Jahrbuch f¨ur Regionalwissenschaft (2010) 30: 23–44 DOI 10.1007/s10037-009-0041-7 ORIGINAL PAPER

The aftermath of openness and integration in the Mexican manufacturing Alejandra Berenice Trejo Nieto

Accepted: 1 December 2009 / Published online: 19 January 2010 © Springer-Verlag 2010

Abstract Mexico is a country whose development process has been traditionally characterised by a substantial unbalanced performance of its regions. Since the early 1980’s it carried out an intense liberalisation and export promotion programme representative of a new national development stratagem. This paper looks at the spatial patterns of Mexican manufacturing at the same time as the shifts after the trade liberalisations are particularly weighed up. Data on manufacturing Gross Domestic Product by state is employed for an examination of shares, agglomeration indices and densities. The data reveals a long-run dispersion tendency, the changes however are small. Although the “largest” geographic dispersion happened after the generalised liberalisation programme and after the entry into force of the North America Free Trade Agreement relocation was already in progress in the 1970’s. States in the north were the initial beneficiaries of the spatial decentralisation of manufacturing but states in the so called Baj´ıo region have had the major upward changes more recently. There is no evidence of a significant change in a dual regional structure. The argument about cumulative causation and the advantages of initial conditions is pertinent for the case of Mexico City. Moreover a multi-agglomerate pattern within the core is consolidating. Keywords Regional disparities · Mexican manufacturing · Economic geography · Agglomeration · Trade liberalisation

Article originally presented at the International Seminar on Globalization, Agglomeration and Inclusive Growth, organized by UNCTAD-JUECON Programme, Department of Economics, Jadavpur University, Kolkata and UNCTAD-India, New Delhi May 21–22, 2009. A. B. Trejo Nieto (u) Centro de Investigaciones Socioecon´omicas, Universidad Aut´onoma de Coahuila, Unidad Campo Redondo, Edificio S, 25280 Saltillo, Mexico e-mail: [email protected]

13

24

A. B. Trejo Nieto

1 Introduction The strong spatial clustering of economic activity has been a remarkable characteristic of many economies in the contemporary world. The distribution of economic activity in space is extremely ‘lumpy’ and characterised by significant variations in the intensity and character of economic activities from one place to another. Such unevenness is manifest at diverse geographic levels. In the world economy a coreperiphery structure is identified: high income countries constitute a core which is situated in the North Hemisphere whereas productivity per capita declines with distance from this core. In 2000 more than 80% of the world’s Gross Domestic Product (GDP) was produced by three regional blocs whilst this pattern has become stronger over time. Similarly, 50% of the world’s GDP is produced by 15% of the world’s population while 54% is produced by countries occupying just 10% of the world’s land area (Henderson et al. 2001). In Europe the United Kingdom, Belgium, Germany, Switzerland and Italy form the core of that continent’s economic activity (Brackman et al. 2001). Spatial disparities are rather evident within countries where there are industrial centres with a high concentration of economic activity and employment, and regions whose contribution to GDP and employment is marginal (Venables 2003). Regional disparities are even stronger in less developed countries where metropolitan concentrations of activity and population are typical. Henderson et al. (2001) points out that agglomeration of economic activity is often reflected in the urban system of those nations. On the other hand, the geographic distribution of economic activities within a country may undergo changes induced by the external shock that events such as trade liberalisation and economic integration represent. Such external shocks are, on the one hand, associated with the processes of globalisation and, on the other, determined by the kind of policies that countries implement. Protectionist policies have been frequently assumed to have led to a strong spatial agglomeration of economic activities provided that in an economy closed to external trade firms tend to locate near the main domestic markets -usually the capital city. Likewise it has been suggested that trade liberalisation and economic integration led to changes the spatial distribution of production and employment within countries where the outcomes can vary from dispersion of economic activities, to increasing agglomeration or simply relocation. In Mexico the economic dominance of Distrito Federal – the administrative region coextensive with Mexico City – and a few other states is emblematic of the geography of manufacturing during the Import Substitution Industrialisation regime (ISI). In the 1980s Mexico moved from an inward-looking industrialisation strategy towards an open economy regime. Far-reaching restructuring policies were expected to be beneficial in terms of regional equity and spatial development only under the premise that backward regions gained from increasing trade liberalisation resulting in a more homogeneous and a less fragmented geography of production. By and large previous research has provided evidence of a northward relocation of manufacturing with a corresponding geographic adjustment that followed the adoption of an extensive program of trade liberalisation (see for instance Krugman and Livas Elizondo 1996; Chamboux-Leroux 2001; Sanchez-Reaza and Jordaan 2002;

13

The aftermath of openness and integration in the Mexican manufacturing

25

Corona J´ımenez 2003; Decuir-Viruez 2003; Sanchez-Reaza and Jordaan 2004). Some of these studies also suggest that the relative fall of Distrito Federal became less apparent over time because this region underwent recoveries in some periods. Hence outcomes are still contentious; in particular two questions remain open: 1) to what extent has the spatial pattern of production in Mexico changed since the opening of the economy? And 2) what have been the role of economic reforms in such a change? This paper attempts to evaluate, in the light of the Mexican data set, mainly the question of how the internal economic geography of Mexican manufacturing has evolved and adjusted to trade liberalisation and integration. We look at regional shares in GDP, GDP densities as well as summary indicators of overall levels of agglomeration. Examining past and current patterns affords the possibility of observing location and agglomeration dynamics. We find a number of stylised facts about location and agglomeration which are the categories used in order to characterise the observed ‘geographical landscape’ of Mexican manufacturing. We also try to identify the actual regional structure. Contrary to the argument of the positive effect of free trade on spatial disparities this paper puts forward evidence that spatial development in Mexico has been limited to and localised in particular regions. Thus a mild impact of external shocks on agglomeration but a number of noteworthy shifts in location is a nuanced outcome to be considered. Certainly, the resulting spatial distribution of manufacturing also depend on the initial conditions whilst additional forces, other than trade or Foreign Direct Investment (FDI), may as well lead to one or another result. In the subsequent sections we present a review of the main characteristics of the transition from a protectionist to a liberalised economy in Mexico followed by a brief theoretical discussion. Then after a methodological note the changes in the geographic distribution of manufacturing are analysed using a series of indicators. Also a characterisation of the regional structure of manufacturing is depicted. Finally there is a section with concluding comments.

