variables and theoriginal predictors.25 With the stepwise selec-tion method we obtain the equation presentedin Table 8.26 The most interesting result is thevariable LOGD4143*Dl, which is statisticallysigni®cant and positively related with perfor-mance. This suggests that large ®rms' cooper-ation with suppliers positively and signi®cantlyin¯uences performance.

From the statistical results presented we canconclude that performance and some forms ofcooperative behavior, namely cooperationwith suppliers, varies with size. These resultsalso con®rm the existence of internal hetero-geneity within the cluster investigated.

7. CONCLUSIONS

The main question addressed in this paperconcerns the impact of market liberalization onthe footwear industry cluster of Guadalajara.The impact was quite strong consisting of ahuge reduction in the number of ®rms and indomestic production and in a large increase inshoe imports. Trade reform however made®rms aware of global competitive pressures.Furthermore, the later return to partialprotection and the Peso devaluation gave some®rms the time to respond with greater cooper-ation with suppliers, buyers and through theentrepreneurial association.

Cooperation within the cluster increases®rms' capability to grow and the indicators ofperformance in our questionnaire can beconsidered as a rough proxy for growth.Therefore, with regression analysis on datafrom a sample of 63 enterprises, we tested thehypothesis of a positive association between®rms' performance and their cooperativebehaviors. The model shows clearly thatperformance is positively and signi®cantlyrelated with some forms of vertical and hori-zontal cooperation. Therefore, one canconclude that cooperation, which is one of themain components of collective e�ciency,contributes to ®rm growth.

External economies are another source ofcollective e�ciency. This e�ect has not beenmeasured in our investigation, but it is a clear

Table 8. OLS estimation of performance equation withmultiplicative dummy variables for small and large size

(stepwise selection model) a;b

C31 0.314�� (0.106)LOGD4143*Dl 2.536�� (1.020)Ds ÿ0.606�� (0.237)Adjusted R-squared 0.361S.E. of regression 0.820P-value (F-statistic) 0.000

a Dependent variable is FAC1; number of observations:63; Constants are not reported. Standard errors inparentheses.b Source: author's survey.�� 5% level of signi®cance.�� 1% level of signi®cance.

Table 7. Performance indicators by size a;b

P 70 employees(N� 20) (%)

21±69 employees(N� 21) (%)

6 20 employees(N� 22) (%)

Chi-squares c

Production 70.0 61.9 22.7 11.105(0.025)

Export 50.0 (N� 16) 23.8 (N� 8) 4.5 (N� 3) 20.185(0.003)

Sales 75.0 52.4 22.7 11.698(0.020)

Pro®ts 25.0 14.3 ÿ 12.422(0.053)

Average price 65.0 42.9 68.2Average deliverytime

35.0 33.3 36.4

Product quality 80.0 76.2 (N� 20) 73.0 (N� 21)Employees 35.0 23.8 9.1

a % of ®rms which after liberalization have registered an increase in the listed performance indicators (for averagedelivery time it is indicated a decrease).b Source: author's survey.c The signi®cance level is in parenthesis. If the chi-squares are not reported the size distribution is not statisticallysigni®cant.


®nding of the qualitative research. Forinstance, there are positive externalities gener-ated by cooperationÐcalled by Nadvi (1997)``external economies of joint action''Ðbetweensome shoe manufacturers and suppliers:improvements in quality, fashion content,speed in delivery which percolate through thecluster. Important externalities also emergefrom cooperation between the few exportingenterprises and their foreign buyers: learningfrom exports takes place at individual levelamong exporters but also spills over to the restof the cluster.

A further e�ect of the structural changescaused by liberalization is the increasingheterogeneity within the cluster. Internalheterogeneity by size in¯uences ®rms' perfor-mance, as con®rmed by regression analysis.Nonetheless, the more successful ®rms generateexternality gains to others in the cluster.

