Report No. 19864-ME Mexico Export Dynamics and Productivity Analysis of Mexican Manufacturing in the 1990s September 15, 2000 Document of the World Bank CURRENCY EQUIVALENTS Currency Unit: Mexican Peso (N$) N$1.0 - US$0.108 (September 1, 2000) WEIGHTS AND MEASURES Metric System FISCAL YEAR July 1 - June 30 ACRONYMS ALADI Asociacion Latinoamericana de Integracion ALTEX Empresas Altamente Exportadoras BANCOMEXT Banco Nacional de Comercio Exterior BBV Banco Bilbao Viscaya CIM Censo Industrial Maquiladora CIMO Programa Calidad Integral y Modernizacion CIPI Comision Intersectorial de Politica Industrial CONACYT Consejo Nacional de Ciencia y Tecnologia EIA Encuesta Industrial Anual EIM Encuesta Industrial Mensual ENESTYC Encuesta Nacional de Empleo, Salarios, Tecnologia, y Capacitacion GATT General Agreement on Tariffs and Trade GDP Gross Domestic Product INEGI Instituto Nacional de Estadistica, Geografia e Informatica NAFIN Nacional Financiera NAFTA North American Free Trade Agreement PEMEX Petroleos Mexicanos PITEX Programna de Importacion Temporal para Exportacion RFC Registro Federal de Causantes SECOFI Secretaria de Comercio y Fomento Industrial SME Small- and medium-scale enterprise TFP Total Factor Productivity IBRD Vice President David de Ferranti Chief Economist Guillermo Perry Country Director Olivier Lafourcade Sector Leader Fernando Montes-Negret Task Manager Kristin Hallberg This report was prepared by Kristin Hallberg, Hong Tan, and Leonid Koryukin in the Private Sector Development Department The authors would like to express their appreciation for the close collaboration and assistance offered by Enrique Ordaz, Abigail Duran, and Alejandro Cano (INEGI headquarters in Aguascalientes), Gerardo Durand (INEGI Mexico City), Jesus Cervantes (Banco de Mexico), and Sergio Fadl and Samuel Puchot (Bancomext). The report was prepared under the direction of Fernando Montes-Negret (FPSI Sector Leader for Mexico), Olivier Lafourcade (Country Director for Mexico), and Ira W. Lieberman (Senior Manager, PSD). Research and publication assistance were provided by Anna Lee Hewko and Caroline Bouffard- Spira. Table of Contents EXECUTIVE SUMMARY ...........................................................................................................................I 1. INTRODUCTION ..... 1 A. Overview ........ I B. Macroeconomic Context .............................................. 3 C. Trade Policy and Trade Agreements .............................................. 6 D. Trends in the External Sector .............................................. 9 E. Plan of the Report .............................................. 11 2. TRENDS IN EXPORT ORIENTATION ........................................................ 13 A. Databases .............................................. 13 B. Export Products, Origins, and Destinations ........................... ................... 15 C. Firm Concentration of Exports .............................................. 21 D. Characteristics of Exporters .............................................. 23 E. Export Transitions and Export Permanence ............................. ................. 25 F. Conclusions ............................................... 31 3. THE DECISION TO EXPORT ........................................................ 33 A. Case Studies .............................................. 33 B. Enterprise Surveys .............................................. 35 C. Regression Analysis ............................................... 39 D. Conclusions ............................................... 49 4. EXPORTING AND PERFORMANCE: CAUSE OR EFFECT? ..................................................... 51 A. EIA .............................................. 52 B. Maquiladoras .............................................. 62 C. Conclusions .............................................. 64 5. EXPORTING, SUPPLIER DEVELOPMENT, AND PRODUCTIVITY ......................................... 66 A. Overview .............................................. 66 B. Measuring TFP Growth ............................................... 66 C. Production Function and TFP Estimates .................... .......................... 67 D. Analyzing Firm-Level TFP Growth ............................................... 69 E. TFP Effects of Exporting and Subcontracting ........................................ ...... 72 6. CONCLUSIONS AND POLICY IMPLICATIONS ................................................... 78 BIBLIOGRAPHY ................................................. 82 ANNEX A: TRADE AND EXPORT ORIENTATION TABLES ................................................. 85 ANNEX B: DATABASES .99 List of Tables Page No Table 1.1: Contribution of Components of Aggregate Demand to Real GDP Growth ..............4 Table 1.2: Trade Liberalization 1985-88 .....................................................7 Table 1.3: Import Tariffs and Licensing Requirements, 1990-1997 .........................................7 Table 1.4: Composition of Exports .....................................................11 Table 2.1: Value of Exports and Number of Plants in the EIA and CIM ................................ 14 Table 2.2: Exports of Metal Products and Machinery (EIA) ................................................... 16 Table 2.3: Regional Distribution of Exports (EIA) .................................................... 18 Table 2.4: Regional Distribution of Exporters (EIA) .................................................... 18 Table 2.5: Mexico Export Participation in U.S. Imports .................................................... 20 Table 2.6: Export Market Diversification .................................................... 21 Table 2.7: Export Performance .................................................... 26 Table 2.8: Incumbent and New Exporters in 1995 .................................................... 27 Table 2.9: Decomposition of Export Growth (EIA) .............................. 30 Table 2.10: Decomposition of Export Growth Excluding Automobile Industry (3841) ........... 30 Table 3.1: Constraints to Increased Exports .................................,.,.,,.,., . 37 Table 3.2: Sources of Financing for Exporters ................................ 38 Table 3.3: Export Experience Variables ................................ 40 Table 3.4: Independent Variables Used in Regressions ................................ 40 Table 3.5: Probit Models of Export Status ................................ 43 Table 3.6: Predicted Probability of Exporting by Size ................................ 44 Table 3.7: Probit Models of Entry Decision ................................ 47 Table 3.8: Tobit Models of Export Volume ................................ 49 Table 5.1: Production Function Estimates ................................ 68 Table 5.2: TFP Levels and TFP Growth Rates ................................ 69 Table 5.3: Assistance Received by Suppliers from Buyers ................................ 72 Table 5.4: Determinants of Firm-Level TFP - Random Effects Model .................................. 74 Table 5.5: Determinants of Firm-Level TFP - Linked EIA-ENESTYC Sample .................... 77 Table A. 1: Balance of Payments - Current Account ................................................ 86 Table A.2: Balance of Payments - Capital Account ................................................ 87 Table A.3: Import Tariffs and Licensing Requirements ................................................ 88 Table A.4: Exports by Sector of Origin ................................................ 89 Table A.5: Mexico's Major Trading Partners ................................................ 90 Table A.6: Main Products Traded by Mexico ................................................ 91 Table A.7: Composition of Exports ................................................ 92 Table A.8. 1: Firm Concentration of Exports .93 Table A.8.2: Firm Concentration of Exports .94 Table A.9: Finn Concentration of Non-Maquila .95 Table A.10: Export Status by Size of Plant .................................... 96 Table A. 1 1: Export Status by Industry .96 Table A.12: Export Status by Ownership Category .................................... 97 Table A. 13: Entrants and Exits by Size .97 Table A.14: Decomposition of Export Growth .97 Table A.15: Export Transitions .98 Table A.16: Export Status of Plants in EIA ................................... 98 Table B. 1: Number of Plants in EIA by Industry ................................... 99 Table B.2: Results of Plant Re-Classification ................................... 100 Table B.3: CIM Incumbents, Entrants, and Exits ................................... 101 Table B.4: Number of Plants in CIM by Industry ................................... 102 Table B.5: Maquiladora Plants by Region ................................... 103 Table B.6: Number of Plants in ENESTYC 1995, by Industry ................................... 104 List of Figures Figure 1.1: Real Exchange Rate Index ......................................4 Figure 1.2: GDP, Consumption, Capital Formation, and Exports ...................................... 5 Figure 1.3: Trends in GDP Growth ......................................6 Figure 1.4: Total Exports ..................................... 10 Figure 2.1: Composition of Exports (EIA) ..................................... 16 Figure 2.2: Mexican Participation in U.S. Imports ..................................... 20 Figure 2.4: Lorenz Curve for Non-Maquiladora Exporters, 1995 ..................................... 21 Figure 2.5: Firm Concentration of Exports ..................................... 22 Figure 2.6: Export Value by Size (EIA) ..................................... 22 Figure 2.7: Export Status of Plants (EIA) ........................ 23 Figure 2.8: Export Status by Size (EIA) ........................ 23 Figure 2.9: Export Status by Ownership (EIA) ........................ 24 Figure 2.10: Export Value by Ownership (EIA) .24 Figure 2.11: Export Orientation by Size (EIA) .25 Figure 2.13: Entry and Continuation of Exporters .26 Figure 2.14: Export Transition (EIA) .28 Figure 2.15: Export Entrants and Exits by Size (EIA) .29 Figure 2.16: Export Entrants and Exits by Ownership (EIA) .29 Figure 2.17: Decomposition of Export Growth (EIA) .30 Figure 2.18: Decomposition of Export Growth Excluding Automobile Industry .31 Figure 4.1: Total Factor Productivity of Entrants ......................... 54 Figure 4.2: Labor Productivity of Entrants ......................... 55 Figure 4.3: Total Factor Productivity of Quitters ......................... 55 Figure 4.4: Labor Productivity of Quitters ......................... 56 Figure 4.5: Profitability of Entrants ......................... 57 Figure 4.6: Profitability of Quitters ......................... 57 Figure 4.7: Inventory Turnover of Entrants ......................... 58 Figure 4.8: Inventory Turnover of Quitters ......................... 58 Figure 4.9: Imported Material Inputs of Entrants ......................... 59 Figure 4.10: Imported Assets of Entrants .60 Figure 4.11: Imported Material Inputs of Quitters .60 Figure 4.12: Imported Assets of Quitters .61 Figure 4.13: New Investment of Entrants .61 Figure 4.14: New Investment of Quitters .62 Figure 4.15: Total Factor Productivity of Maquiladoras .63 Figure 4.16: Other Performance Variables for Maquiladoras .64 Figure 5.4: Trends in Exporting and Subcontracting (1993-1997) .71 List of Boxes: Box 2.1: The Automobile Industry in Mexico .17 Box 3.1: Madermont S.A. de C.V .33 Box 3.2: Interquimia S.A. de C.V .34 Box 3.3: Forja Collection Mexico S.A ................... 34 i Executive Summary More than ten years ago, Mexico began a process of structural reform that fundamentally changed the economic environment facing productive enterprises. The centerpiece of structural reforms was trade liberalization, beginning with Mexico's entry into GATT in 1986 and continuing with the replacement of quantitative restrictions with tariffs, a reduction in the levels and dispersion of tariffs, the elimination of official prices, and the simplification of regulations governing international trade. Trade policy reform was accompanied by the easing of restrictions on foreign investment, improvements in the regulatory framework, privatization of public enterprises, and financial sector reform. In early 1994, implementation of the North American Free Trade Agreement (NAFTA) was initiated, marking a major opening of trade and investment flows among the three countries. Mexico's export-led growth strategy attempts to achieve higher and more sustained rates of economic growth by expanding the country's participation in dynamic external markets and encouraging higher rates of productivity growth. It was hoped that greater outward orientation and its associated efficiency gains would extend to a large share of productive sector firms -- small as well as large, traditional as well as modern, and those located in less-developed regions of the country -- thus benefiting a broad segment of the population. If the productivity gains are broadly distributed and linkages between the export sector and the domestic economy are strengthened, trade liberalization can lead not only to higher growth but also more equitable growth. - This report examines the export performance of Mexican firms in the 1 990s to determine whether the intended effects of trade liberalization have, in fact, occurred. It investigates: * the depth and extent of changes in outward orientation of firms and the characteristics of firms that became more export-oriented; * the degree of linkage between the export sector and domestic economy, or conversely, the extent to which the export sector is an "enclave" whose behavior is independent from the rest of the economy; * the factors that determine export behavior, and the extent to which exporting is ''permanent" versus "transitory"; and * the extent to which productivity gains from exporting have been realized, whether through learning by direct exporters or through spillovers to indirect exporters and others. ii The report relies on statistical analysis of cross-section/time-series (panel) databases with observations at the level of the individual plant or firm. Three such panel databases were constructed: one from the annual industrial survey (Encuesta Industrial Anual, EIA) conducted by the Mexican statistical agency (Instituto Nacional de Estadistica, Geografia e Informatica, INEGI), covering 1993-98; another from INEGI's monthly census of plants operating under the maquila regime, covering 1990-98; and the central bank (Banco de Mexico) customs database of exports and imports at the level of the firm, covering 1991-98. The report's analysis of productivity growth using micro data is the first the Bank has done since a 1994 report using a smaller 1984-90 database, and it complements the Bank's recent Country Economic Memorandum which analyzes productivity growth using macroeconomic data. Growth in Manufactured Exports The first chapter provides background on the macroeconomic context and aggregate trade flows. Since the opening of the economy began in the mid- 1980s, the external sector has grown rapidly relative to the rest of the economy. The openness of the economy, measured as the sum of exports and imports of goods and services relative to real GDP, grew from 23 percent in 1985 to 42 percent in 1995. Manufactured exports in particular have increased dramatically, achieving growth rates similar to those of Hong Kong and Singapore, the two Asian economic tigers with the greatest