2 Industrialisation and trade regimes in Mexico After its industrialisation take-off in the 1940’s Mexico implemented import substitution industrialisation as their predominant economic scheme for over twenty years. The government explicitly stated that the expansion of the manufacturing sector was essential to modernisation and economic self-sufficiency in the long-run. Therefore it set the conditions for profitable private investment by indirect regulation and through direct intervention in specific and strategic industries. It also established mechanisms by which public resources were allocated to investment projects (Norris et al. 1999). Largely economic, industrial and trade policies were set in accordance with the ISI model (Norris et al. 1999): 1. Macroeconomic policy assured durable economic stability: public deficits, money supply and interest rates were determined by policy. Nominal wages and labour unions were also well under control. 2. Trade policy set up high levels of protection which allowed domestic producers to enjoy high profits. A comprehensive protective apparatus was set in order to in-

13

26

A. B. Trejo Nieto

crease the attractiveness of productive investment. This system relied strongly on quantitative import controls being import-licensing requirements the primary tool for protectionist purposes.1 3. Industrial policy promoted the reallocation of resources to manufacturing activity. It targeted specific sectors, which benefited from tax reductions, credit allocation, preferential interest rates, reimbursement of import duties and access to import quotas. In principle, the progression of early Mexican industrialisation under the ISI strategy does not differ radically from that of other Newly Industrialised Countries. These underwent a conversion in their economic structures from predominantly agricultural to industrial, and a transition from a largely rural society towards an urban population (Fischer et al. 1982). The Mexican economy reached stable and rapid economic growth, an expansion of its domestic productive plant and market as well as price and exchange rate stability (Kalter 1992). However Mexico did not transform its domestic manufactures into successful exports as many East Asian countries did. Instead, it opted for a more restrained approach to export promotion. This consisted of a regional programme in which states at the border with the US underwent industrialisation of some kind through the implementation of the maquiladoras or in-bond industry which reached partial success at the regional level. The establishment of export processing zones (EPZs) along the border with the US favoured only a limited number of industries and regions (Cravey 1998; Norris et al. 1999).2 All this resulted in a low ratio of total trade to GDP (Puyana and Romero 2006). Economic planners were successful with the first phase of ISI but in the 1970s this model revealed its limits as a useful development strategy as Mexico’s highly protective structure created a number of economic inefficiencies and deficiencies (Kalter 1992; Cravey 1998; Norris et al. 1999). The decline of the industrialisation model brought about a change in perspectives and a reversal of the whole economic scheme. Not only national but international conditions brought the need for urgent restructuring in the early 1980s. Globalisation of production and markets was relevant in accounting for the expansion of new outwardoriented policies. An intense unilateral liberalisation – by which Mexico dismantled its trade barriers to a large extent, privatisation and deregulation processes were the expression of the significant turnabout in the development approach. The country progressed towards an open-economy regime on a multilateral basis via accession to the GATT (later the World Trade Organisation, WTO), the Asia-Pacific Economic Cooperation (APEC) and the Organisation for Economic Cooperation and Development (OECD). Due to the slow progress of multilateral negotiations for Mexico the next move in order to guarantee the continuity of its liberalisation programme and long-term ac1

More than two thirds of total imports required licenses. A number of products imposed duties as high as 100%. The highest rates were imposed on final goods, with relatively lower rates imposed on selected intermediate and capital goods; higher rates for consumer durables and lower for intermediate and nondurable consumer goods (Gereffi 2003). 2 For a few industries there were tariff exemptions, tied to specific export targets, for the import of intermediate and capital goods. Drawback provisions for import duties on exported goods were designed.

13

The aftermath of openness and integration in the Mexican manufacturing

27

cess to its most important market was a formal trade integration with North America (Vega and de la Mora 2003). Geographical proximity along with economic dependence made this integration a natural step (Kessel 1995). Mexico’s main motivation was to increase FDI flows which potentially would create employment opportunities and higher remunerations (Cravey 1998). Signed and put into effect in 1994, NAFTA represented the first reciprocal Free Trade Agreement (FTA) joining two industrialised economies, the US and Canada, with a developing country, Mexico (Norris et al. 1999). From the 1990s Mexico negotiated and reached other several FTAs as opposed to sector-by-sector accords. With this it achieved preferential access to fortytwo national markets in Europe, Asia and America and represented a potential for export expansion and diversification, and consequently for production and employment. Although the domestic demand contracted after a severe economic recession in 1995,3 some positive outcomes were reached in terms of exports. The share of nonmaquiladora exports within total GDP increased from 17% in the 1993 to over 34% in 2003. Non-oil exports tripled between 1994 and 2008 growing at an annual average rate of 8.4%. The number of exporting firms rose from 22 thousand in 1994 to 34 thousand in 1998. Mexico became the eighth largest trading nation in the world and the first in Latin America (Vega and de la Mora 2003). Maquiladoras played a predominant role in export expansion (Mari˜na Flores 2004). Their number more than doubled between 1990 and 2001 going from 1500 to 3700 (L´opez Villafa˜ne 2001). In the 1990s they accounted for more than 40% of Mexico’s total exports (85% shipped to the US) and employment passed the one million mark (L´opez Villafa˜ne 2001; Gereffi 2003). As with regard to trade, manufacturing symbolized the core of total external flows. In Mexico exports in manufactures are more than 95% of total non-oil exports which in turn are the 90% of the total. Exports in this sector grew at an annual average rate of 12.8% between 1980 and 2008 whereas total non oil exports had an AAGR of 10.9%. Due to the heaviness of manufacturing in trade, it was the sector by which liberalisation and integration could influence national and regional growth. The main exporting regions were the traditional industrial hubs of Distrito Federal and Nuevo Leon together with the northern states of Chihuahua, Baja California and Tamaulipas. Exporting firms concentrated in the north but expanded increasingly to the Baj´ıo (Jalisco, Guanajuato and Aguascalientes). The top exporting regions were also the states with the highest FDI inflows (L´opez Villafa˜ne 2001). In 1999 only 302 firms accounted for 93% of total exports and foreign firms represented the most substantial part of the exporting sector while domestic firms’ share in total manufacturing exports dropped from 35.8 in 1998 to 25.3% in 2000. With NAFTA Mexico’s attractiveness increased not only for US investors but also for other foreign firms which sought to take advantage of the privileged access to the US market. Between 1994 and 2003 Mexico received US142 billion dollars of FDI 3

In 1994 a series of internal incidents (political assassinations and an indigenous rebellion in Chiapas) and external events (a remarkable reduction in foreign capital flows into Mexico because of increases in interest rates in the US) led to the collapse of the Mexican currency and created the conditions for a severe crisis, from which the country had not recovered by 1996.

