Trade liberalization and the relative importance of domestic and foreign demand

Transcription

1 Trade liberalization and the relative importance of domestic and foreign demand to economic growth in the states of the Mexican federation Gustavo Félix Verduzco 1 and Alejandro Dávila Flores Introduction. International trade grew rapidly in Mexico between 1993 and The value of goods and services exported increased at an annual rate of per cent while imports increased on average at a rate of 9.94 per cent. The rapid removal of trade barriers in the Mexican economy encouraged foreign trade to grow at a rate far above that of the Gross Domestic Product (GDP, which grew at a rate of 2.97) and of its principal components: private consumption (3.16), government expenditure (1.32) and investment (3.15). Thus the proportion of the GDP occupied by exports and imports became steadily greater. Exports rose from 15.2 per cent in 1993 to 40.9 in 2006, while imports, which had accounted for 19.2 per cent of the GDP at the start of the period, provided 45 per cent by the end. During this period of trade liberalization, the concentration of Mexican exports in the market of the United States of America (USA) became more intense, passing from 68.8 per cent in 1990, to 84.8 in In the case of imports, there was a similar concentration at first, and 75.6 per cent of products imported to Mexico in 1996 and 1997 came from the USA. However, since then there has been a greater diversification of goods and services imported to Mexico from other places, and the figure for goods imported from the USA rather than from any other place of origin has gradually come down, to 51.1 per cent in Researcher at the Center for Socioeconomic Research of the Autonomous University of Coahuila, and candidate to the National System of Research in Mexico (SNI). 2 Visiting Scholar in the Department of Economics and Senior Fellow at the Center for North American Studies, American University. Professor at Autonomous University of Coahuila and Member of the National System of Research in Mexico (SNI) 1

2 The greater relative importance of exports to the GDP, and the high degree of concentration of these in the United States market, has strengthened the correlation between the economic cycle of the USA and the evolution of the GDP in the Mexican economy. 3 The interest of the present research paper is basically in three questions that derive from this general appreciation of how the Mexican economy has evolved: 1. Is this new stage of development common to all the states of the Mexican federation? 2. If not, is there then a pattern differentiated according to territory? 3. And if there is a territorially differentiated pattern: which would be the determining factors? 2. Trade policy and the localization of productive factors. Inspired as it happens by the case of Mexico, Krugman and Livas (KL) 4 (1992) developed a theoretical model of economic geography. They concentrated on the relation between trading policies and the localization of economic activity in developing nations. According to these authors, the great concentration of production factors in the big urban zones of developing nations after the Second World War was an unintended result of the industrializing strategy that based on import substitution. Supported by an arsenal of instruments for commercial protectionism, especially duties and other restrictions on foreign trade, the policy of import substitution guaranteed the companies 3 In the Economic Policy Criteria for 2007, of the Mexican Tax Office, the coefficient between the GDP of Mexico and that of the United States is set at over 0.8, a much higher figure than that observed between 1985 and 1995, when it rose to Secretaría de Hacienda y Crédito Público, Criterios de Política Económica para el 2007, SHCP, contenido/documentos/subtema2/cgpe_2007. Pdf, México, p Krugman, P. and Livas, R., Trade Policy and the Third World Metropolis, NBER Working paper series, number 4328, MIT, 1992, USA, pp

3 established in national territory virtually exclusive access to the domestic market of goods and services, while at the same time limiting their sales abroad. For KL, the localization of production involves a tension between economic forces propitiating agglomeration (centripetal forces) and those fomenting territorial dispersion (i.e. centrifugal forces). In the first case, it is principally the interconnections between economies of scale, size of market and transport costs that generate economies linked to strengthening productive chains. Whereas the forces that most make for the dispersion of productive factors are the costs of travel between home and workplace, or else the price of land rents deriving from the distances between home and workplace. 5 As long as the agglomeration economies of a city offset the greater costs generated by a spatial concentration of productive factors, there will be incentives for generating accumulative processes of economic growth. 6 Following the same logic, where there are wage and land rent differentials, there are likely to be agglomeration economies. Such differentials must grow in line with any 5 This way of approaching the subject is very like that of the German school of localization theory, whose logical structure is based on an implicit consideration of economies of scale and an explicit concern with transport costs; Von Thünen (1826), Weber (1929), Christaller (1933) and Lösch (1944) are among its principal exponents (see Richardson, Harry W., Regional Economics, University of Illinois Press, 1979, USA and Keilbach, Max, Spatial Knowledge Spillovers and the Dynamics of Agglomeration and Regional Growth, Physica-Verlag Heidelberg, 2000, Germany, pp ) The KL model is better to the extent that it has the advantage of a more sophisticated mathematics, and makes use of advances in economic theory, especially those to do with the functioning of imperfect markets. 6 Ohlin (1933) subsumes the factors that make an economy tend to concentrate in one place under the term economies of agglomeration. These may be of three types: 1) of scale, referring to those agglomerations that benefit directly from the companies that form them; 2) of localization, that is agglomerations forged by the concentration in one place of establishments from a single industry, and 3) of urbanization, agglomerations deriving from the size of the local economy (quoted in Keilbach, op. cit.). When the function of production or the function of profit of an economic agent is affected, positively or negatively, by external economic agents, economic theory speaks of the existence of an externality. Externalities are classified as technological (when they are not necessarily transmitted by market mechanisms) and pecuniary (those propagated by the system of prices). The principal mechanism for the transmission of externalities is considered to be spatial spillovers of knowledge. There are two types of externality: 1) the Marshall type (1920), linked to the productive specialization of a city on one particular industry, and 2) the Jacobs type (1969), which are the result of the variety of products and technologies that exist in one location. How are externalities related to economies of localization and urbanization? Those distinguished by Marshall are external to the firm but internal to the industry, which links them to economies of localization. However, urbanization economies may appear in a highly specialized or widely diversified local economy. As Keilbach observes (op. cit.), although both agglomeration economies and externalities refer to the phenomenon of the spatial concentration of factors, the two terms have different meanings. 3