export dynamism. Mexico has moved from being primarily dependent upon oil for its export revenues, to being primarily an exporter of manufactured products. The economic crisis of 1995-96 accelerated these trends, as the devaluation of the peso and the sudden drop in domestic demand increased the incentive to export. Export Concentration The second chapter uses the panel databases to show trends in the composition of exports, the characteristics of exporters, and transitions in and out of exporting. Exports of manufactures grew rapidly from N$49 billion in 1993 to N$338 billion in 1998 (using EIA data). Metal products and machinery (the two-digit industry that includes automobiles and autoparts) is clearly dominant, accounting for 70 percent of manufactured exports in 1998. Exports of automobiles nearly equaled the value of oil exports in 1997, in part because of low oil prices, but also because of growing importance of large maquiladora and non-maquiladora automobile assemblers and parts manufacturers. The most important destination for Mexican manufactured exports continues to be the United States, and this tendency was accentuated wvith the implementation of NAFTA in 1994. Mexico has the highest concentration of exports in a single market destination (85 percent to the U.S. in 1997) than any other country in the world. With NAFTA, Mexico's participation in U.S. imports increased significantly. Exports in Mexico are highly concentrated across firms: a few large exporters account for a large share of export value. The degree of concentration is high even when maquiladoras, automobile assemblers, and PEMEX are excluded. There seems to have been an increase in concentration in the early 1990s: the share of the largest 10 exporters iii rose from 16.5 percent in 1991 to 19.2 percent in 1994, and the share of the largest 280 exporters rose from 59.2 percent to 64.4 percent during the same period. Nevertheless, Mexico's export concentration may not be very different from that of other countries. Some recent studies of firm-level export dynamics show that most exporters export only a small share of their production. Export Response During the Crisis In December 1994, a sharp devaluation of the Mexican peso was followed by one of the deepest economic recessions in decades. In 1995, real GDP fell by 6.2 percent, demand by 14.4 percent, and investment by 43 percent. The devaluation and contraction of the domestic market led to a considerable shift in the trade balance: exports increased by almost 31 percent and imports fell by 9 percent in 1995. The strong performance of the export sector moderated the fall in demand in 1995, and acted as a "safety valve" for many firms whose internal markets had all but disappeared. Domestic demand recovered quickly, however, and by the end of 1996 real GDP had regained its pre-crisis level. The EIA panel database reveals a significant increase in the export orientation of the manufacturing sector during the 1993-98 period: the share of plants that export rose from 26 percent in 1993 to 42 percent in 1997. Although export entry cut across all plant size categories, the increase in the number of exporters and in the share of output exported was most pronounced for large-scale plants. Both Mexican-owned and foreign- owned plants became new exporters, but foreign-owned plants continue to account for a high proportion of export volume. As the domestic economy recovered from the crisis in 1997-98 and the peso gained against the dollar, some firms shifted back to their internal market. A large number of these "exits" from exporting after the crisis were SMEs and Mexican-owned firms; larger and foreign-owned plants seem to have a more permanent presence in external markets. Qualitative survey evidence suggests that SMEs have a shorter-term exporting "culture", preferring to return to their home market when demand conditions improve. This conclusion is supported by looking at the average number of years of exporting by firm size: during 1993-98, large-scale plants exported an average of 3.2 years, medium-scale plants 2.3 years, and micro/small-scale plants 1.4 years. This may suggest that the export orientation of smaller and Mexican-owned firms is influenced more by domestic demand conditions, whereas larger and foreign-owned plants are reacting more to trade opportunities opened by NAFTA. Determinants of Export Status and Export Entry In Chapter 3, we use regression analysis to investigate the determinants of export status (exporter versus non-exporter) export entry (moving from non-exporter to exporter status), and export volume. Across several variations of the model, the influence of current macroeconomic conditions on export status is usually as expected: the probability of exporting rises with a decline in domestic demand, a depreciation of the peso, and a decrease in the level and variability of domestic interest rates. However, the iv macroeconomic results are not robust, suggesting that other, perhaps longer-term, factors influence the export decision. Current export status is clearly influenced by previous export experience, perhaps because of the sunk costs associated with starting to export (acquiring market information, establishing distribution channels, etc.) There is a clear association between plant-level efficiency (measured by total factor productivity (TFP) growth) and exporting, even controlling for plant size, ownership, and other plant characteristics. The regression results suggest that high- performance workplace practices and labor skills are associated with export status and export entry, and anecdotal evidence suggests that investments in quality and modernization are made in anticipation of stiffer competition in external markets. The association between exporting and IS09000 and other quality management systems is less apparent, though the regression results may be influenced by collinearity among independent variables. New investment, in particular purchases of imported assets, are associated with subsequent entry into export markets. Qualitative enterprise surveys suggest that financing plays a key role, both in export entry and export growth. Particularly during the crisis, there was a clear "duality" in access to credit: larger, foreign-owned, and export-oriented firms were able to borrow in international financial markets, whereas smaller, Mexican-owned, and domestic- market oriented firms were constrained by the dramatic decline in bank credit to the private sector. Interviews with SME exporters suggested that many chose to limit their growth to that which could be financed with internal resources, aware as they were of the potential for macroeconomic shocks. For larger firms, the cost of credit is seen as a greater constraint than the availability of credit. Exporting and Efficiency: Cause and Effect Many economists and policymakers believe that exporting leads to improvements in technical efficiency: "learning-by-exporting" is said to occur as exporters benefit from the technical expertise of their buyers, strive to meet quality standards in external markets, and discover new ways to reduce costs. This hypothesis seems to be supported by the efficiency gap between exporters and non-exporters that is observed in many countries. However, the association between exporting and efficiency may be due to causality in the other direction, as more efficient firms tend to be the ones that successfully break into export markets. Chapter 4 investigates graphically the dynamic relationship between exporting and enterprise performance, using the EIA panel database as well as the maquiladora database to track individual plants over time. We use a broad range of performance indicators -- total factor productivity, labor productivity, inventory turnover, new investment, and the import share of new assets and material inputs -- to investigate what drives changes in the export status of plants and what happens to them after they begin to export. v The analysis shows that the productivity of non-maquiladora (EIA) plants that start exporting grows substantially in the two to three years prior to entering the export market. Their productivity continues to grow relative to that of permanent exporters and non-exporters for at least two to three years after entry, suggesting that learning-by- exporting effects do exist. This finding contrasts with earlier work on the export- productivity link in Mexico and other countries that fails to find learning-by-exporting effects. Newly entering maquiladoras also seem to improve their performance in the years after entry. This could reflect learning on the part of new maquiladoras, or a process of selection as inefficient maquiladoras exit. One or two years prior to entering the export market, EIA plants seem to experience a downturn in their domestic operations and a modernization or restructuring in anticipation of entering export markets. Their profitability falls, new investment increases, and imports of machinery and equipment intensify. The pre-export downturn in domestic operations is also seen in lower inventory turnover that improves when the plants start exporting. As is the case with entrants, plants that cease to export experience a downturn in their operations for several years before they quit. This is revealed mainly in the rapid inventory accumulation prior to quitting, rather than in a decline in profitability. Re-orientation to the domestic market is associated with two opposite trends: substantial, but short-term, gains in profitability (perhaps due to the resolution of the inventory problem), and a gradual, long-term decline in productivity. Exporting, Supplier Development, and Productivity The report also uses a production function framework to measure the impact of exporting as well as other factors including supplier linkages, spillovers, training, foreign direct investment, and high-performance workplace practices on plant-level TFP growth. The regression results in Chapter 5 confirm that exporters experienced more rapid TFP growth than non-exporters over 1993-98. Export experience is a significant determninant of productivity growth: each year of exporting experience is associated with a 4.5 percent rise in TFP. This again suggests that learning through exporting is taking place: while the immediate productivity gains from exporting are modest, sustained productivity gains accrue as experience accumulates. We use the regression model to test whether there are learning effects (productivity increases) associated with becoming a supplier to other firms, as suppliers may receive technical assistance or technology transfer from contracting firms. To better isolate the various factors that lead to productivity improvements, the plants in the EIA panel were linked to those covered in the 1995 National Survey of Employment, Wages, Technology, and Training (ENESTYC). When variables measuring high-performance workplace practices are included in the regressions, there is evidence of learning from supplier experience. There is an initial dip in TFP when firms first become suppliers, but by the sixth year their TFP levels are about 9 percent higher than that of non-supplier firms. Concern for product and process quality, as manifested in the use of quality-circle practices, is associated with a 14 percent increase in TFP; worker training (in-house formal training in particular) is also an important determinant. vi Information in the ENESTYC survey suggests that outsourcing is associated with transfers of assistance from buyers to suppliers. Over half of suppliers receive some form of assistance, the most common being supplies of raw materials, followed by technical assistance. Less common, but very important in light of the credit constraints faced by small Mexican-owned firms in the mid-1990s, is help with financing. While suppliers that have some foreign capital are more likely to receive assistance from buyers -- many of whom may be foreign-owned firms or maquiladoras -- it is the export-oriented suppliers that tend to receive the most. Policy Implications The objective of this report was to provide a retrospective view of the performance of the export sector in the wake of structural reforms, and thus did not evaluate policy options in depth. Nevertheless, based on our analysis of performance to date, our understanding of the Mexican industrial sector, and lessons from other countries, we can suggest areas that seem to be important, directions for future policy reform, and topics that need further investigation: * The pre-entry phase of preparing to compete in export markets is important. Building labor skills, investing in high-quality equipment and materials, reorganizing the production line, and establishing quality management systems raise productivity and give firms the "credentials" to participate in foreign markets. The Government should focus on extending the coverage and quality of its support for enterprise training, technology diffusion, and information. e Buyer-supplier relationships are an important channel for extending the productivity benefits of exporting to a wider group of Mexican firms. The Government has long recognized the need to strengthen the linkages between the export sector and the domestic economy, and has a number of supplier development programs to encourage large firms to purchase from or subcontract to smaller Mexican firms. In general the impact of these programs has been limited. They should be re-evaluated in light of international experience, and more effective ways of facilitating private-to-private transactions should be designed. * While "transitory" exporters may be able to move in and out of foreign markets as a reaction to changes in domestic demand, it appears that the productivity benefits of exporting increase as export experience accumulates. We need to better understand what encourages or discourages "permanent" exporting behavior, particularly among SMEs and Mexican-owned firms. * Lack of access to credit, which particularly affects SMEs and Mexican-owned firms, is a constraint to modernization and entry as well as to export permanence and growth. Although it is beyond the scope of this report to propose specific measures, improving access to finance seems to be key to increasing export orientation across a broad spectrum of firms. 1 1. INTRODUCTION A. Overview 1.1 More than ten years ago, Mexico began a process of structural reform that fundamentally changed the economic environment facing productive enterprises. The centerpiece of structural reforms was trade liberalization, beginning with Mexico's entry into GATT in 1986 and continuing with the replacement of quantitative restrictions by tariffs, a reduction in the levels and dispersion of tariffs, the elimination of official prices, and the simplification of regulations governing international trade. Trade policy reform was accompanied by the easing of restrictions on foreign investment, improvements in the regulatorv framework, privatization of public enterprises, and financial sector reform. 1.2 Negotiations leading to the North American Free Trade Agreement (NAFTA) began in 1991 and the agreement became effective in January 1994, marking a major liberalization of trade and investment flows among the three countries. The agreement specified a gradual elimination of tariffs and non-tariff barriers, common norms and standards, customs improvements, and mechanisms for the resolution of disputes. The Mexican authorities hoped that NAFTA would open the U.S. and Canadian markets to Mexican exports, lower the cost of capital goods and materials imported into Mexico, and increase foreign direct investment in Mexico from NAFTA partners as well as third countries. 