13

28

A. B. Trejo Nieto

and became the fourth recipient of FDI among developing countries and the second in Latin America only behind Brazil. The US was the main source of foreign investment followed by the European Union. By and large FDI in manufacturing was directed towards the most dynamic exporting sectors (Secretar´ıa de Econom´ıa 2004). On the other hand, imports showed a parallel boom. The ratio of imports to GDP grew from 22% in 1994 to more than 37% in 2003. The imported component of total supply increased considerably and reached 16.2% in 2000 whereas the imported component in manufacturing goods reached 26%. Moreover final goods were the growing component in total imports illustrating an exacerbating import competition in consumer goods. Exports were significantly affected by the US economic slowdown in 2000 which caused a decline of this country’s imports. Mexican total and manufacturing production contracted after the fall in export levels (Figs. 1 and 2). Although the export strategy reached partial achievements, as with the ISI, some structural problems soon were evident. Mexico developed a secondary exporting model characterised by its fragility and dependence on maquiladora exports which had a limited impact on the economy. The economy was heavily reliant on the strategies of transnational companies and on the US’ economic activity. Furthermore, economic growth was attained to the detriment of employment and wages. Exports responded more to the lack of internal demand during recessive periods and to large devaluations of the peso than to export promotion itself. NAFTA signified a limited trade and investment integration due to the asymmetries among members, principally due to the weak stability of the Mexican economy (cyclic patterns of contractions and slow recoveries, price instabilities, extremely high interest rates and inflation rates, and a currency of questionable steadiness). Even though Mexico expanded its network of FTAs and investment treaties it did not advanced far into the diversification of exports in any sense. Instead the success of the export-oriented model was limited to a small number of sectors, companies, countries of destinations and regions of origin. In addition to this, structural de-

Fig. 1 Total and Manufacturing GDP Annual Growth Rates, 1994–2006 (%)

13

The aftermath of openness and integration in the Mexican manufacturing

29

Fig. 2 Exports’ Annual Growth Rates, 1994–2008 (%)

pendence on imported intermediate and capital goods and technology was hardly overcome. Indeed Mexico underwent a very particular kind of globalisation. Mexican globalisation has been very close to an exclusive relationship with the United States. It has represented privileged access to a big market by means of preferences within NAFTA, through which Mexico achieved partial international positioning.

3 Internal economic geographies and economic globalisation Old and recent literature have highlighted the significance of studying the geographical aspects of the economy because of their link with urbanisation processes and the evolution of regional development (see for instance Henderson et al., 2001). Similarly, the role of economic globalisation in shaping the geographical landscape of economies and development is regarded as relevant. With this respect two fundamental transformations in the global context remade old perspectives: firstly, the major transformation in how the international economy – markets and production – was organised, and secondly a shift in national economic strategies from Import-Substituting Industrialisation (ISI) to Export-Oriented Industrialization (EOI). By and large economic globalisation has been associated with the increasing integration of trade, production and finance across countries and has been described as the process in which economic transactions assume a transnational character. It has been to some extent the product of the opening up of economies through trade and investment liberalisation both of which have been attained by means of multilateral negotiations, integration schemes, bilateral agreements and unilateral liberalisation programmes. Economic globalisation is likely to have had an impact on the spatial allocation of production and on its efficiency in a similar way as it has an impact on other relevant parameters of productive activities. In this situation, territories at

13

30

A. B. Trejo Nieto

different scales undergo transformations leading to processes of adjustment to new scenarios. From the point of view of traditional trade theory all forms of economic integration are fundamentally beneficial. But the presence of significant external economies and increasing returns, which is highlighted by the new trade theory and geographical economics theory, may cause trade and integration to have adverse effects such as the uneven distribution of benefits across regions and industries. Similarly, initial leading regions may capture a disproportionate share of the benefits of increased integration as a result of path dependence advantages. In a multi-country setting: “a country with a strong initial position in some industry may find itself with an advantage that cumulates over time.” (Krugman and Venables 1996, p. 961). A number of theoretical models have as their setting a core-periphery structure concerned with a dualism between regions within a country. Typically the backward regions host agriculture, whereas developed regions host the bulk of manufacturing activity which brings the associated benefits of industrialisation. When economic globalisation drives the formation or the strengthening of a few dynamic regions – understood these as sub-national geographic units – which are cores of economic activity, then its effects are geographically localised because regions do not benefit equally. That is, the effect of economic globalisation on the internal distribution of economic activity is related with the path of regional together with national development. Economic development efforts undertaken by national, regional or even local governments consistently focus on attracting new and/or expanding existing investment by creating the conditions for economic activity to locate in their territories. Hence while trade and investment liberalisation might offer opportunities for a number of sub-national regions they may also adversely affect other regions. When there are large shifts in the location of activity as the industrialisation policy changes from being inward-looking to being export-oriented then liberalisation and integration produce asymmetric shocks which, in turn, can have temporary or persistent impacts on the spatial structure of the economy. A body of theoretical and empirical literature where globalisation is assumed to be an external shock that affects the internal geography of national economies has been built by the Geographical Economics Theory (GE). Within GE the link between trade liberalisation and integration and the internal economic geography of countries has been dealt with assuming different contexts and diverse purposes, some from a theoretical stance (Krugman and Livas Elizondo 1996; Krugman and Venables 1996; Venables 2000; Paluzie 2001; Crozet and Koening-Soubeyran 2002; Behrens 2003; Behrens et al. 2003; Venables 2003), whereas some are more empiricallyoriented (Hanson 1998; Hanson 2000; Fujita and Hu 2001; Overman and Winters 2003; Br¨ulhart et al. 2004). The results are mixed and inconclusive in stating the actual effect of economic globalisation on the spatial structure of economies. Thus the debate called attention to the likely geographic shift that would be a consequence of a changing regime. In the Mexican case there is the standard idea of an adjustment from a centralised industrial core to relocation to the states at the US border as a consequence of the transition from import substitution to export-oriented production. However, the issue of regional imbalances is more complex than the simple centre-north relocation.