4 increase in transportation costs between the industrial center or centers and other places in the country. In the case of Mexico, foreign trade liberalization allowed producers established in national territory to increase the size of their market, and it facilitated the development of a second industrial center in the states on the northern border. New market opportunities propitiated increasingly rapid changes in the structures of aggregated supply and demand, altering the previous balance between centripetal and centrifugal forces. This created conditions favorable to the localization of new companies in regions with easy access to the new center and also to a territorial reaccommodation of production factors, especially in footloose industries. 7 In short, the KL model allows two predictions to be made about the impact of trade policies on the patterns of localization of manufacturing employment in developing countries: 1. In an economy with high foreign trade barriers and a comparatively narrow internal market, agglomeration economies will compensate for the disadvantages of urban congestion, and produce a pattern of highly concentrated manufacturing employment. 2. The removal of trade barriers will weaken the impact of agglomeration economies on the spatial concentration of manufacturing employment in the industrial center or centers of the country that were formed under protectionism. As already mentioned, in the Introduction, the strategy in Mexico of removing trade barriers increased the part played in the GDP by imports and exports, and altered the structures of aggregated supply and demand. These transformations coincide with a 7 In terms of inputs, production factors, and products, economic activities are classified as: 1) intensive in the use of mobile factors (footloose industries); 2) producers of non-traded goods, and 3) those making intensive use of natural resources. The latter tend to localize close to exploitable deposits; the non-traded industries normally set up near their consumers; and footloose industries tend to concentrate in the largest urban areas. 4

5 greater concentration of Mexican foreign trade, in transactions with the USA. To this flourishing foreign market should be added the considerable purchasing power of economic agents established in the states of the north and in the north of the central part of Mexico. The expansion of business opportunities for companies located in Mexico during the period studied came, basically, from the USA economy. It would therefore be reasonable to expect: 1. A higher rate of growth of the GDP in those states in the north of the country that share a border with the USA, or in those located in the hinterland between the two industrial destinations most important to producers established in Mexico, namely Mexico City and the USA; 2. For the same reasons, it would be logical to anticipate that in these states, external demand will have a greater impact on the growth of the GDP than in other states of the federation. 3. The removal of trade barriers and the regional dynamics of the GDP in Mexico. In order to try and establish the relation between the rate of growth of the GDP and the distance of industrial centers from the USA a graph was drawn using Cartesian co-ordinates. The ordinates drawn on the x-axis represented first the distances from each state capital to the nearest USA border crossing 8, and then to Mexico City. The abscissa marked on the y-axis represented the average real growth rate of the total GDP, the GDP for manufacturing, the GDP per capita (both for GDP overall and for GDP in manufacturing), the sums of direct foreign investment channeled to each state during the 8 The distances considered were to the principal land crossings: Tijuana, Mexicali, Nogales, Ciudad Juárez, Piedras Negras, Nuevo Laredo, Reynosa and Matamoros. The source of information was: Guía Roji de México, México Atlas Turístico de Carreteras , Guía Roji, S. A. de C. V., 2005, México, p