1.3 Mexico's export-led growth strategy attempts to achieve higher and more sustained rates of economic growth by expanding the country's participation in dynamic external markets and encouraging higher rates of productivity growth. The latter is expected to result both from gains in static efficiency (more efficient resource allocation across sectors and more efficient use of resources within sectors) and dynamic efficiency (continuing reductions in costs and improvements in product quality). It was hoped that the greater outward orientation and its associated efficiency gains would extend to a large share of productive sector firms -- small as well as large, traditional as well as modern, and those located in less-developed regions of the country -- thus benefiting a broad segment of the population. 1.4 This report assesses the export performance of Mexican firms in the 1990s to investigate whether the intended effects of trade liberalization have, in fact, occurred. It investigates: * the depth and extent of the changes in outward orientation of firms and the characteristics of firns that became more export-oriented; * the degree of linkage between the export sector and the rest of the economy, or conversely, the extent to which the export sector is an "enclave" whose behavior is independent from the domestic economy; 2 * the barriers and opportunities faced by firms that attempt to increase their exports or break into exporting, and the extent to which exporting is "permanent" versus "transitory"; * the extent to which productivity gains from exporting have been realized, whether through learning by direct exporters or spillovers to indirect exporters and others. 1.5 Investigating these questions required us to trace the performance of individual enterprises over time, linking their actions and performance with their individual characteristics as well as with their external environment. Thus the report relies on careful statistical analysis of cross-section/time-series (panel) databases with observations at the level of the individual plant or firm. Three such panel databases were constructed for this report: one from the annual industrial survey conducted by INEGI, covering 1993-98; another from INEGI's monthly census of plants operating under the maquila regime, covering 1990-98; and Banco de Mexico's customs database of exports and imports at the level of the firm, covering 1991-98.1 Using these three databases gives us a fairly complete picture of the performance of Mexican enterprises over time. The report's analysis of productivity growth using micro data is the first the Bank has done since the 1994 report based on a smaller 1984-90 panel database2, and it complements the Bank's recent Country Economic Memorandum which analyzes productivity growth using macroeconomic data.3 1.6 By law, the information on individual firms that INEGI gathers through its questionnaires (which firms are required by law to answer) is confidential, and INEGI is unable to give the raw data to outside agencies. We discussed with INEGI various ways of using the survey data without violating confidentiality restrictions, and established a procedure in which we did the data analysis in INEGI's Aguascalientes headquarters (for the industrial survey) and the Mexico City office (for the maquiladora survey) in close collaboration with INEGI staff and management. Our results -- tabulations of aggregated data and regression results -- were checked by INEGI staff to ensure that confidential data were not being taken out.4 Not only did our practice of working inside INEGI give us access to the data, working closely with them helped us to refine our questions and interpret the results. Nevertheless, when reading the report, the reader should bear in mind the limitations imposed by this institutional arrangement: since the data analysis was done only during our infrequent trips to Mexico, there remain unanswered questions and possible variations on the models that would be interesting to explore in a next phase of research in INEGI. 1.7 We received valuable support from other government agencies in addition to INEGI. Bancomext helped us refine the issues to be investigated, contributed the results ' A detailed description of these databases and how they were cleaned is contained in Annex B. 2 "Mexico: Reform and Productivity Growth", Report No. 12605-ME, 1994. 3 "Mexico: Enhancing Factor Productivity Growth" (Country Economic Memorandum) Report No. 17392- ME, 1998. 4 The procedures are similar to those used by the U.S. Bureau of the Census to allow researchers to access the Longitudinal Research Database. 3 of their surveys of exporters, and organized interviews of exporters by Bank staff. Banco de Mexico also contributed their ideas for the issues to be investigated and then provided evidence using their own panel database. The report would not have been possible without support and collaboration of all three institutions. 1.8 We anticipate that this report will be the first of a number of pieces on industrial policies and performance in Mexico that take advantage of the wealth of existing microeconomic data. Mexico's newly-created Industrial Policy Commission (CIPI) is planning to establish a Research and Planning Unit to evaluate industrial sector performance and constraints, in order to guide the design of policies and programs supporting industrial sector development. We hope that the report makes an initial contribution to this important effort. 1.9 The rest of this chapter provides background for the analysis contained in subsequent chapters: the macroeconomic context in the 1990s; trade reform and trade agreements undertaken since Mexico's structural reform program began in the mid- 1980s; and trends in aggregate trade flows during this period. It concludes with the plan of the rest of the report. B. Macroeconomic Context 1.10 In December 1994, a sharp devaluation of the peso (Figure 1.1) was followed by one of the deepest economic recessions in decades. In 1995, real GDP fell by 6.2 percent, demand by 14.4 percent, and investment by 43 percent (Figure 1.2). The decline of the industrial sector was even more pronounced than for the economy as a whole, in part because the construction industry was the hardest hit by the dramatic reduction in public and private investment in 1995. The devaluation and the contraction of the internal market led to a considerable shift in the trade balance: exports increased by almost 31 percent and imports fell by 9 percent in 1995. 1.11 The strong performance of the export sector moderated the fall in demand in 1995, and acted as a "safety valve" for many firms for whom the domestic market had all but disappeared. It was also an important force behind the subsequent recovery, though domestic demand began to recover in early 1996 and contributed to the recovery in GDP growth (Table 1.1). By the end of 1996, real GDP had regained its pre-crisis level, as did private investment and consumption in 1997. The three aggregates, especially private investment, continued their fast expansion during 1998. 4 Figure 1.1: Real Exchange Rate Index | Based On Unit Labor Costs in the Manufacturing Sector I 990=1 00 and period averages 8120 - 1 i.o100- t 60 - _ _ - 4O 197577 79 81 83 85 87 89 91 93 II IV IIV 11 IV 11 IV 1994 1995 1996 1997 Table 1.1: Contribution of Components of Aggregate Demand to Real GDP Growth (Annual percentage change) Growth originated in: Period Real GDP Growth Domestic Demand* Foreign Demand (1)=(2)+(3) Net Exports ** 1995 -6.2 -9.2 3.0 1996 5.2 3.6 1.6 1997 7.0 5.8 1.2 *Domestic demand net of imports. **Exports net of their imported content. Source: Banco de Mexico and INEGI. 1.12 Even as domestic demand recovered in 1996-98, domestic bank lending to the private sector continued to decrease. Mexican banks were said to be lending to their best clients, and an atmosphere of extreme caution still pervades the banking sector. Foreign- owned firms and the larger and more export-oriented firms obtained their financing abroad, and foreign capital flows increased significantly. Many of the smaller and domestic market-oriented firms had difficulty obtaining new domestic credit; for them, the 1996-98 period seems to have been a "credit-less recovery".5 5 "Mexico: Strengthening Enterprise Finance", Report No. 17733-ME, September 25, 1998. 5 Figure 1.2: GDP, Consumption, Capital Formation, and Exports Annual percentage change 1 2 ---------- 661 -6-715 -67 7 5 6 4 6 22 15714 . 2 3 9E , I1it -3044 4 |GROSS DOMESTIC -8 .9 1 7 04 PRODUCT -13 11 III lV I II II I I 11 1 III I 11 1 lV I III 1 lV 1994 . 1995 1996 1997 0 '- ---6 -- 7:29 587 6-79 -52- 8 4.26 4.38 2.1 3.65 3.63 2,74 -5 -2.19 -4.29 I I T -10O - 11 101 -8.0317 I 11 II IV I 11 III IV I 11 III IV I 11 III IV 1994 1995 1996 1997 45 30 26.3 27.35 24.4422 77 17.41 , u 8.51 18.31 105 -3 10.56 9.02 10.43 m .l -15 4 ROSS FIXED CAPITAL -30 I -8 FORMATION 45- -- - 34.01-3368 I II III IV I II 111 IV I 11 111 IV I 11 III IV 1994 1995 1996 1997 40 35 28 3 5 3 30. 86 EP SO GOD I 3 0 28 9 I ADSRVICES I 30~~~~~~~~~ 25.58 9 25 2 2097 9 4 20 15 24 15.81 9 173116 8 195 1 5 10.32 11-~~~~~~~~~~~~~~~~~~~~~~~I 09 10 1 11 III IV I II 1II IV I 1I III IV I 11 III IV 1984 1995 1996 1997 6 1.13 Two other external shocks hit the Mexican economy after the 1995 economic crisis: the decline in international oil prices in 1997 and early 1998, and the Asian financial crisis that began in July 1997. The decline in oil prices had a significant effect on fiscal revenues -- public sector oil revenues account for more than a third of public sector revenues - and the Government was forced to respond with major cuts in fiscal expenditure as well as more restrictive monetary policy. The authorities feared that the Asian financial crisis could affect the Mexican economy through a number of channels: an appreciation of the peso vis-a-vis the currencies of the Asian countries with which Mexico competes in the U.S. market; lower growth in industrialized countries, decreasing demand for Mexican products; greater penetration of Asian imports into Mexico; and a discouraging effect on capital inflows into emerging economies. According to a Banco de Mexico/SECOFI survey of almost 700 exporters in August- September 1999, 44 percent of maquiladoras and 42 percent of non-maquiladoras reported that their exports suffered from greater competition from Asian exports, but more maquiladoras than non-maquiladoras (57 percent versus 25 percent) reported that the Asian crisis significantly affected their exports.6 1.14 Looking over the past twenty years, it appears that the comprehensive economic liberalization program that Mexico launched in the late 1980s and early 1990s created a shift in trend of economic growth. Even factoring in the 1995 economic crisis, real GDP growth has averaged 2.5 percent per year since 1990 compared to an average of 0.3 percent per year between 1980 and 1989 (Figure 1.3).7 Figure 1.3: Trends in GDP Growth 1,450 Billion 1,350 _ -- 1993 pesos 120- 1,150 1,050 950 1980 1982 1984 1986 1988 1990 1992 1994 1996 19987 - _ Real GDP GDP trend C. Trade Policy and Trade Agreements 1.15 The first important trade policy reforms in the 1980s were initiated in early 1983 with the implementation of the temporary import regime for exporters. The Governrment then began to reduce the coverage of quantitative restrictions on imports, from 100 6Banco de Mexico/SECOFI (I 999b), p. 6. 7 Mexico Country Assistance Strategy, May 1999. 7 percent coverage of domestic production in 1983 to about 92 percent in 1985. Nevertheless, in early 1985 the economy still remained relatively closed. In July 1985 the Government began to make serious cuts in the coverage of QRs, tariff rates, and official reference prices, and adopted an essentially floating exchange rate. The depth of the reforms are clear from Table 1.2: the coverage of QRs fell to 40 percent at the end of 1986 and 25 percent at the end of 1987; tariff rates were cut in half; and official reference prices were eliminated by 1988. 1.16 Tariff rates have continued to decline in the 1990s. About half of imported products now carry tariffs below 10 percent, and 95 percent are below 20 percent (Table 1.3). Non-tariff restrictions affected about 8 percent of tariff categories in 1997, down from 13 percent in 1990. Table 1.2: Trade Liberalization 1985-88 1985 1986 1987 1988 June Dec. June Dec. June Dec. March Import licensing 92.2 47.1 46.9 39.8 35.8 25.4 25.4 Reference prices 18.7 25.4 19.6 18.7 13.4 0.6 0.0 Tariffs 23.5 28.5 24.0 24.5 22.7 11.8 11.8 1/ Percent coverage of 1986 production. 2/ Average tariffs weighted by 1986 production; excludes 5% surcharge. Source: World Bank (1988). Table 1.3: Import Tariffs and Licensing Requirements, 1990-1997 (% of total tariff items) 1990 1994 1997 Tariff rates 0% 2.5 9.6 14.4 1-10% 49.7 42.5 38.3 11-20% 47.7 47.1 42.2 >20% -- 0.7 5.0 Prohibited 0.1 0.1 0.2 Licensing Requirements Unrestricted 86.4 89.4 91.8 Restricted 13.6 10.6 8.2 Source: Banco de Mexico (1998), from SECOFI. 8 1.17 In addition to the general decline in tariff rates, NAFTA specifies a staged reduction in tariffs on imports from the U.S. and Canada, starting with the tariffs in force in 1991 (a maximum of 20 percent and an average of 13.2 percent). Since there was some attempt to reduce output tariffs faster than input tariffs, effective rates of protection sometimes declined faster than nominal rates. Final elimination of tariffs is to be achieved in zero to fifteen years, depending on the product. As of January 1998, about 60 percent of tariffs had been eliminated. By 2003, most manufacturing products will have zero tariffs; those remaining (mainly agricultural products such as corn) will experience staged reductions in tariffs ending in 2008. 1.18 A number of industries in the Mexican productive sector are expected to gain from NAFTA, including fruits and vegetables, chemicals, textiles (especially clothing and synthetic fibers), leather, glass, steel, copper and lead, electronics, appliances, and automobiles. Those expected to lose competitiveness vis-a-vis the U.S. include medical equipment, sophisticated machinery, and capital goods. In the long run, the authorities expect to see increased trade in "niche" markets, rather than an expansion in exports across broad industry groups.8 1.19 The implementation of NAFTA has implications for other trade policies. The temporary import regime (PITEX) was eliminated for exports to the U.S., which is now governed by NAFTA; PITEX is retained for exports to third countries. The maquila regime is in the process of being eliminated. The share of a maquiladora's production that may be sold in the Mexican market has been increased to 75 percent, and will be raised to 100 percent by 2001. (In practice, most maquiladoras export all of their production.) Essentially, after 2001 the maquila and PITEX regimes will be merged. 