13

The aftermath of openness and integration in the Mexican manufacturing

31

Fig. 3 Traditional manufacturing hubs in Mexico. Source: Author

The dominance of Distrito Federal during the protectionist period is a well-known fact of Mexico’s economic geography. The agglomeration in Distrito Federal and, in a less degree, in the neighbouring Estado de Mexico was due to the accelerated urbanisation process and the rural-urban migratory flows. The other relatively sizeable proportion of manufacturing was located in Nuevo Leon and Jalisco which host the other two largest metropolitan areas in the country, Monterrey and Guadalajara. In contrast, the rest of the states experienced mild to poor industrial agglomeration during this stage (Sanchez-Reaza and Jordaan 2002; Sanchez-Reaza and Jordaan 2004). In 1970 Distrito Federal alone accounted for more than 32% of manufacturing. Together the four leading states accounted for more than 66%. This represented an outstanding phenomenon because these regions comprise only 12% of the country’s area (Fig. 3). At the beginning of the 1980s this picture had not changed much as there was only a minor decrease in these regions’ participation. Sanchez-Reaza and Jordaan (2002) and (2004) find that approximately 45% of total manufacturing employment was observed in Estado de Mexico and Distrito Federal during the early 1980s. With regard to the geographic patterns of manufacturing after trade liberalisation Corona J´ımenez (2003) finds that the bulk of the biggest manufacturing companies4 were located in Distrito Federal, Estado de Mexico, Nuevo Leon and Jalisco. Some geographic dispersion took place but the progress was not linear. Between 1980 and 1989 Distrito Federal’s participation fell, whereas those of Estado de Mexico and Nuevo Leon also decreased but to a lesser extent. Between 1986 and 1990 the states with the biggest increases were northern regions such as Coahuila, Chihuahua, Sonora, Tamaulipas and Baja California, but the participation of states in the Baj´ıo

4

According to the firm’s sales volume.

13

32

A. B. Trejo Nieto

such as San Luis Potos´ı and Guanajuato also increased. In the period 1990–1999 the major winners were the northern states in addition to Distrito Federal and Jalisco. Sanchez-Reaza and Jordaan (2002) and (2004) point at a northward localisation/relocalisation process of manufacturing employment. Similarly, ChambouxLeroux’s (2001) analysis of industrial employment shares shows that between 1985 and 1998 the group of states bordering the US increased their participation by 11% while regions in the centre lost 10%. Moreover, between 1988 and 1998 73% of net new manufacturing jobs were generated in only ten states, of which six were northern states and four were in the centre and Baj´ıo.5 According to Decuir-Viruez (2003) the industrial decentralisation was a shift in which economic activity moved not only towards the border but also to neighbouring states, being a movement from Distrito Federal to its peripheries. Sobrino (2003) points out that, even though they experienced a contraction in their participation, Distrito Federal and Estado de Mexico kept their places as major producers of industrial value added. Jalisco and Nuevo Leon remained at the top, whereas Coahuila, Chihuahua and Puebla joined the list of states with the highest participation in industrial activity meaning this that a localised dispersion of manufacturing took place. In the same token Corona J´ımenez (2003) indicates that the economic decline of Distrito Federal was not substantial. Between 1986 and 1990 this region lost importance, but in the period 1990–1999, and particularly in 1994, it underwent a recovery. This resurgence was reversed only in 1999. On the contrary Cravey (1998) actually argues that the transition from a state-led to a liberalised economy was facilitated by a geographical shift of industry from production sites centred in and around the capital to new sites at dispersed northern locations through maquiladoras. In the 1970s and 1980s, the geographical shift northward and the dispersion of industrial sites allowed the state to reverse industrial policy not the other way around. In the following sections we carry out a review of changes in the spatial structure of Mexican manufacturing.

4 The contemporary economic geography of Mexican manufacturing 4.1 A note on concepts and methods In dealing with the patterns and changes in the spatial structure of the economy a clarification of concepts is particularly meaningful. Traditionally, agglomeration has been associated with the clustering of economic activity or population in a particular place without making any demarcation of the geographic scale or industrial aggregation that the concept refers to. It is, therefore, a concept that requires delimitation of some kind. Agglomeration and concentration are terms that have been used interchangeably to refer to the clustering of economic activity in general. Therefore one differentiation that we make for the purpose of analytical clarity is that existing among Agglomeration, Concentration and Localisation as these concepts turn out to 5 These ten states are, in descending order, Chihuahua, Baja California, Jalisco, Guanajuato, Nuevo Le´ on, Puebla. Estado de M´exico, Tamaulipas, Coahuila and Sonora.

13

The aftermath of openness and integration in the Mexican manufacturing

33

be useful in empirical work. An example of working definitions that take account of such differentiation is that of Brakman et al. (2001) where: • Geographic agglomeration shows the distribution across space of economic activity at a broadly aggregated level, for instance manufacturing as a whole. It shows ‘how much’ or to what extent the aggregated sector is localised in a particular place. • Geographical concentration is the concept that accounts for the spatial distribution of individual industries. Thus the difference is just a matter of scale since both agglomeration and concentration account for the geographical distribution of economic activities, although their focus is different. • Localisation, on the other hand responds to a different issue. It indicates ‘where’ economic activities or industries locate. Empirical research uses a large variety of indicators to measure the geographic distribution of economic activities. Regional shares are considered a detailed indicator of the complete distribution and its use is well extended even though they present limitations (they do not provide a measurement of the overall degree of agglomeration nor they account for regional sizes). However they are useful for a first look at the spatial distribution of economic activity and to account for geographic patterns: Sij = sij/snj where sij is some indicator of the level of economic activity (GDP, employment, value added and so forth) of region i in industry j, and snj is the same indicator but at national level in industry j. The Herfindahl index shows to what extent an economic activity is distributed across a number of geographic units accordingly is an indicator of agglomeration/concentration. It is an absolute measure as it does not compare among activities. The maximum value it takes is 1, indicating a high degree of agglomeration; alternatively, values close to 1/N (where N is the number of regions) reflect a high degree of dispersion of industries across states: HERF k =