6 period, and finally, the differences recorded in the GDP index between GDP per capita most recently observed (2004) and the figure first observed (1993) 9. As expected, there was a negative result for the coefficients of correlation between distance from the border and product variation, as also for coefficients of correlation between distance from the border and product per inhabitant (both with respect to GDP overall and to the figure for only the manufacturing sector). Figures 1 and 2 below illustrate the results of this exercise for all aspects of the GDP. Figure 1 6% Average Rate of Real Growth 5% Baja California 4% Nuevo León Coahuila Sonora Tamaulipas Chihuahua San Luis Potosí Querétaro Guanajuato Aguascalientes Durango Tlaxcala Puebla State GDP Rate of Growth ( ) vs Distance of Capital City to Northern Border. Correlation Coefficient= R2=.1108 Baja California Sur Quintana Roo Yucatán 3% Zacatecas Michoacán Jalisco México Colima Chiapas Campeche 2% 1% Veracruz Hidalgoo Distrito Federal Morelos Sinaloa Guerrero Nayarit Oaxaca Tabasco SOURCE: INEGI, CONAPO and Guía Roji. 0% Kilometers ,000 1,200 1,400 1,600 1,800 2,000 As in those bordering on the USA, there was considerable growth in several states in the north of the central region of the country, especially in Aguascalientes, Querétaro and Guanajuato. Figures are also good for traditional tourist destinations such as Baja California Sur, Quintana Roo and Yucatán. The state of Yucatán, which is very close to the USA market by air and sea, though not by land, also shows relatively high rates of growth. The opposite can be seen in certain states of the center and south (Federal District, Hidalgo, Veracruz, Morelos, Nayarit, Guerrero, Oaxaca and Tabasco). 9 The index is made by relating the GDP per capita of each state to the national average, using projections of the population from , elaborated by CONAPO (Comisión Nacional de Población 6

7 Figure 2 3.0% Average Rate of Real Growth Guanajuato Aguascalientes Coahuila Querétaro Zacatecas 2.5% Tamaulipas Durango Nuevo León San Luis Potosí Michoacán 2.0% Chihuahua State GDP per capita. Rate of Growth ( ) vs Distance of Capital City to Northern Border. Correlation Coefficient= R2= Puebla Yucatán 1.5% Sonora Tlaxcala Distrito Federal Baja California 1.0% Veracruz Hidalgoo Jalisco Colima Morelos Chiapas Campeche 0.5% Sinaloa México Guerrero Nayarit Oaxaca Tabasco Baja California Sur Quintana Roo 0.0% Kilometers ,000 1,200 1,400 1,600 1,800 2,000 There are three explanations for these tendencies: greater activity of manufacturing for exports, localized mainly in the north of Mexico and in the hinterland between the principal internal market and the USA market; less activity of manufacturing for the internal market, localized principally in the states of the central part of the country, and; the great distance and the separation of the states in the south from Mexico City and the points of entry into the USA on the northern border. 10 The population of the states of Quintana Roo and Baja California grew at much higher annual rates (4.27 and 3.46) than the national average (1.4). For this reason, when estimating the correlation between GDP per capita and the distance from the border, both figures appear below the line showing the overall tendency (Figure 2). The correlation sign is still negative and its value, like that of the coefficient of determination (R 2 ), is greater than that of the relation seen in the first figure. Partly as a result of emigration from the states of Zacatecas, Durango and Michoacán, their case is 10 Deficiencies in logistical and transport infrastructure increase the impact of simple physical distance. The relation between the two variables was also inverse in the case of the manufacturing sector, with coefficients of correlation ( ) and of determination (R 2 =0.1754) that were greater than in the previous case. 7

8 different, and they showed a lower population growth (0.41, 0.74 and 0.76, respectively). 11 The hypothesis that increased activity is associated with the greater proximity of different regions to the USA market, and lessening activity with regions having stronger links to the old industrial center of import substitution times, is supported by a consideration of the relation between growth rates for the GDP and the distance from Mexico City (Figure 3). The values of the coefficients of correlation and of determination are low (0.29 and an R 2 of ) and the sign is positive (the greater the distance, the greater the growth of product). Once again a good many of the states of the north and of the northern part of the center appear above the line representing the overall tendency, whereas the opposite applies to the Federal District, the states of the center, the states of the south and two of the states of northern part of the centre (Nayarit and Sinaloa, which have Pacific shorelines). Figure 3 6% 5% 4% 3% Tasa media de crecimiento anual real Aguascalientes Querétaro Tlaxcala 2% HidalgooVeracruz Distrito Federal Guerrero 1% Coahuila Tamaulipas Guanajuato Nuevo León Puebla San Luis Potosí Durango México Michoacán Zacatecas Morelos Tabasco Nayarit Quintana Roo Yucatán Campeche Jalisco Colima Chiapas Oaxaca Sinaloa Chihuahua Crecimiento del PIB ( ) vs distancia a Ciudad de México. Coeficiente de correlación= 0.29; R2= Sonora Baja California Baja California Sur FUENTE: Elaborado con información del INEGI, CONAPO y la Guía Roji. 0% Kilómetros ,000 1,500 2,000 2,500 3,000 3,500 4,000 4,500 Figure 4 shows the GDP index per capita for each state of the federation for the years 1993 to For the most recent year it was possible to review (2004), the 11 In the case of GDP per capita, the coefficients of correlation ( ) and of determination (R 2 =0.2121) for the GDP of manufacturing, with respect to distance from the northern border, were smaller than those of the GDP overall. 8