1.20 During the past five years, in addition to NAFTA, Mexico entered into commercial agreements with other countries in the Western Hemisphere: * Mexico-Chile (1992): Trade liberalization agreement in the ALADI tradition, covering rules of origin and market access. The initial agreement covered all products except petroleum and some agricultural products, and in 1999 was extended to include services and international property rights along NAFTA lines. Implementation of the agreement began with the elimination of non-tariff barriers (except those covered under ALADI). An initial 10 percent maximum tariff was established, to be reduced by 25 percent each year. In 1996, tariffs had been eliminated for 97 percent of tariff categories; the rest were to be eliminated by January 1998. Tariffs on automobiles were eliminated in January 1996. * Mexico-Costa Rica (1995): Liberalization of trade and investment beginning with the elimination of tariffs on 70 percent of non-agricultural exports from Mexico in January 1995 to 90 percent in January 1999 and the remainder by January 2004. According to Bank staff interviews with SECOFI authorities in 1998. 9 * Mexico-Colombia-Venezuela (1995): The G-3 Agreement gradually eliminates tariffs on products traded among the three countries, with the exception of sensitive sectors. It also covers illegal trade practices and resolution of disputes. * Mexico-Bolivia (1995): This agreement began a rapid process of liberalization of goods traded between the two countries. When it entered in force in January 1995, tariffs were eliminated on 97 percent of Mexican exports to Bolivia and 99 percent of Bolivian exports to Mexico. * Mexico-Nicaragua (1998): The agreement provides for reciprocal reduction in tariffs and non-tariff barriers, to be completed in 2005. * Mexico-Israel (2000): Reductions in tariffs on agricultural and industrial products, and virtual elimination of non-tariff barriers. Israeli tariffs on industrial products, currently averaging 7.6%, are to be eliminated by 2003 with some exceptions (which are either to be eliminated immediately, or postponed to 2005). * Mexico-European Union (2000): Negotiations began in 1995, culminating in the implementation of a free trade agreement in July 2000. The agreement calls for a gradual, reciprocal reduction in tariffs and the elimination of quantitative restrictions. Recognizing the asymmetries between Mexico and the EU, the timetable for phased reductions extends to 2003 for the EU and 2007 for Mexico. Tariff reductions for some agricultural and agro-industrial imports extend up to ten years, and there are some products on the "wait list". * Others: Negotiations with El Salvador, Guatemala, and Honduras concluded in mid- 2000 for implementation of a free trade agreement starting January 1, 2001. Negotiation processes leading to free trade agreements have been established with Ecuador and Peru, and with Panama and Trinidad and Tobago; they are being developed with Brazil. D. Trends in the External Sector 1.21 Since the opening of the economy in the mid-1980s, the external sector has grown significantly relative to the rest of the economy. The openness of the economy, measured as the sum of exports and imports of goods and services relative to real GDP, grew from 22.8 percent in 1985 to 31.8 percent in 1990, 38.6 percent in 1994, and 41.8 percent in 1995. Exports of goods and services as a proportion of real GDP grew from 15.3 percent in 1985 to 26.9 percent in 1995. Private investment in the export sector also increased faster than investment in the rest of the economy. The public sector's share of international trade transactions fell from 44 percent in 1985 to 13 percent in 1995. 9 J. Cervantes, "Cambio estructural en el sector externo de la economia mexicana." Comercio Exterior (March 1996, p. 180), and Banco de Mexico, Informe Anual 1996. 10 1.22 The volume of exports has grown rapidly, from US$26.8 billion in 1985 to US$110.4 billion in 1997. Since 1987, the annual growth in exports in U.S. dollar terms has averaged 16 percent (Figure 1.4). The rate of export expansion that began in 1985 was impressive even by East Asian standards. From 1985 to 1995, the growth of non-oil manufacturing in Mexico was similar to that of Hong Kong and Singapore, the two "tigers" with the greatest export dynamism. ' Figure 1.4: Total Exports 120 80* 60 51.9 609 35240.7 42.7 462 40 26.5 27.6 30.7 3. 21.8 20 0 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 40 - 306 30 26.6 2266 2 S0 *112 14.6 158 17.3 14 158 152 10 - - - 419 8.2 _12- -3 185 -10 9. l-20 -Annd Apaoerdag1e dw -30 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1.23 The composition of exports has shifted from a high dependence on oil to primarily manufactured products. In 1985, oil accounted for more than 55 percent of total exports; in 1995, the share had fallen to less than 11 percent, and further to 7.3 percent in 1999 (Table 1.4). Some of the decline was due to lower oil prices, but even holding oil prices at 1985 levels, the share of oil in total exports in 1995 would have been 16 percent. Manufactures increased from 38 percent of total exports in 1985 to over 89 percent in 1999. Both maquiladoras and non-maquiladoras participated: their shares of manufactured exports were roughly equal throughout the period. '1 ibid., p. 179. 11 1.24 Manufactured exports contain a wide range of products, from high-technology products (computers and computer components, high-precision instruments, optical goods) to intermediate-technology items (chemical products, pharmaceuticals, metal alloys, industrial molds) and traditional products (food, textiles, wood products, etc.). There has been some increase in the importance of high-technology exports, primarily by non-maquiladora firms; maquiladoras tend use simpler technologies with a high labor component. An important export industry for both non-maquiladoras and maquiladoras is the automotive industry - auto assembly using imported components as well as the production of autoparts exported for assembly in other countries (principally the U.S.). 1.25 Parallel with the growth of manufactured exports has been an increase in imports of intermediate inputs and capital goods used in the production of exports. This phenomenon seems to be true for both maquiladoras and non-maquiladoras. According to a 1995 enterprise survey, 82 percent of exporting firms used imported inputs, compared to 49 percent of non-exporters. "I Table 1.4: Composition of Exports (% of total) 1985 1990 1995 1999 Oil exports 55.2 24.8 10.6 7.3 Non-oil exports 44.8 75.2 89.4 92.7 Agriculture 5.3 5.3 5.0 3.0 Extractive industries 1.9 1.5 0.7 0.3 Manufactures 37.6 68.4 83.7 89.4 Maquiladoras 19.0 34.1 39.1 46.6 Non-maquiladoras 18.6 34.3 44.6 42.7 Total 100.0 100.0 100.0 100.0 Source: Cervantes (1996) and Banco de Mexico data. E. Plan of the Report 1.26 The rest of this report analyzes the performance of the export sector using micro data: observations over time at the level of the plant or firm. In the next chapter, we begin by tabulating trends in exports derived from the micro data and what they tell us about changes in export orientation, the product and market diversification of exports, the characteristics of exporters, and their transitions in and out of exporting. The purpose is to set up hypotheses to be tested in later chapters. 1 Cervantes (1996), p. 177. 12 1.27 The third chapter discusses the constraints to exporting as reported by firms themselves and presents several case studies of firms that exported during the 1995 economic crisis. It then analyzes more carefully the determinants of a firm's export behavior in a regression framework. Chapter 4 presents a graphical comparison of exporting firms, before and after starting to export, with control groups of other firms, to examine the direction of causality between exporting and enterprise performance. Chapter 5 uses regression analysis to analyze the impact of exporting on total factor productivity growth, as well as the effects of supplier linkages, spillovers, training, foreign direct investment, and high performance workplace practices. In the final chapter we propose directions for future policy reform that are suggested by the results of this report. 13 2. TRENDS IN EXPORT ORIENTATION 2.1 This chapter describes trends in the export sector using the panel databases constructed for this report. The data give a picture of the changes in the structure of exports that have occurred in the 1 990s, the characteristics of exporters, the firm concentration of exports, and transitions in and out of exporting. Looking at these trends suggests hypotheses that are tested in later chapters. We begin with a description of the data (Annex B contains a more detailed description of the sources of data and how we constructed the panel databases). A. Databases 1. EIA 2.2 The Instituto Nacional de Estadistica, Geografia e Informatica (INEGI) conducts an annual survey called the Encuesta Industrial Anual (EIA). The survey covers about 6500 manufacturing establishments (plants) throughout the country that account for about 80 percent of production in each six-digit industry group.12 Since the objective of the survey is to cover a large proportion of manufacturing production, the sample frame includes all of the largest plants in the population of firms and a significant share of medium-scale plants, but has fewer small-scale plants (in proportion to their population) and very few microenterprises.13 2.3 It is important to note that the EIA does not include plants that operate under the special maquila regime. The latter are covered by a separate census, described below. The two surveys are not strictly comparable in terms of product groups (industrial classification) and the definition of value added. 2.4 The design, sample size, and plant identification numbers in the EIA were changed in 1993, and the pre-1993 and post-1993 surveys have not been linked. However, we were able to link the plants in the 1993-97 EIA into a panel database using the unique plant identification numbers contained in the survey data. For 1998, since the EIA is not yet available, we constructed a subset of variables from the monthly manufacturing survey (the Encuesta Industrial Mensual, or EIM), which covers the same sample of plants. The EIA contains 100 variables measuring sales and other income (including sales to the domestic market as well as exports), material inputs (including those purchased domestically and imported), employment, capital stock, and investment. 12 The industry classification system is the Clasificacion Mexicana de Actividades y Productos (CMAP), 1994. 13 INEGI defines four plant size categories based on the number of employees: microenterprises (0-15), small-scale (16-100), medium-scale (101-250), and large-scale (over 250). 14 2. CIM 2.5 INEGI collects monthly data on plants that operate under the special maquila regime. The plants are identified for INEGI by the Ministry of Trade and Industry (SECOFI), which registers firms under the maquila prograrn. In contrast to the EIA, the Censo Industrial Maquiladora (CIM) includes the entire population of maquiladoras - it is not just a sample. There are fewer variables in the CIM compared to the EIA and EIM, and there is no informnation on the stock of capital (which is useful for estimating total factor productivity). Nevertheless, there is information on the consumption of electric energy, which can be used as a proxy for the capital stock. Due to the importance of the maquila industry as a source of employment, the CIM contains detailed data on employment and wages by type of contract and gender. For this report, we constructed annual values of variables from the monthly CIM, and used the unique plant identification numbers to link the data across the years 1990-98. 2.6 Table 2.1 shows the number of plants and their export value in the EIA and the CIM for 1993-98. The value of exports in the CIM (a census) is equal the to entire maquiladora sector's exports; the value of exports in the EIA (a survey) is less than the total of non-maquiladora exports, but since the EIA covers the largest plants in the sector, the excluded export value is small. Note that EIA measures direct manufactured exports only, not manufactured goods exported through trading firms, another reason why the total value of exports of plants in the EIA is less than the manufactured export total in balance of payments statistics. Table 2,1: Value of Exports and Number of Plants in the EIA and CIM (export value in N$ million) Year CIM (Maguilador) EIA (Non-Maguiladora) Total Exports No. plants Exports No. plants Exports No. plants 93 39,364 2,351 49,328 6,862 88,692 9,213 94 89,539 2,391 63,418 6,856 152,957 9,247 95 173,237 2,427 164,922 6,783 338,159 9,210 96 271,625 2,722 245,926 6,684 517,551 9,406 97 348,230 2,980 293,644 6,438 641,874 9,418 98 445,051 2,308 335,848 6,229 780,889 8,937 3. ENESTYC 2.7 INEGI fielded a special module attached to the EIA called the Encuesta Nacional de Empleo, Salarios, Tecnologia, y Capacitacion (ENESTYC) in two years: 1992 (covering calendar year 1991) and 1995 (covering 1994). The questionnaire was designed jointly with the Ministry of Labor and gathers additional information on training, technology, wages, forms of labor contracting, and internal plant organization. The sample includes a greater proportion of microenterprises and SMEs than the EIA. For this report we were able to link the plants in the 1995 ENESTYC with those in the 1994 EIA. This allowed us to combine ENESTYC data on training, quality control, and technology use in calendar year 1994 with EIA variables for the plants common to the 15 two surveys, and use them in our estimates of total factor productivity and exporting behavior. 14 The number of plants common to the ENESTYC and EIA was only about half of those contained in the EIA (i.e., about 3000). 4. Customs data 2.8 The Departamento de Medicion of Banco de Mexico maintains a database on international trade transactions that contains information received from Customs. It has data on individual import and export transactions by firm and 6-digit product, aggregated to monthly totals. Firms are identified by their tax identification number (the Registro Federal de Causantes, or RFC) - so that the observation is the firm rather than the plant. The RFC also contains information on the month and year the firm began operations. Again, this data covers the universe of firms that (legally) export and import, rather than just a sample. The data cannot be linked with the EIA panel because the observations are at the level of the firm rather than the plant, and no concordance is available. Nevertheless, the Banco de Mexico database can be used to investigate trends in the composition and concentration of exports. B. Export Products, Origins, and Destinations 2.9 The product composition of exports for the plants in the EIA sample is shown in Figure 2.1. During 1993-98, total exports grew rapidly from N$49.3 billion to N$338.1 billion. Across broad (two-digit) industry groups, the sectors with the largest share of exports were metal products and machinery (which includes automobiles) and chemical products (including petroleum products, but not crude oil). The structure of exports across industry groups remained relatively stable, with some increase registered for metal products and machinery, and a decline in chemical products reflecting falling oil prices. 2.10 Clearly, manufacturing exports are dominated by metal products and machinery (Industry 38); the share of this sector in manufactured exports rose to 70 percent in 1998. The majority of Industry 38 exports are in the automobile industry (Table 2.2 and Box 2.1), including automobile assembly, motors, and spare parts. In fact, exports of automobiles have almost reached the value of oil exports in 1997 (Annex Table A.6). However, other types of machinery are increasing even faster, especially office machinery and informatics -- particularly computers, which now rank next to autoparts in export value. 