N  [Yki/Yk national]2 n=1

where Y is the gross product (alternatively value-added or employment); k = industry; N = number of regions; i = region. The Herfindahl index is among the most common measures employed in empirical research to assess geographic patterns despite the large number and variety of indices that have been developed. Ellison and Glaeser (1997) and Maurel and S´edillot (1999), for example, propose more informative indices which in addition control for the effect of plant-size. However, frequently the selection of indices is affected by the availability of geographic data. A different measure of the distribution of economic activity is production densities: Di = GDPi/Kmi2

13

34

A. B. Trejo Nieto

where GDPi is the manufacturing GDP of region i and Kmi2 is the region i surface area in square kilometres. Density indicators, by which we mean the amount of the level of economic activity per unit of territory – such as production or employment per square kilometre –, are a good approximation to geographical distribution through which one can control for the regional size effect. As we have pointed out from the early 1980s trade liberalisation, export expansion alongside privatisation and deregulation of the economy was the general approach for fostering economic growth. We use data on manufacturing GDP regional shares from the National Institute of Statistics, Geography and Informatics (INEGI) to evaluate the geographic change of manufacturing production. Annual data from 1993 to 2006 are available, but only quinquennial rather than twelve-monthly records exist from previous to 1985. The basic geographical units of analysis are the thirty one Mexican states and Distrito Federal. This choice is based to some extent on the consideration that statistical information on yearly GDP is only available at state level rather than at city or local levels. Moreover the data available allow us to compute the Herfindhal index rather than other more detailed index. We look only at agglomeration and location patterns rather than the concentration of individual industries. 4.2 A shifting geography of manufacturing? Location adjustments vs. pervasive agglomeration Figure 4 shows the evolution of regional shares within manufacturing GDP. Between 1980 and 1985, the transition period when the generalised trade liberalisation took place, the data shows that:

Fig. 4 Regional shares in manufacturing GDP in selected years (%)

13

The aftermath of openness and integration in the Mexican manufacturing

35

1. The dominant industrial state decreased its participation by almost 4% going from 29 to 25%. The other leading states Jalisco, Estado de Mexico and Nuevo Leon had modest improvement. Jointly all four states decreased by approximately 2.5%. 2. States at the US border, other than Nuevo Leon, did not perform uniformly. Chihuahua, Coahuila, and Sonora’s shares grew whereas Tamaulipas and Baja California’s decreased. Therefore their joint increase was marginal (0.75%). 3. The shares of all the states along a corridor formed by Aguascalientes, San Luis Potosi, Queretaro and Guanajuato, the so called Bajio, increased the joint improvement being of 2.4%. 4. States in the South and Yucatan Peninsula, excepting Quintana Roo, decreased their contribution to manufacturing GDP. Some of these tendencies, however, go back to at least the 1970s: 1. Distrito Federal’s participation declined from 32 to 29% between 1970 and 1980. 2. Estado de Mexico had a slight improvement. 3. Coahuila and Chihuahua, at the US border, had growing shares whereas Baja California’s participation declined. 4. Baj´ıo states, with the exception of Guanajuato, improved their participation. 5. Campeche, Guerrero and Yucatan in the south underwent declines of various extents. Interestingly, Tabasco, Chiapas and Oaxaca had increases which were likely due to an expansion of a number of industries that were linked to the processing of oil. However, like the oil boom in the late 1970s, the sudden dynamism of these industries was temporary. In the period 1993 to 2006 the data reveals a series of tendencies: 1. A declining participation of Distrito Federal, Estado de Mexico and Jalisco alongside increasing shares of Nuevo Leon after 1996. 2. Increases in most north border states with a joint improvement of 4% (Sonora and Chihuahua followed an inconsistent tendency). 3. A steadier tendency of manufacturing shares to raise in the Baj´ıo states, particularly in Queretaro and Guanajuato. 4. Conversely, all southern states remained with marginal shares in manufacturing which, in addition, were declining over the period (excepting Yucatan that had minor improvements). This first approximation to geographic changes in manufacturing gives us evidence of the relative increasing importance of the Bajio and the Border states and the falling of traditional industrial regions. Even though Distrito Federal had declining participation and other states benefited from such a process the aforementioned tendencies tell us little about the extent to which agglomeration in fact changed over time. By employing a method of analysis similar to that for convergence in per capita income we regress the relative change in regional shares on manufacturing GDP against the shares in the initial period to get an idea of an eventual catch up in GDP shares over time. If the coefficient on the initial shares is negative, it will indicate a process in which the differences between regional shares tend to decrease, which could translate into a reduction in agglomeration. Table 1 shows a summary of the estimates

13

36

A. B. Trejo Nieto

Table 1 Convergence in states’ GDP shares Constant SGDPt−11 R2 Observations

1985–1970

2006–1993

1.2855 (0.1050) −0.0183 (0.0150) 0.046 32

0.0805 (0.0049) −0.0362 (0.0879) 0.0056 32

Standard Errors in parentheses

Fig. 5 Herfindhal index of agglomeration, 1970–2006

for two broad periods, 1970–1985 and 1993–2006.6 The results show a negative coefficient indicating that regions’ shares in manufacturing GDP tended to converge between 1970 and 1985 and between 1993 and 2004. However, the coefficients are not statistically significant and the R2 s values are near zero. To check whether agglomeration levels actually showed the declining tendency that a process of convergence would imply, the σ convergence (decreasing Standard Deviation SD)7 and the Herfindhal index of geographic agglomeration were computed. However in Fig. 5 only the Herfindhal index is shown since both indicators follow exactly the same long run tendency of decreasing agglomeration. Between 1970 and 1985 as well as from 1993 onwards there is a steady declining trend but no substantial variations are noticeable. The largest changes are observed between 1980 and 1985 and between 1994 and 1995 just after unilateral liberalisation and NAFTA entered into force. Increases in agglomeration is identified only in 1997 and 2002. 6

This period are used because there is no GDP data between 1986 and 1992 and the methodology for calculating GDP in the two periods presented are different. 7 Upward changes in the standard deviation indicate wider differences in the values of a distribution. In other words, a higher SD shows that there are more disparities among observations. Therefore with a decrease in this statistic the distribution becomes more homogeneous. In the case of densities a declining SD is likely to indicate that GDP is spreading spatially among a bigger number of regions (dispersion is taking place).