9 figure for the Federal District (2.45) was the highest, while Oaxaca (0.41) came at the very bottom of the list. In effect, the average income for an inhabitant of the Federal District was nearly six times that of an inhabitant of Oaxaca. The figure for inhabitants of the Federal District capital stood at a point 145 per cent above the national average, while the figure for people in Oaxaca was at a point 59 per cent below the national average. Figure Mexican States: GDP per capita index y Distrito Federal / Oaxaca = 5.35 (1993) 5.93 (2004) National Average SOURCE: INEGI. Finally, Figure 5 shows the relation between variations in the GDP index per capita and the distance from the northern border. Of the seven states with a positive difference of 0.1 or more, three are from the north-east (Coahuila, Nuevo León and Tamaulipas), and four are from the north of the central region of Mexico (Aguascalientes, Querétaro, Guanajuato and Durango). With the exception of Guanajuato and Durango, the population in these states grew at a higher rate than in the country as a whole. Of the states which come in below the overall tendency, most are in the center and the south of Mexico, but there are also three from the north (Baja California, Sonora and Chihuahua), which have a higher rate of population growth than the rest of the country, as Quintana Roo does, also. 12 The GDP figures for states of the federation were obtained from INEGI ( mx/). They were standardized to a common base value of the peso in

10 Figure Variation on the GDP per cápita index Nuevo León Coahuila Aguascalientes Querétaro Variation on the GDP per capita index of Mexican States vs distance to northern border. ( ) Coefficient of correlation=-0.626; R2= Zacatecas Tlaxcala Chihuahua Yucatán Michoacán Puebla Sonora San Luis Potosí Distrito Federal Colima Chiapas 0.0 Distance: Kilometers Veracruz Jalisco Oaxaca ,200 1,500 1,800 2, Baja California Tamaulipas Guanajuato Durango Hidalgoo Guerrero Sinaloa Morelos Tabasco México Nayarit Baja California Sur Campeche Quintana Roo -0.3 SOURCE: INEGI, CONAPO and Guía Roji. The figures therefore seem to support the first hypothesis derived from the KL theory; the removal of trade barriers in the Mexican economy propitiated a greater growth rate for the GDP in the northern states sharing a border with the USA, as well as in some states located between the northern border and Mexico City. The next question is, to what extent these changes are due to the different weights of internal and external demand as determinants of growth in the states of the Mexican federation. In the following sections, tools of econometric analysis will be used to try to find out. 4. Relative sizes of domestic and foreign demand in the state economies of the Mexican Federal Republic. Because with the removal of trade barriers the objective market of companies is diversified and because the USA is the main trading partner of Mexico, it becomes a matter of interest to identify the relation of the economic performance of the regions to the dynamics of the internal market and to those of the USA market for goods produced in Mexico. In this section we review the dynamics of the Gross Domestic Product of the States of the Federation (GDPS) for manufacturing and of the GDPS overall. 10

11 The hypothesis is that regions with greater access to the USA will respond better to changes in that market; whereas those regions which are more linked to the domestic economy, will respond more strongly to what happens in the domestic market. To show the differences, we propose the following general econometric models, which represent the stochastic relation between the GDPS and the dynamics of the two markets: (1) Lipibe3 it = 0 + 1*Llipib3s3e it + 2*Lipibus t + U t (2) Lipibet it = 0 + 1*Llipibste it + 2*Lipibus t + t With i= 1,, 32 and; t= 1993, In the first equation (1), Lipibe3 it is the logarithm of the real manufacturing GDP (at 1993 prices) for each state. For all indices, 1993 is the starting year and in this equation it takes one as the base value. Llipib3s3e it is the logarithm of the aggregated index of the GDP for manufacturing in the other states of the federation (i.e. excluding the state for which information is being taken into account); Lipibus t is the logarithm of the GDP rate of the USA and; U is the term of contingent disturbance under the usual assumptions. Similarly, in the second equation (2), Lipibet it is the logarithm of the GDPS rate overall; Llipibste it is the logarithm for the overall GDPS of the other states; Lipibus t the logarithm of the GDP of the USA; and is the term of contingent disturbance. The study was of a series of data from all 12 of the years for each of the 32 states of the federation (including the Federal District and for the GDP of the USA. 13 As a technique for estimates, panel data models with fixed and with contingent effects were contrasted one with another, with and without self-correlated contingent deviations. The Haussman test allowed us to use the model with contingent effects and 13 Information for Mexico comes from INEGI, data for the USA from the OECD. 11