14 INEGI has also linked the 1992 and 1995 ENESTYC surveys in a panel. 16 Figure 2.1: Composition of Exports (EIA) (%)I 100% 400,000 -0% Es - = :; - -350,000 S0%- :1 1 ;u_ _ 300,000 60% 250,000 200,000 40% ____ ___- 150,000 20% _ - ~~~~~~~~~~~~~~~~~~~~~~100,000 50,000 1993 1994 1995 1996 1997 1998 _38 Metal products, machinery _31 Food, bev., tobacco 35 Chemical products i37 Basic metal industries _Other (32, 33, 34, 36, 39) Total Table 2.2: Exports of Metal Products and Machinery (EIA) (% of Industry 38 total) 1993 1994 1995 1996 1997 1998 Total 3811 Forging of metallic parts 0.4 0.4 0.5 0.5 0.6 0.6 0.6 3812 Metallic structures, tanks, etc. 0.2 0.3 0.3 0.3 0.3 0.3 0.3 3813 Metal fumiture & repair 0.3 0.3 0.3 0.3 0.4 0.4 0.4 3814 Other metal products excl. mach. & equip. 2.0 1.8 2.7 2.5 2.5 2.4 2.4 3821 Mach. & equip. for specific uses 1.9 2.2 1.7 1.5 1.8 1.9 1.8 3822 Mach. & equip. for general uses 1.5 1.6 1.3 1.2 1.4 1.3 1.3 3823 Office machinery and informatics 8.1 9.8 10.0 11.9 16.4 17.9 14.4 3831 Electric mach. & equip. 3.1 4.0 4.1 3.8 3.6 3.7 3.7 3832 Electronics (TV, commun., medical) 1.0 1.8 1.6 1.5 1.5 1.6 1.5 3833 Household equipment excl. electronics 3.8 3.4 3.2 2.9 3.0 3.0 3.1 3841 Automobile industry 77.2 73.8 73.7 72.9 67.8 65.9 69.8 3842 Transport equip. excl. cars and trucks 0.2 0.1 0.1 0.1 0.1 0.1 0.1 3850 Precision instruments excl. electronics 0.3 0.4 0.3 0.5 0.7 0.8 0.6 Total 100.0 100.0 100.0 100.0 100.0 100.0 100.0 17 Box 2.1: The Automobile Industry in Mexico NAFTA has drastically modified the policy regime that detennined the previous structure of the automotive sector in Mexico. The trade agreement gradually eliminates barriers to trade in vehicles, engines, and autoparts over a ten-year period ending in 2004. It also eliminates (i) restrictions on foreign investment in the sector, particularly in Mexican autoparts producers; (ii) domestic sales and trade balance requirements; and (iii) domestic value added requirements. The elimination of restrictions applies to trade and investrnent to and from the United States and Canada, not with third countries such as Japan. The liberalization of the industry brought about by NAFTA has made Mexico a key participant in the globalization of the automobile industry. Mexico exports assembled vehicles as well as autoparts. Production for export is concentrated in the northern part of the country (along the Monterrey-Saltillo corridor) and production for the domestic market is located in the central region. The big three U.S. firns (General Motors, Ford, and Chrysler) specialize in the production of six- to eight-cylinder models while the non-U.S. firns (Volkswagen, Nissan, Mercedes- Benz, BMW) produce mainly smaller four-cylinder vehicles. The autoparts industry includes over 500 firns. There is a network of principal components suppliers that produce high value-added components (e.g., engine blocks) for the large assemblers, achieving high standards of quality, technical specifications, and delivery time. Many of the principal components suppliers are Mexican-U.S. joint ventures organized for just-in-time production for GM, Chrysler, or Ford. There is also a group of smaller Mexican firms that produce replacement parts using more standardized technologies. Sources: Ramirez and UInger (1997,), Mercado and Sotomczyor (1 996,). 2.11 The regional distribution of exporters and of export value shows a high concentration in the northern part of the country (the states of Chihuahua, Coahuila, Durango, and Nuevo Leon) as well as the central-south region (which includes Mexico City as well as the states of Mexico, Morelos, Puebla, and Tlaxcala). Based on manufactured export data in the EIA, together these two regions accounted for 52 percent of export value (including 54 percent of exports of metal products and machinery) and 64 percent of exporters in 1998 (Table 2.3). Their dominance declined somewhat over the 1993-98 period: the Central and Pacific Central regions gained export share at the expense of the Central South, Pacific North, and other regions. However, the Central South region gained the largest number of exporters during the period (Table 2.4). 18 Table 2.3: Regional Distribution of Exports (EIA) Change in Share of Export Rank Export Rank Export Export Share Share Share Growth Region 1993 1993 1998 1998 1993-98 1993-98 North 24.8% 2 23.4% 2 -1.3% 23.2% Pacific North 10.6% 3 6.6% 5 -4.0% 5.90/a Pacific Central 9.5% 4 16.7% 3 7.2% 17.9% Pacific South 0.3% 9 0.5% 8 0.2% 0.5% Central North 8.6% 5 5.4% 6 -3.2% 4.8% Central 5.1% 7 14.0% 4 8.9% 15.5% Central South 33.2% 1 29.0% 1 -4.2% 28.2% Gulf 7.2% 6 4.3% 7 -2.9% 3.8% Southeast 0.6% 8 0.2% 9 -0.4% 0.2% Total 100.0% 100.0% 0.0% 100.0% Table 2.4: Regional Distribution of Exporters (EIA) Change in No. Rank No. Rank No. exporters exporters exporters Region 1993 1993 1998 1998 1993-98 North 299 2 371 2 72 Pacific North 72 6 80 7 8 Pacific Central 160 4 205 4 45 Pacific South 8 9 9 9 1 Central North 66 7 101 6 35 Central 168 3 213 3 45 Central South 832 1 982 1 150 Gulf 81 5 102 5 21 Southeast 44 8 43 8 -1 Total 1730 2106 376 19 Definition of Regions North Chihuahua Central Guanajuato Coahuila Hidalgo Durango Queretaro Nuevo Leon Central South Distrito Federal Pacific North Baja California Estado de Mexico Baja California Sur Morelos Nayarit Puebla Sinaloa Tlaxcala Sonora Gulf Tabasco Pacific Central Colima Tamaulipas Jalisco Veracruz Michoacan South East Campeche Pacific South Guerrero Chiapas Oaxaca Quintana Roo Central North Aguascalientes Yucatan San Luis Potosi Zacatecas 2.12 The most important destination for Mexican goods exports has traditionally been the U.S., and this tendency was accentuated with the implementation of NAFTA in 1994. In 1997, Mexico had the highest concentration of exports in a single market destination (85 percent to the U.S. in 1997) than any other country in the world.'5 The reasons behind this concentration are obvious: the geographical proximity of the U.S., which has encouraged commercial ties and familiarity with U.S. markets; the size and dynamism of the U.S. market; the development of the maquiladora industry; and the volume of U.S. foreign direct investment in Mexico (e.g., in the automobile industry). Quite some distance behind the U.S. are market destinations in Europe (4.0 percent of total exports) and the rest of Latin America (3.8 percent). NAFTA also has encouraged an inflow of FDI from third countries seeking to export to the U.S. under NAFTA preferences. 2.13 Viewed from the perspective of U.S. imports, Mexico's market share has increased significantly since the mid- 1990s, both as a result of NAFTA and the devaluation of the peso. According to Banco de Mexico customs data, the share of Mexican exports in U.S. imports rose from 6.8 percent in 1993 to 10.3 percent during January-November 1998 (Figure 2.2). Of the approximately 1,250 products imported by the U.S. in 1998, Mexico exported 1,071 - and had the largest share of the U.S. market for 102 of them (Table 2.5). '5 Banco de Mexico (1999). 20 Fig. 2.2: Mexican Participation in U.S. Imports 12.0 10. 10.0~~~~~~~~~~~~~~1. 10.0 9-3 0 .3 8.3 8.0 724l 6.1 6.3 6 .8 6.0 4.0 2.0 0.0 - _ i 1 1 1 1 1 - Year Table 2.5: Mexico Export Participation in U.S. Imports Share of U.S. Market Position in U.S. Market No. of Value Market No. of Value Products (US$m.) Position Products (US$m.) Less than 10% 775 15,896 1 102 26,328 10-20% 147 28,089 2 112 18,022 20-30% 67 13,372 3 50 7,222 30-40% 35 9,058 4 29 9,065 40-50% 15 1,718 5 3 33 More than 50% 32 10,301 Other 775 17,764 Total 1071 78,434 Total 1071 78,434 Data for Jan.-Oct. 1998. Source: Banco de Mexico. 21 2.14 A common hypothesis relating to export development is that an exporting firm moves along a learning curve that leads to greater market diversification over time. In the beginning, a new exporter chooses a market with low costs of information, transactions, and entry; over time, as the firm learns how to export, it exports to a greater number of countries. This hypothesis is supported by the data in Table 2.6, which shows the annual number of market destinations for non-maquiladora exporters during 1993-97. For example, the group of all exporters in 1993 exported to an average of 2.33 countries. Those that continued exporting sold to an average of 2.74 countries in 1994, 2.86 in 1995, and 3.11 in 1996. In every year, incumbent exporters were more diversified across market destinations than new exporters. In addition, the number of market destinations for all exporters has increased somewhat over time, from 2.33 in 1993 to 2.65 in 1996. Table 2.6: Export Market Diversification (average number of country destinations) 1993 1994 1995 1996 1993 2.33 2.74 2.86 3.11 1994 2.34 2.72 2.98 1995 2.41 3.02 1996 2.65 Data for all non-maquiladora firms exporting >=$50,000 per year. Diagonal entries: average number of export market destinations for all firms exporting that year. Off-diagonal entries: average number of export market destinations for survivors from previous year. Source: Banco de Mexico, from customs data. C. Firm Concentration of Exports 2.15 Exports in Mexico are highly Figure 2.4: Lorenz Curve for Non- concentrated across firms. On one side of Maquiladora Exporters, 1995 the distribution, a few large exporters 1.00 -.-------. account for a large share of export value; on a 2 0.80 the other side, hundreds of small exporters .0 account for a small proportion of export 0 o 60 value. The degree of concentration is high & 0.40 - even when maquiladoras, automobile , 0.20 assemblers, and PEMEX are excluded (Annex Table A.9). This situation has been d o.00 characteristic of Mexico for a number of 0.00 0.20 0.40 0.60 0.80 1.00 years, but in the early 1990s there was an Cumulative percentage of firms increase in concentration: the share of -. export value accounted for by the largest 10 exporters rose from 16.5 percent in 1991 to 19.2 percent in 1994, and the share of the largest 280 exporters rose from 59.2 percent to 22 64.4 percent during the same period.'6 Plotting this data in a Lorenz curve (Figure 2.4) shows, for example, that the top 20 percent of exporters account for about 60 percent of export value. The large majority of exporting firms sell less than US$500,000 per year (Figure 2.5), and the number of these small exporters increased during and after the crisis. Figure 2.5: Firm Concentration of Exports (by value of exports, US$000) 10,000 - _____.__ _ 8,000 - 6,000- .4,000- 2,"000 0 1992 1993 1994 1995 1996 1997 Value of Exports L o> 50 * 50-500 [3500-1,000 .1,000-5,000 w5,000-25,000 *25,000+. Figure 2.6: Export Value by Size (EIA) 350000 300000 .. JE 250000 l 200000 0 150000 _ la 100000 50000 011 1993 1994 1995 1996 1997 1998 * Large * Medium a Small/Micro 2.16. The degree of export concentration in Mexico may not be very different from the situation in other countries. Firm-level studies of export dynamics in Colombia, Venezuela, and Morocco have shown similar patterns: most exporters export only a small share of their production (between 0 percent and 10 percent), so that a few large exporters account for a large share of export value. A recent study of Colombia"7 finds three categories of firms: major exporters, firms that export on a small scale, and non- 16 Using Banco de Mexico data on non-maquiladora exporters excluding automobile assemblers and PEMEX. 17 Das, Roberts, and Tybout (draft, 2000). 23 exporters. Movement of firms between the second and third categories is common, but does not have a large effect on total export volumes. D. Characteristics of Exporters 2.17 The EIA panel database gives information on the characteristics of exporters and non-exporters during the 1993-98 period. According to this data, during the economic crisis the number and share of plants that exported rose significantly, from 26 percent of plants in 1993 to 42 percent in 1997, then falling back to 37 percent in 1998 (Figure 2.7). Figure 2.7: Export Status of Plants (EIA) 100% __ 80%- -60%-_ 0 40% l L 20% 1993 1994 1995 1996 1997 1998 *FMqporters X Non-expoers] _ _ _ _ _ --7 ___ Figure 2.8: Export Status by Size (E1A) 0.70 0.60 __ _ 0 0.50 _ -- - F m0.40 0.30 L L * * G 0.20 0.1 0.00 1993 1994 1995 1996 1997 1998 g_e Lrg *Medium D SmallM icro 2.18 The increase was most pronounced for large-scale plants: from 43 percent of plants in 1993 to 63 percent in 1997 -- a 48 percent increase (Figure 2.8). The shares of medium-scale and micro/small-scale plants that exported both increased by 37 percent between 1993 and 1997. As expected, the proportion of plants that export increases with plant size: between 1993 and 1998, the share of plants that exported some of their 24 production was about 21 percent for small/micro-scale plants, 38 percent for medium- scale, and 53 percent for large-scale plants. 2.19 Figure 2.9 shows export status by ownership categories. The data suggest that the share of plants that export is higher among joint ventures (averaging 58 percent over 1993-98) than for foreign-owned plants (averaging 42 percent), but this may be due to the fact that maquiladoras are not covered in the EIA. Nevertheless, foreign-owned plants account for the largest share of export value (Figure 2. 10). Figure 2.9: Export Status by Ownership (EIA) 0.80 0 O. 0.70 x 0.60 __ ___ _ _ 'S 0.50 E 0.40 - _ 0.30 0 -0.20 0.10 0.0.00 1993 1994 1995 1996 1997 1998 * Domestic * Joint Venture C Forein Figure 2.10: Export Value by Ovnership(EIA) 250000 ?200000 _ ____ _ _ _ _ 1, 50000 ______ _ ____ > 100000 0 50000 1993 1994 1995 1996 1997 1998 * Domestic mJoint Venture E Foreign 2.20 The greatest numbers of exporters are in industries 38 (metal products and machinery) and 35 (chemical and petroleum products), and lowest in industries 33 (wood and wood products), 34 (paper and paper products) and 37 (basic metal products). Those industries showing the largest gain in the number of plants that exported during 1993-97 25 were 32 (textiles and garments) and 38 (metal products and machinery); only industry 34 (paper and paper products) was had fewer exporters in 1997 than in 1993. 2.21 The export orientation of plants -- defined as the value of exports divided by total sales -- rose from an average of 21.7 percent in 1993 to 27.4 percent in 1998 (Figure 2.11). It is interesting to note that the share of production that is exported is not very different for large-scale versus medium-scale versus small/micro-scale plants. However, large plants increased their export orientation during the period -- especially during 1995 -- more than did smaller plants. Increases in the share of output exported were apparent for all ownership categories, but particularly for joint ventures (Figure 2.12). Figure 2.1 1: Export Orientation by Size (EIA) 35% 30% _ _ _ _ __ _ _ _ .