13

The aftermath of openness and integration in the Mexican manufacturing

37

How can we account for such small changes in agglomeration? Although the properties of the indicators may explain this, Sobrino (2003) and others call attention to the minor contraction in Distrito Federal’s shares that has guaranteed its sustained supremacy over time (cumulative causation and initial conditions). Distrito Federal shares in manufacturing expand when agglomeration increases (1997 and 2002) whereas its reduction after 1998 is rather small. Unquestionably relocalisation of manufacturing production from Distrito Federal to other states has taken place, which confirms previous studies. However, what is striking about the results on agglomeration is summed up in a couple of stylised facts: 1. There is a persistent agglomeration of manufacturing activity. Agglomeration levels show little variation and Distrito Federal has sustained its predominance. 2. A process of geographic dispersion was having effect before trade liberalisation. States at the US border started to build an important industrial base since the 1960s. States in the Baj´ıo also began to enjoy some of the spread of manufacturing before the generalised trade liberalisation. These findings concur with Corona Jimenez’s observation that liberalising policies triggered an adjustment involving relocation to other states and a decline in Distrito Federal. However, with a new set of reforms the subsequent changes are smaller and firms’ movements toward the border become less likely. Instead, there is increasing relocation toward states in the Baj´ıo, and none major movements towards southern regions. This makes a persistent agglomeration but in different locations. The decline in the traditionally industrialised Mexican states favoured only states at the border and in the Baj´ıo. Density indicators are another way of looking at geographic differences in order to see how regions are unbalanced in the distribution of economic activity relative to their surface areas. In this case regional differences are also manifest when allowing for regional sizes since GDP densities vary significantly from state to state. Moreover between 1993 and 2006 manufacturing GDP became more geographically dense in the traditional manufacturing hubs of Distrito Federal and Estado de Mexico, and densities also increased significantly in the Bajio (Figs. 6 and 7). There are extreme cases Distrito Federal on the one hand, and Baja California Sur and Campeche on the other with approximately 28 841 and 5 thousand Mexican pesos per square kilometre respectively in 2006. Distrito Federal, Estado de Mexico, Morelos and Aguascalientes have had the biggest production densities and have generally been ranked in the same positions throughout the period 1993–2006. As a matter of fact, no major changes in the ranking occurred in the whole period. The ten most crowded states are located mainly in the centre of the country and in the Baj´ıo, with the exception of Nuevo Leon which is in the north but has been part of the traditional industrial hubs even so. The states at the US border have medium levels of GDP densities, whereas the south has mostly low-density states. Looking at the 1993–2006 period the evolution of the GDP densities standard deviation illustrates a different tendency of regional disparities. It indicates that when considering states’ territorial size dispersion of production among states was not the governing tendency. Instead an inverted U shape is observed between 1995 and 2005 (Fig. 8).

13

38

A. B. Trejo Nieto

Fig. 6 Manufacturing GDP densities, 1993

Fig. 7 Manufacturing GDP densities, 2006

In contrast to the regional shares standard deviations and the Herfindhal index we observe an overall increase in agglomeration between 1993 and 2006. Between 1995 and 2001 a solid process of spatial reconcentration of production is clear but reversed afterwards. This turnaround, however, does not demonstrate that substantial regional similarity has been reached as a result of trade liberalisation or that dispersion has been widespread. For Spain Brand (2004) finds that trade liberalisation resulted in a reorganization of industrial structures in the agglomerates of Spanish clusters. In the Spanish case the increased trade liberalisation and integration was not sufficiently strong to un-

13

The aftermath of openness and integration in the Mexican manufacturing

39

Fig. 8 Regional GDP densities standard deviation, 1993–2006 (Mexican pesos)

dermine the existing geography of production. Similarly, an important proposition about the Mexican economic geography and the adjustment after trade liberalisation and NAFTA emerges: the diminishing but persisting agglomeration of overall manufacturing is evidence of an outcome in which integration did not affected the pervasive clustering of production significantly, although it altered locational patterns. In addition, as the Distrito Federal’s predominance has been continual the argument about cumulative causation and the advantages of initial conditions is reinforced in a within-country regional setting where the supremacy of the manufacturing hub has been sustained over time.

5 The spatial configuration of manufacturing production Rodr´ıguez-Pose and S´anchez-Reaza (2003) characterised the internal territorial pattern during the ISI period as a dual economy that emerged before the ISI period. The intrinsic characteristics of the ISI model accentuated but did not create the economic backwardness of southern states, which remained relatively isolated from the whole process of industrialisation and reliant mostly on primary activities. Likewise, ISI sustained and reinforced the already existing metropolitan areas where the main industrial poles were located and which were the hub markets of the domestically protected industries. The maquiladora programme contributed to the configuration of a spatial pattern in which the regions in the north border wereincorporatedinto an industrialisationprocess. In line with the idea of duality, Bassols Batalla (1998) proposes that six out of eight Mexican regions have been en route to reaching relatively high integration with

13

40

A. B. Trejo Nieto

each other tending to form a single integrated core.8 In Bassols’ words the dual structure was originated by a longstanding chaotic developmental process in which two economic regions subsisted. Due to the particular way in which the process of economic development in Mexico operated, the biggest region kept the supremacy that was historically established even before the Spaniards conquered the country. In order to show how the spatial structure of Mexican manufacturing shifted from a ‘core-periphery’ dual geographic pattern some questions must be addressed. Is the core formed by one big homogeneous region with a unique production structure? Or, instead, can the core include various dynamic regions that cohabit geographically with secondary centres? Is there the possibility of duality within the core or within the periphery? These questions make any straightforward account debatable and controversial. The grouping of the thirty two states into broad regions can give a picture of the various alternative regional structures. Firstly, we consider the dual pattern in which the states located from the centre to the north are put together into a single region and the states in the south and the Yucatan peninsula form another region. According to this regional classification the participation of the periphery and core within manufacturing GDP remained virtually the same (about 4% the former and 96% the later); the south did not improve at all its industrial shares. The duality expressed as a core-periphery or north-south divide between an industrial hub and an agricultural periphery has historically existed and continues. The core has become a highly developed and industrialised economy compared to the south. This duality is a sign of the persistence in disparities with a clear regional and geographical character in which the south remained behind. Hence when stating that the significant move from the centre towards the north was the great contribution of trade liberalisation and integration to the geographic structure of production in Mexico, their virtues in terms of reducing regional differences and promoting spatial development were overestimated. The grounds of this assertion is twofold: on the one hand, the south was left out of the analysis on many occasions; on the other, dispersion had taken place before the trade liberalisation and so other forces were likely to be already in motion. Although it was expected that trade liberalisation would bring profound alleviation of regional disparities, the outcome is rather heterogeneous and complex. Domingues and Machado Ruiz (2005) find evidence that industrial dispersion in Brazil, like in other developing countries where high levels of spatial clustering is a common feature, has been limited to a few regions. Likewise, the announced industrial spread of Mexican manufacturing has been geographically restricted, only the traditional industrial nuclei and newly industrialised states have tended to equalise; moreover location and relocation has been among states in the core. Indeed this has become more homogeneous and its industrial geography more even.