12 the test of Bhargava et al. (1982) 14 suggested a structure of self-correlation AR(1) in the deviations. 15 Four regression equations for each model were evaluated using Chow contrasts, 16 each one contrasting a hypothesis to explain the difference of the regression parameters with respect to the geographical heterogeneity related to the emphasis on one market or the other. Thereby equation (1) proposes an equality of coefficients for the 32 federal entities; equation (2) proposes differential coefficients for the six states bordering the USA and those without a border to that country; equation (3) contrasts a difference in coefficients depending on which states had greater and which had lesser direct foreign investment (DFI) 17 and; equation (4) proves the difference in coefficients between two groups of border states, those of the north-east (Coahuila, Nuevo León and Tamaulipas) and those of the north-west (Baja California, Sonora and Chihuahua), while at the same time contrasting the effects of the maquiladora export industry at the end of the year 2000, through the use of temporary dummies. Figures 6-11 show the dynamics of the GDPS for manufacturing and overall. Performances that accord with the hypothesis are distinguished, and the contrasts are proposed. The Border States in the north-west show their high dependence on the export maquiladora export industry (MEI); a fall in their figures coincides with the crisis of employment in the industry at the end of This behavior should not come as a 14 Bhargava, et al., Serial Correlation and the Fixed Effects Models, Review of Economics Studies, 49, 4, , The proof statistic has a value of , thus the hypothesis can be rejected that the rho parameter associated with the AR(1) process is equal to zero. 16 All the tests are available upon request from the authors. 17 The geographical reference for DFI should be taken as an approximation, as in official records it is determined by the address of the legal representative or that of the administrative office of each company. The amount of DFI can therefore be overestimated for some states (for example, the Federal District) and for others it would be underestimated. Bearing this in mind, states with the greatest amounts of DFI were considered to be those in which the accumulation of DFI in the last five years was greater than the average minus 30% standard deviation (excluding the Federal District to make the calculations). These states are, then: the six states in the north with a USA border, plus Aguascalientes, Baja California Sur, Guanajuato, Jalisco, Estado de México, Puebla, Querétaro, San Luis Potosí and the Federal District Mexico City. 12

13 surprise as, according to INEGI information, just over 50 per cent of employment in the MEI is concentrated in Baja California, Sonora and Chihuahua, while at the same time the employment in maquiladoras in these states accounts for a greater proportion of employment in manufacturing than elsewhere in the country (75.8 per cent in the year 2003). 18 By contrast, the border states of the north-east show a dynamic of greater growth, as distinct from elsewhere in the country, and, furthermore, less dependent on the maquiladora industries: the tendency in Nuevo León is unaffected by the maquiladora crisis; Coahuila springs back in 2001 and Tamaulipas in 2001 and This behavior is consistent with the smaller amount of employment in maquiladoras in these states. Figures 6, 7, 8, 9, 10 y 11 Indice del PIB Manufacturero: estados fronterizos, Mexico y EUA (total) 1993= FUENTE: Entidades de México, elaborado con información del INEGI; los de EUA con datos de la OECD BC Coah Chih NL Son Tam Mx Mx_Sfr EU 18 For the country as a whole the proportion was 25.2 per cent, and for the Border States as a whole, 54.9 per cent: Félix, G., Localización y Demanda Laboral en la Industria Maquiladora de Exportación, working paper, CISE-UAdeC,

14 Indice del PIBE manufacturero: otros estados con mayor IED, 1993= Ags Bcs DF Gto Jal EMex Pue Qro SLP Fuente: Elaborado con datos de INEGI Indice del PIBE Manufacturero; resto de entidades, 1993=1 Fuente: Elaborado con datos de INEGI Cam Col Chia Dgo gro Hgo Mich Mor Nay Oax Qroo Sin Tab Tla Ver Yuc Zac Indice del PIB: estados fronterizos, Mexico y EUA (total), 1993= FUENTE: Entidades de México, elaborado con información del INEGI; los de EUA con datos de la OECD BC Coah Chih NL Son Tam Mx Mx_Sfr EU 14

15 Indice del PIBE: otros estados con mayor IED Ags Bcs DF Gto Jal EMex Pue Qro SLP Fuente: Elaborado con datos de INEGI. Indice del PIBE; resto de entidades, 1993= Cam Col Chia Dgo gro Hgo Mich Mor Nay Oax Qroo Sin Tab Tla Ver Yuc Zac The 6 Border States also form part of the group of 15 receiving the most direct foreign investment (DFI). The other 9 states in this group, presented in figures 7 and 10, show a tendency that is clearly not linked to the MEI and within this sub-category a group of more dynamic non-border states can be distinguished, which stand out at national level (Guanajuato, Aguascalientes, Querétaro, Baja California Sur and Puebla) as well as another group, of less dynamic non-border federal entities (the Federal District, Jalisco, Estado de México and San Luis Potosí). 15