~25% __ 20% 0% 1993 1994 1995 1996 1997 1998 I Large u Medium [ Small E. Export Transitions and Export Permanence 2.22 The number of exporting firms in a given year is the sum of those that exported the year before, plus those that began exporting less those that ceased exporting. An analysis of the entry and continuation of exporters was undertaken by Banco de Mexico. Table 2.7 shows, on the diagonal, the number of non-maquiladora firms (RFCs) that exported US$50,000 or more during the year. The entries above the diagonal show the number of firms that continued exporting from the previous period, and the entries below the diagonal show the number of firms that ceased exporting (compared to the year on the diagonal). For example, in 1992 there were 8,596 firms that exported at least US$50,000 each. Of these, 5,401 continued exporting in 1993, 4,534 continued exporting in 1994, etc. Looking below the diagonal, of the 8,596 firms that exported in 1992, 3,195 did not export in 1993, 4,062 did not export in 1994, etc. 26 Table 2.7: Export Permanence (no. of firms) 1992 1993 1994 1995 1996 1997 1992 8596 5401 4534 4196 3598 3097 1993 3195 8451 5752 5117 4229 3621 1994 4062 2699 8765 6286 5100 4294 1995 4400 3334 2479 11789 7428 5976 1996 4998 4222 3665 4361 12179 7855 1997 5499 4830 4471 5813 4324 11557 Non-maquiladora firms that exported at least US$50,000 during the year (US$25,000 Jan.-June for 1997). Source: Banco de Mexico. Figure 2.13: Entry and Continuation of Exporters 12,000 10,000 8,000 No. firms 6,000 4,000 2,000 period I per. 2 per. 3 per. 4 per. 5 per. 6 _- _992- 1993-- 1994-- 1995--) 199 . 2.23 The first thing to note is the important increase in the number of exporters in the years during and after the crisis: from 8,765 in 1994 to 11,789 in 1995 and 12,179 in 1996. The number of exporters then fell back somewhat in 1997 to 11,557. From the table it is also possible to see that (a) the largest numbers of "drop-outs" occur in the first year after exporting, and (b) the number of drop-outs decreased in 1995-96 compared to years before and after the crisis. 27 2.24 Table 2.8 takes one year, 1995, and shows the breakdown of exporters between incumbents and new exporters, and for the latter, the proportion who were new versus existing firms. In that year, excluding possible errors (mainly due to keypunching errors), about 80 percent of exporters were incumbents and 20 percent were new exporters. Of the new exporters, (a) 4,404 were non-maquiladoras and 167 were maquiladoras; (b) of the non-maquiladoras, only 297 were new firms in the sense that they initiated operations in that year18; (c) of the 167 new maquiladora exporters, only 43 were new firms (implying that 124 were firms that moved from non-maquiladora status to maquiladora status); (d) for both maquiladoras and non-maquiladoras, new exporters have significantly lower average export values than incumbents, and firms that were born as exporters had higher export values than existing firms that became exporters; and (e) the number of possible errors is substantial, but the value of exports accounted for by these firms is small. Table 2.8: Incumbent and New Exporters in 1995 No. of firms Exports Avg. Exports (US$m.) (US$m.) All firms 29,608 79,823 2.696 Incumbents 18,388 76,508 4.172 New exporters 4,571 2,487 0.544 New firms 340 322 0.947 Existing firms 4,231 2,165 0.512 Possible errors 6,699 828 0.124 Non-maquiladora 27,335 48,512 1.775 firms Incumbents 16,489 45,981 2.789 New exporters 4,404 2,020 0.459 New firms 297 187 0.630 Existing firrns 4,107 1,833 0.446 Possible errors 6,442 511 0.079 Maquiladoras 2,273 31,311 13.775 Incumbents 1,849 30,527 16.510 New exporters 167 467 2.796 New firms 43 135 3.140 Existing firms 124 332 2.677 Possible errors 257 317 1.233 Note: 114 firms exported as maquiladoras and non-maquiladoras. For this exercise, these parts were considered as two separate firms. Source: Banco de Mexico. 2.25 Export transitions were also calculated using the sample of exporters in the EIA. We define an "export transition" variable as follows: 18 Since the firm identification number (RFC) includes the date of birth of the firm, it is possible to distinguish new firms from those that existed in prior years. 28 Export transition category Export Status (t) Export Status (t-l) 1 =incumbent exporter 1 1 2=entrant exporter 0 1 3=exiting exporter 1 0 4=non-exporter 0 0 2.26 Export transitions are defined by looking at two periods only. '9 This means that "entrants" may be new exporters or re-entering exporters. Though incumbent exporters far outweigh entrants in all years, the number of entrants increased sharply in 1995 and remained high through 1997. Figure 2.14: Export Transitions (ETA) 100% 80% .60% 40% o20%-__ .. 0% 1994 1995 1996 1997 1998 1 1-Incumbent * Entrant j Exit 1 Non-exporter 2.27 Though all sizes of firms broke into export markets in 1995, the increase was more pronounced among smaller plants; larger plants seem to have more stable patterns of entry into and exit from exporting (Figure 2.15). This conclusion is supported by looking at the average number of years of exporting by plant size: over the six-year period 1993-98, large-scale plants exported an average of 3.2 years, medium-scale plants 2.3 years, and micro/small-scale plants 1.4 years. 2.28 We can decompose the change in the value of exports from one year to the next, to that attributable to incumbents, entrants, and exits (by definition, non-exporters do not contribute to export growth).20 Table 2.9 and Figure 2.17 show the results of this exercise. Over the 1993-98 period, total manufactured exports grew N$273.2 billion. Of this amount, incumbents accounted for the great majority, N$245.3 billion (89.8 percent), entrants for N$34.8 billion (12.7 percent), and exits N$-6.9 billion (-2.5 percent). Although the largest increase in total exports was in 1995, the largest contribution of 9 Note that this two-period definition of "entrant" and "exit" differs from that of Chapter 4, which looks across the entire 6-year period. 20 This calculation follows the methodology in Roberts and Tybout (1997a). 29 entrants (30.1 percent) was in 1996. A closer look at the contribution of entrants to 1996 export growth reveals that most of entrants' exports -- N$18.5 billion out of N$20.5 billion -- came from the automobile industry in the central region of the country. If the automobile industry is excluded, the contribution of entrants to export growth in 1996 is smaller and more sirnilar to the other years in the sample (Table 2.10 and Figure 2.18). 2.29 As would be expected, entrants are less export oriented than incumbent exporters -- on average, the export share of total sales for entrants is 10.4 percent versus 28.2 percent for incumbents. Plants that cease to export also have lower export orientation than incumbents (9.5 percent). Figure 2.15: Export Entrants and Exits by Size (EIA)- 600 400 __ -200 _ _ -400 L -600 1994 1995 1996 1997 1998 Lo Large * Medium a Micro/Small Figure 2.16: Export Entrants and Exits by Ownership (EIA) 800 6200 400 Z200 -200 -400 1994 1995 1996 1997 1998 | Domestic * Joint Venture o Foreign 30 Table 2.9: Decomposition of Export Growth (EIA) Year Incumbents Entrants Exits Total N$m. % N$m. % N$m. % N$m. % 1994 13,411.2 97.4 1,789.9 13.0 -1,433.0 -10.4 13,768.2 100.0 1995 96,349.3 95.1 5,965.2 5.9 -992.0 -1.0 101,322.5 100.0 1996 48,371.4 70.9 20,543.2 30.1 -669.7 -1.0 68,244.9 100.0 1997 41,353.4 90.7 5,090.9 11.2 -842.0 -1.8 45,602.2 100.0 1998 45,830.6 103.5 1,398.8 3.2 -2,931.2 -6.6 44,298.1 100.0 Total 245,315.9 89.8 34,788.0 12.7 -6,867.9 -2.5 273,235.9 100.0 Figure 2.17: Decomposition of Export Growth (EIA) 120000 100000 80000 SOOo __ _____ __ x 60000 _ 0> t 40000 _ * 20000 -20000 1994 1995 1996 1997 1998 * Incumbwents Entrants 6 Exits Table 2.10: Decomposition of Export Growth Excluding Automobile Industry (3841) Year Incumbents Entrants Exits Total N$m. % N$m. % N$m. % N$m. % 1994 8,024.7 98.1 1,498.2 18.3 -1,341.0 -16.4 8,181.8 100.0 1995 52,754.1 92.1 5,409.2 9.4 -886.2 -1.5 57,277.1 100.0 1996 30,454.5 95.4 2,090.2 6.5 -611.3 -1.9 31,933.4 100.0 1997 28,224.7 87.4 4,823.4 14.9 -741.3 -2.3 32,306.9 100.0 1998 24,682.2 104.4 1,336.0 5.6 -2,369.5 -10.0 23,648.7 100.0 Total 144,140.2 94.0 15,156.9 9.9 -5,949.3 -2.5 153,347.8 100.0 31 Figure 2.18: Decomposition of Export Growth, Excluding Automobile Industry (3841) 70000 60000 ,Z 50000 _ __ * I , 40000 __ 3~0000__ 2 30000 0. -10000 1994 1995 1996 1997 1998 L .cimbent lEnransE as its F. Conclusions 2.30 Simple tabulations of the data in the 1993-98 EIA panel and the 1990-98 customs data panel suggest the following trends in the export sector during the 1990s: * Within manufacturing, the distribution of exports across two-digit industries has remained relatively stable over 1993-98. Industry 38 (metal products) is dominant; within that sector, the automotive industry accounts for the majority of exports, though the contribution of computer equipment is rising. * The U.S. is overwhelmingly dominant as the destination of Mexican exports. Nevertheless, there seems to have been some diversification in destinations during the 1 990s, reflected in an increase in the average number of destinations per firm. Mexico's share of the U.S. import market has increased with NAFTA, and the country has the leading market position in a large number of products. * Exports appear to be highly concentrated across firms, although we should have cross-country data to be able to draw strong conclusions. Export concentration increased through 1994, but the number of small exporters (defined by export value) increased during and after the crisis. * The number and proportion of non-maquiladora firms that exported rose significantly during the 1993-98 period: customs data shows an increase from about 8,500 exporting firms before the crisis to around 12,000 in 1995-97; the EIA shows that the share of plants that export rose from 26 percent to 42 percent between 1993 and 1997, before falling back to 37 percent in 1998. The increase was most pronounced for 32 large-scale firms, but the share of SMEs that exported also increased. Ownership matters: joint ventures and foreign-owned firms (including maquiladoras) are more likely to export, and export value is dominated by foreign-owned firms. Though the increase in the number of new exporters in 1995 was apparent for all size categories, the increase was most pronounced arnong small plants; larger plants seem to have more stable patterns of entry into and exit from exporting. Over the 1993-98 period, incumbent exporters accounted for almost 90 percent of export growth. Entrants accounted for a significant part of export growth in during and after the crisis, though during 1995 most of the contribution from entrants came from the automobile industry. 33 3. THE DECISION TO EXPORT 3.1 Why do exports in Mexico continue to be concentrated among a relatively small number of firms? Why haven't more firms taken advantage of the export opportunities created by trade liberalization and regional trade agreements? In this chapter, we examine the constraints to exporting as reported by firms themselves, using recent enterprise surveys conducted by Banco de Mexico/SECOFI and Bancomext/INEGI. We then use the EIA panel to empirically investigate the determinants of current export status and the decision to move from non-exporter to exporter status. A. Case Studies 3.2 First, to give some insight into the process of breaking into exporting and the choice of financing, we present three "stories" of SME exporters. They are based on a series of interviews of SME exporters conducted by Bank staff during 1997, when the economic crisis and favorable exchange rate had led many non-exporters to consider entering export markets.21 Box 3.1: Madermont S.A. de C.V. Madermont produces rustic-style furniture for the Madermont's equipment, which it imports from home. Its single plant in Mexico City has been in Germany, is fairly sophisticated for the industry. operation since 1993. The owners had emerged A Total Quality Management system is in place, from a previous business, making wooden doors and the company places a high priority on worker and toys, that went bankrupt in the late 1980s. training and flexible organization. The owners The firm has four owners (not of the same family) feel that their main constraints to growth are and 48 employees. related to public sector inefficiencies and over- regulation. From the beginning, Madermont's strategy has been to penetrate external markets by using Madermont has never used bank credit, in part classic Mexican designs produced with because of the owners' previous experience with standardized techniques and high quality. Other a firm that went bankrupt. In that prior case, an manufacturers of this type of furniture use unanticipated devaluation of the peso left them "artesan" techniques, making each piece unable to meet debt service obligations. Now, the individually. Over 90 percent of the firm's sales owners have decided to meet their investment and are for export, mainly to retailers in the U.S. and working capital needs internally. Europe and increasingly to Asia. 21 The interviews were conducted as part of the research for "Mexico: Strengthening Enterprise Financing", Report No. 17733-ME, September 1998. 34 Box 3.2: Interquimia S.A. de C.V. A family-owned firm with 300 employees, Interquimia found export markets by attending Interquimia produces over 200 products for trade fairs in the U.S., with some assistance from construction and maintenance including sealants, Bancomext. It has invested in equipment and paints and anti-corrosives, and roofing materials. quality control systems, and is getting IS09000 Its single plant in Tecama, Mexico has operated certification, mainly to be able to participate in for 30 years. external markets. It meets quality requirements in the U.S. market, and competes primarily on the Most of the company's sales are to the domestic basis of price. Products are transported by truck market, but during the economic crisis, when the and train to the U.S. and by sea to Cuba and the domestic construction market declined sharply, Caribbean. Inefficient port facilities are a major exports rose from 3 percent to 15 percent of total constraint to exporting, though port services have sales. The firm's largest external market is the improved somewhat since privatization. U.S., but it also sells to Cuba, Central America, Puerto Rico, and other Latin American countries; Based on his experience with previous recently it has sold roofing materials to Singapore devaluations and other macroeconomic shocks, and Hong Kong. The general director's son is the general director has a very cautious financing credited with pushing the company to expand to strategy. Most investments as well as working new markets. capital are financed from the company's own resources. In 1994 the company contracted During the 1995 crisis, capacity utilization fell NAFIN-financed loans through a credit union, but from 80 percent to 20 percent, employment fell ran into payments difficulties during the crisis by about half, and sales by about a third. that forced it to sell assets. Since then, it has been Nevertheless, as domestic demand recovered in current on its remaining debt. The firm is the first six months of 1996, many former interested in finding a minority foreign partner, employees were hired back, and the firm began to but is unwilling to sell a controlling interest. earn a reasonable profit. Box 3.3: Forja Collection Mexico S.A. The company has produced metal furniture in its are well-known companies like Ralph Lauren. single plant in Tlalnepantla, Mexico (near Mexico Bancomext supported the firm's attendance at City) since 1992. Forja is a family-owned trade fairs, and additional customers were found company and has a sales subsidiary in San via recommendations of current customers. Forja Antonio, Texas. The furniture is mainly for hopes to expand its exports and to open a plant in indoor use but can be painted for outdoor use. Monterrey, Mexico. The main structure of the furniture is assembled in the plant; small parts including glass table tops The main barrier to faster expansion has been a and fabric seats are subcontracted out. The firm lack of credit: the owner estimates that the firm has 130 employees (most of whom have just a would be twice as large by now if it had received primary education) and produces about 1300 adequate investment financing. It also would pieces per week. Production does not require have integrated more rapidly, producing all sophisticated technology, but quality control is furniture parts itself instead of subcontracting. important. The firm had sought a line of credit with Banco Cremi, but its application was denied when Cremi Forja was established as a dedicated export was purchased by BBV. Credit is available from company. Nevertheless, as the exchange rate Bancomext but approval can take up to six weeks, became overvalued prior to 1994, it increased its and the application process is costly in terms of sales to the domestic market to 70 percent of total the owner's and his accountant's time. As a sales. During the economic crisis, exports result, Forja's investments and working capital returned to 90 percent of total sales. Export sales are financed mainly from internal resources along are to the U.S. (Miami, North Carolina, New with suppliers' credit. Advance payments are York, and Texas) as well as the Virgin Islands sometimes received for small export orders from and Italy. Several of the company's U.S. buyers the U.S., but this is not common for larger orders. 