8 These regions are Baja California, Baja California Sur, Sonora, Sinaloa and Nayarit in the northwest; Chihuahua, Coahuila, Durango, Zacatecas and San Luis Potos´ı in the north; NL and Tamaulipas in the northeast: Jal, Aguascalientes, Colima, Michoac´an and Guanajuato in the west; Queretaro, EM, DF, Morelos, Hidalgo, Tlaxcala and Puebla in the centre-east; Veracruz and Tabasco in the east. The two remaining regions are Guerrero Oaxaca and Chiapas in the south and Campeche, Yucatan and Quintana Roo in the Yucatan Peninsula.

13

The aftermath of openness and integration in the Mexican manufacturing

41

Fig. 9 Regional participations in GDP manufacturing, 1970–2006 (%)

The dual pattern should be dealt with cautiously as well because intraregional differences also exist. The economic landscape within cores and peripheries may include a mixture of diverse economic structures; geographic heterogeneity and fragmentation can be identified in the Mexican manufacturing. If allowing for an alternative setting the evidence reveal a ‘multi-agglomerate’ geographic structure, best described as a within-country ‘polycentric production structure’ where several industrial centres and peripheries are found. There are three defined geographic manufacturing ‘agglomerates’ within the core: the north-border, the Baj´ıo and the traditional hub which exist geographically with a group of peripheral states. This grouping allows us to perceive how the country’s periphery and the core’s periphery have not undergone any substantial relocalisation, whereas the decline of the traditional hub has certainly favoured the border in some periods but consistently more the Bajio states. We also see that such process does not have its origins in the generalised liberalisation or NAFTA (Fig. 9). This multi-centred context make the panorama rather complex because it implies agglomerates which can be geographically removed from each other but which have similar production densities or intensities.

6 Concluding comments In this paper we find that while the economic policy reorientation had an impact on the geography of Mexican manufacturing production some progression towards a new geographical configuration started prior to the generalised policy shift. Despite changes in industrial location the large and sustained regional differences remain and have an impact on development at a more comprehensive level. The large-scale conclusion that emerges is that the slow-moving rearrangement of manufacturing was not set off by the broad schemes of economic liberalisation. The generalised programme of export promotion strategy and liberalisation did not give rise to relocation of manufacturing and the relative increasing importance of northern and Bajio states but supported a process already in progress. Data point to an outcome in which there were minor changes in the degree of geographic agglomeration making the geographical disparities continual. With regard to this two conclusions are suggested:

13

42

A. B. Trejo Nieto

1. From a dynamic perspective there has been a tendency towards dispersion of manufacturing, yet this dispersion has been minimal. 2. From a static point of view manufacturing activity is still very much agglomerated. As relocation and declining agglomeration were ongoing phenomena they cannot be entirely attributed to the economic reforms although they seem to have had an impact. The larger fall between 1980 and 1985 may represent the adjustment after the first set of efforts to liberalise external trade was implemented. The second significant step was entry into GATT in 1986 and the concomitant broader program of liberalisation, yet we are unable to verify the effects of this period. Finally, there was also a decrease in agglomeration after NAFTA but not very substantial. Therefore economic reforms were a contributing factor for location shifts but had a limited impact in reducing agglomeration of manufacturing. A number of forces, other than institutionalised free trade and investment, are likely to have been playing a role previous to the change in economic policy. The resulting scenario is closer to an outcome where the opening up of the economy to trade and investment produced some relocation of manufacturing with a mild effect on agglomeration. Accordingly the unequal performance of Mexican states is still a reality as the trade liberalisation’s impact in promoting widespread regional development has been negligible and adverse for southern states. In contrast relocation has led to the formation of new manufacturing sub-centres in northern and Bajio states. The course that the discussion on the geographical dispersion and decentralisation of the Mexican manufacturing industry takes will also hinge on what geographical configuration we allow for. Even though the findings go some way towards explaining a widening core-periphery split or a north-south divide, on an intraregional scale regional disparities might involve far more than just a dual pattern because intraregional differences within the core and the periphery can lead to greater industrial territorial complexity. One issue of importance explaining the expansion of Bajio states is the impact of the imported component of trade relative to exports. Due to trade liberalisation resulting from changes in economic policies, a large number of firms producing for the export market generate a pull effect in which the manufacturing sector moves towards those regions in which exports are higher by reason of better access to external markets. But on the other hand, a considerable amount of foreign products flowing into the domestic market might favour the relocation towards internal regions far from Mexico’s borders in an attempt to avoid foreign competition. Further research may perhaps look at this and other issues.

References Bassols Batalla A (1998) Geograf´ıa Econ´omica de M´exico. Editorial Trillas, Mexico City Behrens K (2003) International Trade and Internal Geography Revisited. Paper presented at the Third Spring School in Economic Geography, Dijon, France Behrens K et al. (2003) Inter-regional and International Trade: Seventy Years After Ohlin. CEPR Discussion Papers 4065 Brackman S et al. (2001) An introduction to geographical economics. Cambridge University Press, Cambridge