16 The other states, with smaller amounts of DFI, show a generally poor economic performance; not so bad in Yucatán, Tlaxcala, Michoacán and Durango, but very worrying in Nayarit, Chiapas, Tabasco and Campeche (Figures 8 and 11). 5. Econometric results of the dynamics of the GDPS for manufacturing. The hypothesis of this research project would lead one to suppose that the regions most linked to the USA market will show regression coefficients that are higher for the Lipibus variable; and one might expect to find something similar for the variable showing how the domestic market has developed. The Chow contrasts mentioned above show that the models adjust better when geographical differences are proponed in the regression coefficients. 19 To revise the results of the estimations in greater detail, we start with those that correspond to the GDPS for manufacturing, and will then consider those for the GDPS overall. The results of equation (1) suggest that, on average, the manufacturing industry of the federal entities responds positively both to manufacturing impulses in the domestic market and to those in the USA market. On average, the link to the domestic market is 2.1 times higher than it is for the USA (See Table 1). When the distinction is made between Border States and other States, in the columns corresponding to equation (2), it is possible to observe on the horizontal plane how the effect of the dynamics of the USA economy on the manufacturing of the regions increases, with the coefficient associated to the dynamics of the USA economy 2.2 times higher in the Border States in the north. Also, those states respond better to manufacturing impulse in Mexico. This result is possible because of the high 19 The contrast consists of elaborating a statistical proof which is compared to the critical value of a contingent variable having a distribution F. The statistic is constructed from the sum of the errors squared of both the restricted model (combined data) and the unrestricted model (equations for each geographical variant). The details of the proof are available upon request from the authors. 16

17 heterogeneity of the non-border states, where there is a different degree of integration to domestic and USA markets. Table 1 Panel Data Model with random effects and autocorrelated AR(1) disturbances. Dependent Variable: Logaritm of Manufacturing GDP index of regions (1993=100) Variables Constant mymx myus mymxd myusd Model 1 Model 2 Model 3 Model 4 + All No Border Border Low FDI High FDI Northest Northwest **0.664 **0.315 **0.590 ** **0.632 **0.253 **2.237 ** **0.830 ** ** **0.920 ** **0.604 ** *0.418 ** R 2 Whitin R 2 Between R 2 Overall Wald n States t observed values in parentheses. **, *: significant at 5% and 10%. + For this estimation, mymxd and myusd are the correspondent variables combined with a dummy with value of "0" for the years before 2000, and "1" for the years of 2000 and after. Equation (3) distinguishes between states with more and states with less DFI. Among the first are the 6 Border States in the north, and 9 other federal entities. One would expect these 15 states to respond more to the dynamics of the foreign market, and this tendency is confirmed by the regression results. The states receiving most DFI are more linked to the dynamics of the USA economy and to those of national manufacturing. Whereas for the states with the least direct foreign investment, the coefficient associated to the dynamics of the GDP of the USA is not statistically different from zero (See column 6 in Table 1). Equation (4) distinguishes between the coefficients for the border states of the north-west and those for the north-east. It is important to make the distinction because, as noted in the previous section, the dynamics of manufacturing in the north-west are more closely bound to any decline in the maquiladora export industry. From the year 17

18 2000 onwards maquiladoras delocalized from these Border States in the north-west, and because DFI is prominent in their economies, changes to the amounts of DFI received by these states may induce structural changes in their economies. To catch the possibility of this effect, the regression corresponding to these two regions was divided into the period before these changes were observed ( ) and the subsequent period ( ). To do so, dummy variables were used iteratively. The manufacturing of the north-east is more closely bound to the USA economy than it is to the manufacturing sector in the rest of the country (See Table 1). The relation remains virtually unchanged during the second period ( ), and this confirms the hypothesis that this region has suffered the crisis in the maquiladora export industry less intensely. On the other hand, the results for the north-west region show an important structural change. Up until 1999, there is a significant systematic dependence of regional manufacturing on the USA economy. The presence of many maquiladoras in the region might lead one to suppose that the dynamics of the MEI would respond more intensely to variations of demand in the USA market. However, from 2000 onwards considerable delocalization of MEI companies was observed from this region to countries with lower labor costs. As a result, the dynamics of the maquila now respond more to the competition of lower wages in other countries, and not so much to demand in the USA market (at least, systematically). The figures suggest that the north-west region is undergoing an important structural change with regard to its links to the USA economy. 6. Econometric results of the GDPS overall. The next step was to study the dependence of the dynamics of the state GDP overall, on the USA economy and on the national economy. The procedure was like that 18