35 3.3 The case studies paint a picture of SME exporters that aware of the benefits of selling in external markets, but ready to shift their sales to the domestic market should there be a change in the relative conditions between internal and external markets. Their experience suggests that the decision to export is more a response to recessionary markets at home than the potential profitability of external markets on their own, and that recovery in the domestic market causes a shift back to non-exporter status. There seems to be a set of firms that "switch" between domestic and export markets but that prefer to maintain a presence at home. In fact, other interviews suggested that firms that had expanded their exports during the crisis feared that, if they failed to return the domestic market quickly as it recovered, they would lose domestic market share to competitors. 3.4 The case studies also suggest that there is a period of investment in IS09000 certification and quality control systems before or early in the process of entering export markets, and that finns recognize that quality control is important to be competitive abroad. Investment in training may also be an important precursor to entering export markets. It is not clear what role subcontracting plays in the decision to export or the firms' competitiveness; the scope of the firm's activities is based on a broader competitive strategy. 3.5 The firms interviewed during the crisis were conservative in their financing strategies, having lived through previous macroeconomic crises and witnessed the impact of macroeconomic shocks on their ability to service debt. Investment in new equipment and expansion of capacity are often financed internally, at least by SME exporters. Some firms seem to choose a growth strategy within their internal resource availability, avoiding indebtedness even if credit is available. Others say that they would expand more rapidly if credit were available, so that their growth is constrained by a lack of credit. The latter view is often reflected in firms' responses to questions about export constraints in enterprise surveys - many surveys conclude that lack of access to credit (or the high cost of credit) is the most important constraint to exporting. B. Enterprise Surveys 3.6 Two recent enterprise surveys provide insight into the constraints and opportunities facing firms attempting to enter external markets: * BancomextlINEGI22 : a survey of exporters and non-exporters in twelve cities, designed by Bancomext and implemented in February-March 1998. The sample included 277 direct exporters, 55 "indirect exporters" (suppliers to direct exporters) and 89 "potential exporters" (firms in industries identified by Bancomext as having significant export potential). The Bancomext survey focuses on medium-scale exporters - those that exported between US$40,000 and US$4 million per year. * Banco de Mexico/SECOFI23: the most recent in a series of biannual surveys of exporters covering 414 non-maquiladora exporters that participate in the PITEX and 22 "Encuesta sobre la problematica del sector exportador y el conocimiento y satisfaccion de los productos y servicios ofrecidos por Bancomext." Bancomext, June 1998. 23 Banco de Mexico/SECOFI (1999). 36 ALTEX temporary import regime, plus 342 maquiladoras. The first group accounts for 65 percent of non-maquiladora exports and the second accounts for 25 percent of maquila exports; together they account for 45 percent of the total value of non-oil exports. The firms in the Banco de Mexico survey are highly export-oriented; their exports account for an average of 59 percent of total sales. 3.7 The Banco de Mexico survey finds that a majority of firms, both non- maquiladoras and maquiladoras, are optimistic about their opportunities in external markets. Non-maquiladora exporters expect to expand their exports by an average of 7.7 percent in the first six months of 1999 compared with the same period in 1998. Most expected to expand their imports of materials even more, by an average of 8.9 percent for non-maquiladoras. Both the non-maquiladoras and maquiladoras, particularly the larger exporters, planned increases in investment in 1999. 3.8 Bearing in mind that the Banco de Mexico survey covers only the highly export- oriented firms, the following factors were seen by them as limiting export growth (Table 3.1): * for non-maguiladoras, falling international prices for their products, high cost of domestically-purchased inputs, and scarcity and high cost of domestic credit; * for maguiladoras, the weakness of demand in external markets, greater competition from Asian products, scarcity of qualified labor, and bureaucratic red tape in Mexico. 3.9 The Bancomext survey provides evidence that exporters survived the 1995 economic crisis better than non-exporters, as seen in their sales performance and capacity utilization. Nevertheless, although over 90 percent claimed to have a "permanent" exporting strategy, nearly 40 percent planned to shift toward the domestic market as demand recuperated. 3.10 Most of the direct exporters surveyed had started to export during the past five years, and the majority said they had no problems exporting. More than half were able to start exporting within two months of initiating the process. Reported problems in exporting varied by firm size (30 percent of large-scale firms, 38 percent of medium- scale, and 45 percent of small-scale). Smaller firms said they faced greater problems in the availability of credit, but larger firms (which are more highly leveraged) placed greater emphasis on the cost of credit. Firms that did not export directly - the indirect and potential exporters - indicated their main reasons for not exporting were lack of interest (27 percent), lack of production capacity (26 percent), and lack of knowledge of export markets and specific clients (14 percent). 37 Table 3.1: Constraints to Increased Exports (% distribution of responses) Non- Maquiladoras Maquiladoras External Factors 22.7 24.8 Lower demand for exported products 7.2 11.4 Greater competition from Asia 6.5 8.5 Fall in international prices 9.0 4.9 Financial Factors 28.9 23.2 Scarcity of credit in Mexico 7.9 6.2 Lack of internal resources 4.3 5.8 High cost of domestic credit 8.4 4.7 Scarcity of foreign credit 0.8 0.4 High cost of foreign credit 0.7 0.3 Scarcity of suppliers credit 1.2 1.3 Uncompetitive exchange rate 5.6 4.4 Costs 26.1 26.7 Higher cost of imported inputs 6.5 3.4 Higher cost of domestic inputs 9.9 5.7 Low quality of domestic inputs 3.7 3.4 Increases in labor costs 2.8 5.5 Scarcity of qualified labor 3.2 8.8 Other factors 22.3 25.3 Bureaucratic red tape in Mexico 6.5 9.8 Tariffs and other barriers in foreign markets 4.4 3.2 Lack of official support in Mexico 4.0 4.2 Transportation problems 2.6 3.5 Communication security problems 4.8 4.6 Total 100.0 100.0 Source: Banco de Mexico/SECOFI survey of 756 exporters, Jan.-Feb. 1999. 3.11 Asked what they needed in order to export, the majority of firms - both exporters and non-exporters -- emphasized the cost and availability of credit, particularly long-term credit. Some firms also cited the importance of information on clients abroad (14 percent) and marketing of their products in external markets (9 percent). 38 3.12 The two surveys provide information on the sources of financing used by different groups of firns. According to the Bancomext survey, direct exporters are more highly leveraged, borrow a larger share of their credit abroad, and use more long-term financing than the indirect exporters and potential exporters. The Banco de Mexico survey highlights the importance of external financing both from foreign banks as well as from the foreign headquarters of Mexican affiliates (Table 3.2). For non-maquiladoras, retained earnings are by far the most important source of finance, for large-scale exporters as well as smaller ones. Table 3.2: Sources of Financing for Exporters (% distribution of responses) Large Medium Small Total Non-Maquiladoras Retained earnings 57.9 52.7 61.2 57.2 New investment from domestic partners 0.3 4.3 7.8 1.4 Credit from Mexican banks 3.3 7.9 2.1 4.1 Non-bank financing in Mexico 1.3 4.8 5.4 2.2 Credit from foreign banks 13.7 17.3 2.9 13.6 New investment from foreign partners 1.0 2.6 8.7 1.8 Suppliers credit 0.9 5.3 4.3 1.9 Credit from foreign headquarters 20.7 4.4 6.8 16.9 Own resources located abroad 0.0 0.0 0.3 0.0 Other foreign financing 0.9 0.7 0.5 0.9 Total 100.0 100.0 100.0 100.0 Maquiladoras Retained earnings 4.1 28.6 24.2 4.9 New investment from domestic partners 0.0 0.0 14.6 0.3 Credit from Mexican banks 0.0 10.6 15.5 0.5 Non-bank financing in Mexico 6.8 1.1 5.8 6.7 Credit from foreign banks 0.2 0.0 0.2 0.2 New investment from foreign partners 19.0 49.7 17.4 19.5 Suppliers credit 3.3 2.2 6.0 3.3 Credit from foreign headquarters 41.6 0.0 7.2 40.2 Own resources located abroad 0.0 0.0 3.1 0.1 Other foreign financing 25.0 7.8 6.0 24.3 Total 100.0 100.0 100.0 100.0 Source: Banco de Mexico/SECOFI survey of 756 exporters, Jan.-Feb. 1999. Responses to question on the financing of investments planned for the first half of 1999. 39 C. Regression Analysis 1. Export Status 3.13 Recent theoretical and empirical work has contributed to our understanding of export behavior and the export supply response to trade liberalization and other macroeconomic conditions. Theoretical models of the "sunk costs" of entry into export markets suggest that current market participation is affected by prior experience.24 This means that transitory policy changes or macro shocks can lead to permanent changes in market structure, so that trade flows may not be reversed when a stimulus is removed. In other words, sunk entry or exit costs produce "hysteresis" in trade flows.25 At the level of the individual firm, this phenomenon is observed as a kind of "permanence" in behavior: even in the face of increasing market incentives to export, the firm may choose to remain in the domestic market because of the fixed costs associated with entering export markets. 3.14 Empirical evidence on the sunk cost-hysteresis framework using aggregate trade- flow data has been inconclusive, but more recent studies that use firn- or plant-level data suggest that prior export experience is an important determinant of current export status and that it can increase the profitability of exporting.26 Using the EIA panel database, we are able to examine this hypothesis for the Mexican industrial sector. 3.15 Suppose that the current export status of plant i (Xjt) is a function of three types of variables: current market variables exogenous to the plant, state variables specific to the plant, and prior export experience: (X)i= C + y (R)it + ai (pj)t + E E (S )it + F_it i k where: Xit = 1 if plant i exports in year t 0 if plant i does not export in year t Rit = export experience variable for plant i in year t Pt = set of j market variables in year t Sit set of k plant characteristics for plant i in year t 3.16 The export experience variable requires some explanation. If sunk costs associated with breaking into export markets are important -- in other words, if there is some permanence or "hysteresis" in exporting -- then past export behavior will be a significant determinant of current export behavior, over and above the influence of 24 See the theoretical papers by Baldwin (1988, 1989), Baldwin and Krugman (1989), Dixit (1989), and Krugman (1989) and the empirical studies by Roberts and Tybout (1995, 1997). 25 Roberts and Tybout (1997), p. 545. 26 For example Roberts and Tybout (1997) for Colombia. 40 current market variables and plant characteristics. Some previous empirical studies have used the lagged value of exports, or lagged export status, to reflect sunk costs.27 However, a significant coefficient on once-lagged export status may just indicate serial correlation in export status -- an export-prone firn is likely to export for more than one period -- rather than the importance of sunk costs. Ideally, a test of hysteresis would investigate whether the entry and exit responses to exchange rate movements were asymmetrical. In this paper, we try to improve upon the use of lagged export status by constructing an "export experience" variable defined as the number of current and prior years that the plant exported. We also include the squared value of this variable in the regressions, to see whether accumulated experience has a quadratic effect on current export status. For comparison purposes, we also run the model using lagged export status instead of the export experience variable. Table 3.3 shows an example of the export experience variable for a plant that moves from non-exporter status to exporter status in year t=1, exports for three years, and then ceases to export. The remaining independent variables that we tested, measuring market conditions and plant characteristics, are shown in Table 3.4. Table 3.3: Export Experience Variables Time t O 1 2 3 4 5 Current export status ("dexport") 0 1 1 1 0 0 Export status in previous period ("dexport 0 0 1 1 1 0 (t-l)") Years of export experience ("texport") 0 1 2 3 3 3 Years of export experience, squared 0 1 4 9 ("texport2") 3.17 Since Xit is a binary dependent variable, we use a probit regression model applicable to cross section-time series data.28 The estimated coefficients measure the change in the probability of exporting, per unit change in the independent variable. If entry into exporting is influenced by a decline in domestic demand, the coefficient on real GDP should be negative; however, current exports enter into current GDP so the effect could be positive. An increase in the exchange rate index indicates a devaluation of the peso, so the coefficient on this variable should be positive. Higher real interest rates would be expected to affect exporting negatively if exporting requires new investment, but if production for export comes out of underutilized capacity, interest rates may not be so important. Interest rate variability would be expected to reduce the probability of exporting. Table 3.4: Independent Variables Used in Regressions Independent Variables Definition Market Conditions: GDP (gdp) National real GDP at 1993 prices State-level GDP (gdp-est) State-level real GDP at 1993 prices 27 e.g., Roberts and Tybout (1995). 