From what has been said so far it appearsthat although some external events such as thepeso devaluation and the return to highermarket protection helped the footwear industryof Guadalajara, the increase in collective e�-ciency also played an important role in therecent recovery of the cluster. Nevertheless, thepositive impact of collective e�ciency a�ectsonly part of the cluster: those ®rms which enterinto cooperative actions and those which areable to bene®t from some externalities gener-ated by cooperation. The main challenge forthe future will be to transform the static part ofthe cluster. Is it possible to assist the lessdynamic manufacturers to ``switch gears'' inorder to meet the challenges of the increasinglycompetitive and rapidly evolving internationalmarket? How can they be helped to cooperateand upgrade? These are crucial questions forpolicy makers and donors.

NOTES

1. For a summary of the argument and evidence see

Schmitz (1995).

2. A turning point may be de®ned as a break with the

past that opens up new opportunities or poses new

threats.

3. The other main cluster is Leon, specializing in men's

and children's shoe production.

4. For details on the questionnaire used, see Rabellotti

(1998).

5. It is not realistic to expect trade liberalization to lead

to a substantially greater e�ciency and competitiveness

of the domestic industry in a short period of time.

Following Ten Kate (1992), one should expect a lag of at

least 5±10 years before the bene®ts of trade reform fully

materialize.

6. A very good and comprehensive survey of this

literature is presented in Rodrick (1995).

7. In a study on trade reform in Africa Lall (1998)

adopts an alternative approach to standard neoclassical

theory based on evolutionary or ``neo-Schumpeterian'',

theory to analyze how import liberalization may be

expected to a�ect technological activity in developing

countries.

8. External economies are de®ned as the by-product of

some activities undertaken within the clusters; while

cooperation e�ects are the result of explicit and volun-

tary cooperative behaviors (Rabellotti, 1997).

9. According to the last available census (1990), the

population of the metropolitan area of Guadalajara is

2.9 million.

10. Although most of the ®rms selected were a�liates

to the local trade association, a few ®rms, which did not

belong to the Camara, were also interviewed.

11. In the previous survey carried out in Guadalajara,

the sample was also randomly taken from the register of

members of the local entrepreneurial association.

According to the association and to sector experts

interviewed, the lists are representative of the formal

enterprises because when ®rms decide to register o�-

cially for paying taxes and social bene®ts, they usually

become members of the entrepreneurial association,

which helps them in dealing with bureaucratic proce-

dures (Rabellotti, 1997).

12. The questionnaire was elaborated together with

Peter Knorringa, Khalid Nadvi and Hubert Schmitz and

also adopted for ®eld work carried out in clusters in

India, Pakistan and Brazil. For details see Rabellotti

(1998).


13. Figures on the impact of liberalization on suppliers

were collected only in the tannery sector. According to

estimates of the local Chamber of Leather Producers,

there was a 40% decrease in leather production (inter-

view with Rogelio Alferez, September 1996).

14. According to the 1989 Mexican Industrial Census

there were more than 19,000 footwear retailers, the large

majority of them individual retailer stores or very small

chains.

15. Both the Camara del Calzado in Guadalajara and

Leon have a data base which includes information

respectively on some 1,300 and 4,300 retailers about

their payment history and their behavior (e.g., if he/she

accepts orders when delivered, how often he/she returns

orders for quality problems).

16. These estimates are provided by the local Camara

del Calzado.

17. In the 1990s a new initiative for promoting

agrupamientos industriales was launched by the Camara

del Calzado.

18. In Leon, the largest footwear cluster in Mexico,

institutionalized cooperation has recently increased

rather strongly for two main reasons: the local political

situation and the very strong specialisation of the local

economic system. Both at the state and the municipal

level there are respectively a governor and a mayor who

began their professional life in the footwear industry and

who are therefore very receptive to the industry's

problems. Furthermore, given that in 1992 the footwear

®li�ere accounted for 15.5% of the GDP of Guanajuato

(the state of Leon) and for 68% of the GDP of the city of

Leon (CEESP, 1993), the crisis of the sector hit the local

economy strongly. This called for urgent policy inter-

ventions.

19. Horizontal cooperation through the entrepreneur-

ial association has to be considered cautiously, given

that the sample is selected from the list of the Camara's

members. Among the Camara's members however there

is a wide variety of opinions concerning the activities of

the entrepreneurial association. This result is also

con®rmed by our previous inquiry (Rabellotti, 1997).