19 used for the GDPS for manufacturing; that is, models for different geographical groupings were contrasted with regard to the possible degree of linkage to one or the other market. For any regional level, when we compare the results of Table 2 with those for manufacturing (Table 1), we can see that on average the regional linkage to the foreign market is lower for economic activity in general and considerably higher with respect to the domestic market. This finding is logical as in the case of the GDPS overall there is a relatively greater presence of non-trade economic activities. Table 2 Panel Data Model with random effects and autocorrelated AR(1) disturbances. Dependent Variable: Logaritm of GDP index of regions (1993=100) Variables Constant ymx yus ymxd yusd Model 1 Model 2 Model 3 Model 4 + All No Border Border Low FDI High FDI Northest Northwest **0.784 ** ** **0.768 **0.185 ** **0.886 ** **0.661 ** **0.925 ** **0.852 ** **0.921 **0.471 (0.013) (0.061) (0.063) (0.640) (0.547) (0.014) (0.068) (0.070) (0.943) (0.812) (0.011) (0.102) (0.102) (0.014) (0.081) (0.082) (0.016) (0.082) (0.083) (0.014) (0.140) (0.135) (0.019) (0.167) (0.168) R 2 Whitin R 2 Between R 2 Overall Wald n States t observed values in parentheses. **, *: significant at 5% and 10%. + For this estimation, ymxd and yusd are the correspondent variables combined with a dummy with value of "0" for the years before 2000, and "1" for the years of 2000 and after. Equation (1) in Table 2 shows that the state economies react to the impulse from both markets, but as noted before, they respond more to what happens in the domestic market. In concrete terms, it may be appreciated that the economies of the different regions of the country receive a greater impulse from the domestic economy than from the foreign (3.1 times more, in fact; and 2.1 times more for manufacturing). 19

20 The states in the north, with a border to the USA, show a stronger link to both markets than that observed for non-border states (See equation (2) in the same table). It is also clear that there is a relatively stronger link to the USA market in the Border States. The same tendency is seen in equation (3): greater linkage to both markets by the regional economies with the highest levels of DFI. Columns 6 and 7 of Table 2 show the results of regressions for the Border States of the north-east and the north-west. Once again, dummy temporal variables were introduced repeatedly for the year 2000 and after. For total economic activity, the results confirm a greater linkage of the states in the north-east to the USA economy. For the years 2000 and after the effect of the growth of the GDP for the USA on the growth of the GDPS of the north-east region is greatly strengthened, indicating the possibility of greater interdependence between the region and the USA economy. As in the case of manufacturing, the systematic link of the GDPS in the northwest region to the USA economy appears to have weakened. This is the combined result of: a) the importance of the MEI in the region, and b) the existence, from 2000 onwards, of the possibility of a structural change, deriving from the lesser wage efficiency of the region compared to Mexico s principal international competitors (China, India and Central America), which has manifested itself in a certain amount of delocalization of economic activity in favor of these other nations. 7. Conclusions and final remarks reflections Regional economies dedicated to exports (those in the north with a border to the USA and those with the highest amounts of DFI) show higher coefficients of linkage to both markets. The other regions show little or no relation to the USA market and weaker articulation to the national economy itself. As a result, the economies of these states are more dependent on their local markets. 20

21 The economies of the border states of the north-east react more intensely to changes in demand from the other side of the Rio Bravo than they do to changes in demand from the national economy. This development is even clearer in the manufacturing sector. Because of its great dependence on the MEI, the north-west region appears to be undergoing a structural change, as its dynamics since 2000 have responded to a large extent to delocalization decisions following the loss of their wage competitiveness, a circumstance that undermines their presence in the USA market. In spite of trade liberalization, the economies of the other states of the republic are more sensitive to the impulse from the domestic market and less to the performance of the USA economy. In short, with the one exception of the north-est of Mexico, the domestic market continues to have a greater specific gravity than the development of foreign demand does in the economic growth of most states of the federation and that of the country as a whole. It is particularly important for defining regional development strategies in a world economy, to distinguish the orientation of markets that affect the dynamics of demand to companies, and thus the regional economies where they are sited. In an article on the economic performance of regions in the USA, Michael Porter 20 provides data and conclusions that should be taken into account. He finds that just under a third of the employment in their economies (31.8 per cent), is generated by export companies. The criteria for localization are therefore strongly determined by the degree of competitiveness in the places selected, which leads to a geographical concentration of firms looking for the positive externalities of agglomeration economies. Whereas non- 20 Porter, Michael, The Economic Performance of Regions, Regional Studies, Vol. 37.6&7, 2003, USA, pp