13

The aftermath of openness and integration in the Mexican manufacturing

43

Brand H (2004) The Economic Geography Effects of Trade Liberalisation on the National Regions of Spain. European Trade Seminar Group (ETSG) Conference, Nottingham, England Br¨ulhart M et al. (2004) Enlargement and the EU Periphery: The Impact of Changing Market Potential. World Econ 27(6):853–875 Chamboux-Leroux JI (2001) Efectos de la Apertura Comercial en las Regiones y la Localizaci´on Industrial en M´exico. Comercio Exterior (Bancomext) 7:600–609 Corona J´ımenez MA (2003) Efectos de la Globalizaci´on en la Distribuci´on Espacial de las Actividades Econ´omicas. Comercio Exterior (Bancomext) 53:48–56 Cravey AJ (1998) Women and work in Mexico’s maquiladoras. Rowman & Littlefield Publishers, Inc., Lanham, MD Crozet M, Koening-Soubeyran P (2002) Trade Liberalization and the Internal Geography of Countries. Serie des Documents de Travail du CREST 2002-37:1–18 Decuir-Viruez L (2003) Institutional Factors in the Economic Growth of Mexico. 43rd ERSA Congress 2003 on Peripheries, Centres, and Spatial Development in the New Europe, University of Jyvaskyla, Finland Domingues EP, Machado Ruiz R (2005) Industrial cores and peripheries in Brazil. Textos para Discuss˜ao Cedeplar-UFMG td261, Universidade Federal de Minas Gerais, Cedeplar Ellison G, Glaeser EL (1997) Geographic Concentration in U.S. Manufacturing Industries: A Dartboard Approach; J Polit Econ 105(5):889–927 Fischer B et al. (1982) Growth, employment, and trade in an industrialising economy: a quantitative analysis of Mexican development policies. J.C.B. Mohr, T¨ubingen Fujita M, Hu D (2001) Regional disparity in China 1985–1994: The effects of globalisation and economic liberalization. Ann Reg Sci 35:3–37 Gereffi G (2003) Mexico’s Industrial Development: climbing ahead or falling behind in the world economy? Confronting development. Assessing Mexico’s Economic and social policy challenges. a. E. Z. Kevin J Hanson GH (1998) Regional adjustment to trade liberalization. Reg Sci Urban Econ 28:419–444 Hanson GH (2000) Firms, workers, and the geographic concentration of economic activity. In: Gordon MPF, Clark L, Gertler MS (eds) The Oxford Handbook of Economic Geography. Oxford University Press, Oxford, pp 477–494 Henderson JV et al. (2001) Geography and development. J Econ Geogr 1(1):81–105 INEGI. Banco de Informacion Economica, http://dgcnesyp.inegi.gob.mx/?c=72. Accessed Apr 2009 Kalter E (1992) The Mexican Strategy to Achieve Sustainable Economic Growth. In: Loser C, Kalter E (eds) Mexico: the strategy to achieve sustained economic growth. International Monetary Fund, Washington, pp 3–13 Kessel G (1995) Liberalizaci´on Comercial y Crecimiento Econ´omico. In: Rubio L, Fdez A (eds) M´exico a la hora del cambio. Cal y Arena, Centro de Investigaci´on para el Desarrollo, A. C., M´exico, pp 196– 228 Krugman P, Venables AJ (1996) Integration, specialization, and adjustment. Eur Econ Rev 40(3–5):959– 967 Krugman P, Livas Elizondo R (1996) Trade policy and the Third World metropolis. J Devel Econ 49:137–150 L´opez Villafa˜ne V (2001) NAFTA and the Dynamics of U.S.–Mexico Economic Integration. www.delfos.org.mx/proyectos/norteamerica/vlopez.pdf, Accessed 11th March 2005 Mari˜na Flores A (2004) Balance y perspectivas de la industria manufacturera mexicana tras veinte a˜nos de reestructuraci´on neoliberal: Integraci´on subordinada a Estados Unidos, desindustrializaci´on y precarizaci´on del empleo. IX Jornadas de Economia Critica, Madrid, Espa˜na Maurel F, S´edillot B (1999) A measure of the geographic concentration in French manufacturing industries; Regional Science and Urban Economics, No. 29. Middlebrock. Stanford University Press, Stanford, pp 195–240 Norris CC et al. (1999) North American Economic Integration. Theory and Practice. Edward Elgar Publishing, Gloss, UK Overman HG, Winters A (2003) Trade Shocks and Industrial Location: The Impact of EEC accession on the UK, CEE Papers. Paper No’ CEPDP0588. http://cep.lse.ac.uk/pubs/download/dp0588.pdf, Accessed 23th October 2004 Paluzie E (2001) Trade Policy and Regional Inequalities. Pap Reg Sci Urban Econ 80:67–85 Puyana A, Romero J (2006) Trade Liberalization in Mexico: Some Macroeconomic and Sectoral Impacts and the Implications for Macroeconomic PolicyConference on Post Liberalisation Constraints

13

44

A. B. Trejo Nieto

on Macroeconomic Policies. Muttukadu, Chennai, India, International Development Economics Associates (IDEAS) and United Nations Development Programme (UNDP) Rodr´ıguez-Pose A, S´anchez-Reaza J (2003) Economic Polarization Through Trade, Trade Liberalization and Regional Growth in Mexico. Discussion Paper No. 2003/60, WIDER-United Nations University Sanchez-Reaza J, Jordaan J (2004) Industrial Diversity and Specialisation in Mexican Regions under Free Trade, Centro de Investigaci´on y Docencia Econ´omica CIDE, pp 1–15 Sanchez-Reaza J, Rodriguez Pose A (2002) The impact of trade liberalization on Regional disparities in Mexico. Growth Change 33:72–90 Sanchez-Reaza J, Jordaan J (2002) Resource Endowments, Externalities and Transport Costs: An Exploration of Alternative Explanations of Regional Production Structures in Mexico. Worshop on Agglomeration and Regional Labour Markets, German Institute for Economic Research, Berlin, Germany Secretar´ıa de Econom´ıa (2004) M´exico: Negociaciones Comerciales Internacionales. M´exico, D. F., pp 1–32 Sobrino J (2003) Competitividad de las ciudades en Mexico. El Colegio de Mexico, Centro de Estudios Demograficos, Mexico City Vega G, de la Mora LM (2003) Mexico’s Trade Policy: Financial Crisis and Economic Recovery. Confronting Development. Assesing Mexico’s Economic and Social Policy Challenges. a. E. Z. Kevin J. Middlebrook. Stanford University Press, Stanford, pp 163–194 Venables AJ (2000) Cities and Trade: external trade and internal geography in developing countries. Background Paper written for the 1999 World Development Report, ‘Entering the 21st century’. London School of Economics and Centre for Economic Policy Research, London, pp 1–16 Venables AJ (2003) Spatial disparities in developing countries: cities, regions and international trade. UNU-WIDER, London School of Economics and Centre for Economic Policy Research, London, pp 1–30

13