20. Information about cooperation with subcontrac-

tors and buyers for the export market was also collected,

but due to the high number of missing values it could not

be included in the statistical analysis.

21. For details on how aggregate indicators of perfor-

mance and cooperation were constructed, see Rabellotti

(1998).

22. The standardized betas are the estimated coe�-

cients in a regression where the original variables have

been divided by their sample standard deviation so as to

clean the estimated coe�cients of their dependence on

measurement units.

23. The share of exports is omitted from the principal

component analysis, due to the high number of missing

values. The test can be done in two di�erent ways: ®rst

of all, with seven OLS separate models in which the

independent variables are always the same while the

dependent variables are the seven performance indica-

tors and secondly, with multivariate regression analysis,

testing the relationship between a set of interrelated

dependent variables (the seven performance variables)

and one group of independent variables. For further

details see Rabellotti (1998).

24. Results are not reported in the paper, but on

request they are available from the author.

25. The full equation estimated with OLS

is: FAC1�b0 + b1Dl + b2Ds + b3C31 + b4C32 + b5

LOGD4143 + b6LOGF61 + b7C31 Dl + b7C31 Ds +

b8C32 Dl + b9C32 Ds + b10LOGD4143 Dl + b11 LO-

GD4143 Ds + b12LOGF61 Dl + b13LOGF61 Ds.

26. The model does not change if backward and

forward methods are adopted for variable selection.

For further details on this methodology, see Rabellotti

(1998).

REFERENCES

Arias, P. ed. (1985) Guadalajara, La Gran Ciudad de laPeque~na Industria. El Collegio de Michoacan,Zamora.

Arias, P. (1992) El Calzado en la Region Jalisciense: LaIndustria y la Camara. Camara de la Industria delCalzado de Jalisco, Guadalajara.

Balassa, B. (1988) Interest of developing countries in theUruguay Round. The World Economy 11, 39±54.

CANAICAL (1996) Per®l de la Industria del Calzado,Camaras de la Industria del Calzado, Mexico City.

CEESP (1993) Boletin informativo. Mimeo, Centro deEstudios Economicos del Sector Privado, Leon.


Dominguez-Villalobos, L. and Grossman, F. (1992)Employment and income e�ects of structural andtechnological changes in footwear manufacturing inMexico. Working Paper, No. 224, World Employ-ment Program Research, International Labor O�ce,Geneva.

ITESM (1995) Jalisco 2000; De frente a las nuevasrealidades. ITESM, Guadalajara.

Lall, S. (1998) The Technological Response to ImportLiberalisation in Africa. Macmillan, London.

Nadvi, K. (1997) The cutting edge: Collective e�ciencyand international competitiveness in Pakistan. IDSDiscussion Paper, No. 360, Institute of DevelopmentStudies, University of Sussex, Brighton.

Rabellotti, R. (1995) Is there an industrial districtmodel? Footwear districts in Italy and Mexicocompared. World Development 23, 29±41.

Rabellotti, R. (1997) External Economies and Coopera-tion in Industrial Districts. A Comparison of Italy andMexico. Macmillan, London.

Rabellotti, R. (1998) Recovery of a Mexican cluster:Devaluation bonanza or collective e�ciency? IDSWorking Paper No.71, Institute of DevelopmentStudies, University of Sussex, Brighton.

Rabellotti, R. and Schmitz, H. (1999) The internalheterogeneity of industrial districts in Italy, Braziland Mexico. Regional Studies 33, 97±108.

Rodrick, D. (1995) Trade and industrial policy reform.In Handbook of Development Economics, eds. J.Behrman, and T. Srinivasan, pp. 2925±2982, Vol.IIIB. Elsevier, Amsterdam.

Schmitz, H. (1995) Small shoemakers and fordistgiants: Tale of a supercluster. World Development23, 9±28.

Ten Kate, A. (1992) Trade liberalization and economicstabilization in Mexico: Lessons of experience. WorldDevelopment 20, 659±672.

Woodru�, C. (1997) Trade associations and coalitionbargaining with incomplete contracts. University ofCalifornia, San Diego.


recovery of a mexican cluster: devaluation bonanza or collective efficiency?

Documents