22 trade economic activities are directed to supplying local markets, a circumstance that makes their spatial distribution one determined by proximity to the population. Finally, activities of exploiting natural resources are sited in the vicinity of corresponding reserves. According to Porter, even though the number of people employed in export activities amounts to less than a third of all those employed, they have better wages and a higher rate of productivity and innovation than are found in other industries and they have a great influence on the prosperity of their region, as their competitive success stimulates the demand for local industry. In his study, Porter demonstrates the existence of a strong association between the average wages in the region and the average wage paid in export industries, with the former reaching 66 per cent of the latter. In non-trade activities, wage levels, productivity and innovation capacity were lower, in spite of the fact that these activities generated over two thirds of employment (67.4 per cent). While companies linked to the exploitation of natural resources had only a marginal participation in employment (0.8 per cent) and had an intermediate level of wages, productivity and innovation. On the basis of these findings, Porter (2003, op. cit.) concludes that it is more important for a region to develop the conditions that allow higher wages in export industries, than it is to increase the coefficient of regional exports. Taking these tendencies into consideration along with the findings obtained in the present work, a series of guidelines can be derived for inclusion in the design of regional development policies: 1. Planners should identify the economic activities in a region that go to make up its export clusters, and consolidate their presence in world markets by fostering their competitive capacity. 22

23 2. The exportable base of a region should be exploited to the maximum in order to optimize its multiplying effects on the local economy. To this effect it should be possible to structure policies of development for suppliers and policies to promote investment that incorporate new tiers in local productive chains, or that allow the aggregated value of those already in existence to increase, in such a way as to raise the regional coefficients of supplying intermediate inputs. 3. It is possible to apply a policy of productive diversification to limit the fragility of a region subject to external shocks, without compromising its competitive viability. To achieve this, it makes sense to explore strategies of regional diversification for clusters supported by economic activities that are simultaneously linked to more than one economic grouping. For example, the productive branch producing articles of rubber and plastic may be a provider, at the same time, to various clusters (automotive, electronics and textiles). These branches of activity, which may be termed hinge, can serve as the beach head of the impulse to develop new clusters. 4. Policies to stimulate competitiveness based on clusters, should not be restricted to export activities, as they can also encourage innovation and the economic growth of activities producing non-traded goods. 5. Policies of infrastructure, training, temporary employment and fighting extreme poverty should be linked to the strategies for improving the competitiveness of local clusters and to those for diversifying their economies. 6. Two of the principal competitive advantages of Mexico are its geographical position and its natural resources, especially of energy. However, they are most absurdly wasted. In the first instance, deficiencies in the transport and telecommunications infrastructure make these services unnecessarily costly. Secondly, 23

24 the public monopolies of the sector (PEMEX, CFE, and Luz y Fuerza del Centro) have inexcusably low levels of efficiency. 7. Strange though it may seem, in the body of the document entitled Informe Anual del Banco de México 2006 (the annual report of Mexican central bank), there is not a single reference to the evolution of the real exchange rate of the Mexican peso. In fact, as Table A 14 of the statistical annex shows 21, for the fourth quarter of 2006 this index is , giving the Mexican peso a rate of real appreciation higher than that observed in 1993 and The appreciation of the real exchange rate has had a devastating effect on the wage competitiveness of the MEI, as on that in shoes manufacturing and textiles. In this last case, the strengthening of Mexican peso explains almost entirely the overpricing of a square meter of Mexican textiles in comparison with those made in China sold in the USA market. 23 The moment has arrived for undertaking a thorough revision of the impact of monetary policy on the appreciation of the exchange rate, on the relative prices of national versus foreign products, and the relative loss of competitiveness of producers established in national territory. 21 Banco de México. (2007), Informe Anual 2006, Abril del 2007, México, p According to the report of el Banco de México (Op. cit., p. 106), the real effective exchange rate is calculated on the basis of prices to the consumer with respect to a basket of 111 countries, weighted with the GDP of each of them. The base of this index is the year If the real exchange rate of the Mexican peso is measured against the American dollar, the figure for the index is 81.7, a rate almost exactly the same as that of November In making this comparison, the base of the index was November Between January and September 2006, the average price per square meter of textiles exported from Mexico to the USA market, was of 3.58 dollars, 26 per cent higher than the 2.84 dollars of equivalent Chinese material ( Eliminating the overvaluation of the peso, estimated to be 22.4 per cent, would allow the price of a square meter of Mexican textiles exported to the USA to come down to 2.92 dollars, not very far off the Chinese price. According to experts, to eliminate the undervaluation of the Yuan, its value would need to be reduced by 38 per cent. This would raise the price of Chinese textiles to 3.92 dollars, which would make them 34.2 per cent more expensive than the revalued Mexican equivalent. In spite of the fact that Mexican labor would still cost twice as much as Chinese, if both currencies were revalued, and costs three times more with the two currencies as they are. 24