28 We also experimented with a logistic model and with a probit model with cross section/time series differentiation; the results were roughly similar. 41 Exchange rate (exchrate) Real export/domestic price ratio at 2-digit level, 1994=100 Interest rate (iavg) Real CETE rate, annual average constructed from monthly nominal rate less monthly inflation rate Interest rate variability (istdev) Standard deviation of monthly real CETE rate over the year Wages (wage) Real wage rate index, Dec. 1992=100 Plant Characteristics: Industry (indl-ind9) Dummy for each 2-digit industry 32-38 (versus 31) Region (regl- reg9) Dummy for each region 2-9 (versus 1) Size (size) Micro/small=3, medium=2, large=1 Size dummy (sizedum) Dummy for medium, large (versus micro/small) Assets (Inassets) Fixed assets, in logarithmic terms Ownership (owner) Mexican ownership 100%=l, 51-99%=2, 0-50%=3 Ownership dummy (fordum) 1,0 for some foreign capital or not IS09000 achieved (iso9OOOa) 1,0 for IS09000-certified or not in 1994 IS09000 initiated (iso9000b) 1,0 for initiating IS09000 certification or not in 1994 IS09000 achieved or initiated =1 if IS09000a=1 or ISO9000b=1, =0 otherwise (iso9000) Quality management (quality) 1,0 for quality mgmt. system or not in 1994 Workplace processes (process) 1,0 for adoption of high-performance workplace practices in 1994 or not Labor training (training) 1,0 for formal in-firm training or not in 1994 Labor quality (Iqual) percent of workers with adequate qualifications (as reported by owner/manager) in 1994 Total Factor Productivity (tf) Estimated as residual in regression on ln(value added) 3.18 Based on the characteristics of exporters we saw in the previous chapter, we would expect the probability of exporting to increase with firm size and the degree of foreign ownership. If exporters are more efficient than non-exporters (regardless of the direction of causality), the coefficient on total factor productivity (TFP) will be positive.29 The presence of a quality management system, high-perfornance workplace practices, IS09000 certification, qualified labor, and in-firm training should also be associated with a higher probability of exporting (again, without being able to determine causality).30 3.19 Table 3.5 shows the results of three versions of the regression model. The first captures time trends with year dummy variables, and uses the plant's export status in the previous year to represent export experience. The second includes macroeconomic variables rather than time trends, uses the accumulated years of export experience, and experiments with different plant characteristic variables. The third model is similar to the second, but excludes plants in the automobile industry to see how much this important export industry is driving the results. Several other variations on the model were tried but are not shown here. 29 See Chapter 4 for a description of how TFP is estimated. 30 The quality and training data is taken from the ENESTYC survey, which covers the year 1994 only. The 1994 values are assigned to those EIA plants that were covered by the ENESTYC survey, for every year they appear in the EIA panel. This means that the regression results reflect the influence of the 1994 values over time. It also means that the number of plants in regressions that use the ENESTYC variables falls by about half compared to regressions that do not use these variables. 42 3.20 Previous export behavior, whether measured as the plant's export status in the previous year or as the number of years of export experience, is highly significant. These results are consistent with empirical results obtained in other studies and suggest that there may be sunk costs involved in exporting. However, these results may be driven mainly by simple time trends. The squared value of export experience is negative and significant: in other words, the probability of being an exporter increases with additional years of export experience, but at a declining rate. This could be consistent with a learning-by-exporting model in which firms find out in the first year or two whether they are able to compete in external markets, and if they are, they become "permanent" exporters. 3.21 The macroeconomic variables enter as expected: the probability of exporting rises when domestic demand falls, rises with a devaluation of the peso, and is negatively related to both the real interest rate and interest rate variability. These results were consistent across other variations of the model, as long as national GDP was used to measure domestic demand. However, when we used state-level GDP instead of national GDP, the signs on the macroeconomic variables changed - suggesting that the results with the macroeconomic variables are not very robust. When the real wage rate index was included in the regression, it was dropped due to its high correlation with other macroeconomic variables. 3.22 The coefficients on the industry dummies measure the change in the probability of exporting for plants in a given industry relative to Industry 31 (food, beverages, and tobacco) after size and ownership are taken into account.31 There are clearly important differences in the probability of exporting across industries, even when macroeconomic conditions and industry and plant characteristics are taken into account. 3.23 Not surprisingly, plant size makes a difference: large-scale plants are the most likely to export, followed by medium-scale and then small-scale plants. Calculations of predicted export probabilities (using Model 2) shows that the expected probability of exporting increases over 1994-97, as do the actual proportions of exporting plants, but the model tends to underestimate the probability of exporting for medium-scale and small- scale plants (Table 3.6).32 For large plants, the model underpredicts export probabilities. 31 Regional dummies were also included in all regressions to control for location effects; as a group they generally were not significantly different from zero. We do not report these results in the table. 32 Predicted probabilities are calculated holding independent variables at their sample means for each of the size groups and years. 43 Table 3.5: Probit Models of Export Status Dependent variable: Current export status (= 0 if non-exporter, = 1 if exporter) Model 1 Model 2 Model 3 (using lagged export status, (using export experience, (excluding auto industry) year dummies) macro variables) Explanatory Variables Coeff. Z-Stat. Coeff. Z-Stat. Coeff. Z-Stat. constant term -2.581 -34.786 ** 26.978 8.708 ** 26.734 8.661 ** Export Experience export status (t-1) 2.281 92.817 ** export experience 2.606 63.861 ** 2.588 64.260 ** export experience squared -0.335 -39.799 ** -0.332 -39.963 ** Market Conditions exchange rate 30.144 3.234 ** 28.983 3.132 ** interest rate -0.505 -6.180 ** -0.497 -6.113 ** interest rate variability -0.273 -9.188 ** -0.269 -9.118 ** GDP -0.021 -10.270 ** -0.020 -10.228 ** Year Dummy Variables 1994 -0.249 -7.533 ** 1995 0.087 2.695 ** 1996 0.015 0.469 ** 1997 and 1998 (a) Plant Characteristics: Size ln(assets) 0.125 16.775 ** large (vs. small/micro) 0.256 5.240 ** 0.236 4.777 ** medium (vs. small/micro) 0.088 2.045 ** 0.085 1.954 * Plant Characteristics: Ownership joint venture (vs. Mex.) 0.161 2.879 ** foreign (vs. Mex.) 0.058 1.779 * jv or foreign (vs. Mex.) -0.100 -2.285 ** -0.104 -2.397 ** Other Plant Characteristics TFP 0.094 7.440 ** 0.073 3.819 ** 0.060 3.367 ** iso9000 certified 1994 -0.087 -1.121 iso9000 initiated 1994 -0.078 -1.339 iso9000 cert. or initiated 0.045 1.265 -0.068 -1.314 quality mgmt 1994 0.037 0.969 in-firm training 1994 0.094 2.429 ** 0.054 1.030 workplaces processes 1994 0.102 2.311 ** 0.085 1.859 * labor qualifications 1994 0.001 2.233 ** 0.001 1.289 research & devel. 1994 0.000 0.155 Industry Dummy Variables 32 Textiles 0.433 10.845 ** 0.049 0.661 0.053 0.732 33 Wood 0.382 5.810 ** 0.378 3.717 ** 0.380 3.762 ** 34 Paper -0.021 -0.408 -0.071 -0.710 -0.080 -0.797 35 Chemicals 0.345 8.969 ** 0.062 1.019 0.068 1.140 36 Non-met. Minerals 0.086 1.573 -0.327 -1.878 * -0.319 -1.845 * 37 Basic metals 0.399 5.072 ** -0.186 -1.086 -0.158 -0.930 38 Machinery 0.477 12.574 ** -0.214 -1.664 -0.199 -1.559 39 Other mfg. 0.565 5.214 ** -0.573 -3.068 ** -0.550 -2.964** Hypothesis tests (chi-sq) qualitycoeffs=0 31.82 ** 16.29 ** 15.84** Log likelihood -7371 ** -3237 ** -3299 ** No. observations 22417 22178 22495 * = significant at 5% level; ** = significant at 1% level. (a) variable dropped due to multicollinearity. 44 Table 3.6: Predicted Probability of Exporting by Size (Actual Proportions in Parentheses) 1994 1995 1996 1997 Large .26 (.45) .53 (.53) .71 (.58) .76 (.63) Medium .09 (.29) .24 (.39) .37 (.44) .44 (.47) Small/Micro .03 (.16) .06 (.22) .09 (.25) .13 (.26) 3.24 The influence of foreign ownership on the probability of exporting is less clear: it enters positively in Model 2 but negatively in Model 3. Other variations of the model suggested that the ownership results are not robust. This may suggest that it is the larger size or industry characteristics of joint ventures and foreign firms, not their foreign participation per se, that determines their greater probability of exporting. 3.25 The three case studies presented at the beginning of this chapter suggest that the probability of exporting increases with investments in training and quality control. We were able to test this by linking the EIA panel with the ENESTYC survey for 1994. For the subsample of plants that were included in both the EIA and ENESTYC in that year, we have information on IS09000 (whether or not the plant was IS09000 certified in that year, or was in the process of initiating IS09000 certification), the presence or absence of a quality management system, investments in research and development, the use of high- perfornance workplace practices, the proportion of qualified labor, and the presence or absence of worker training.33 Across several variations of the model, the variables found to be most consistently significant were the adoption of high-performance workplace practices, labor qualifications, and training. Consistently less significant were R&D and quality management systems. In separate regressions for each size group of plants (not shown in Table 3.5), the effect of training was found to be highest for micro- and small- scale plants. 3.26 We also investigated the relationship between plant-level productivity and the probability of exporting by using estimates of Total Factor Productivity derived as residuals in a production function model.34 TFP is consistently found to be positively associated with the probability of exporting. In separate regressions for each size category (not shown in the table) we found this relationship particularly strong for medium-scale and micro/small-scale plants. The relationship between exporting and TFP will be explored in detail in the next two chapters. 3.27 The automobile industry accounts for a large share of manufactured exports, but taking this industry out of the sample does not seem to alter the regression results very 33 Note that there is only one year of ENESTYC data, so these values were applied to these firms for all years they appear in the EIA panel. 34 Chapter 5 contains a detailed description of how these TFPs were estimated. 45 much (Model 3). The determinants of export status seem to be similar in the auto industry and the remainder of the manufacturing sector. 2. Export Entry 3.28 Instead of looking at current export status independently from prior export status, we can investigate the decision to enter export markets from non-exporter status. These estimations consider only the subset of plants that did not export in year (t-1). The purpose is to identify the factors that determine a move from the domestic market to the export market. The model is: (E)it = c + EZaj(Pj)t + Epk(Sk)it + 'it k where: Eit = 0 if Xit l = 0 and X1t = 0 (do not enter: remain a non-exporter) = 1 if Xit.I = 0 and Xit = 1 (enter: move from non-exporter status to exporter status) Pt = set of j market variables in year t sit = set of k plant characteristics for plant i in year t 3.29 The plant characteristics include size, foreign/domestic ownership, and high- performance workplace practices. Market variables include the real exchange rate, real interest rates and real interest rate variability, real domestic GDP and real U.S. GDP. Finally, we include financial variables (defined in the table below) to test the relationship between the plant's profitability and pre-export modernization, and its entry into export markets.35 Profitability is measured as cash flow-based profitability as well as inventory turnover. Pre-export modernization is measured by the new investment, and the import share of fixed assets as well as materials.36 We also experiment with lagged values some of the independent variables, and disaggregate the exchange rate to an industry-specific ratio of the change in an export price index relative to a change in a domestic price index.37 35 This relationship is analyzed graphically in the next chapter. 36 The construction of these variables is described in more detail in Chapter 4. 37 Export price indices were available at the 2-digit level, and producer price indices at the 4-digit level. 46 Financial Variables Ratio Numerator Denominator Profitability before extraordinary Ordinary income Sales (or non-operating income if items and income tax sales=O) Inventory turnover Cost of sales Average inventories during the period Share of imports in total fixed assets Total imported fixed assets Total fixed assets acquired during purchased during the period the period Share of imported primary materials Imported materials Total materials purchased in total purchased New investment Investment in fixed assets Total fixed assets at the beginning of the period 3.30 The regression results are summarized in Table 3.7, excluding the industry and regional variables for simplicity. We find again that size matters: the larger the plant, the more likely it is to move from non-exporter to exporter status. Firms with some foreign ownership are more likely to enter exporting than 1 00-percent Mexican firns, and total factor productivity is also positively associated with entry. Of the training and quality variables, only some are important, namely IS09000 and high-performance workplace practices. 3.31 The influence of macroeconomic conditions is difficult to interpret, since the results change from model to model. In most models tested, the level of U.S. GDP was a positive and significant determinant of export entry, as expected, but the direction of influence of national GDP depended upon whether national-level or state-level GDP was used, and domestic interest rates were not significant. The exchange rate entered positively only when financial variables were added to the model. When the automobile industry was excluded (not shown here), the results were very similar. 3.32 If there are "export spillovers" from foreign firms to domestic firms, the probability of entry into export markets would be increased with greater foreign presence in the plant's industry and/or region. We tested this by including a spillover variable, the share of four-digit industry exports accounted for by maquiladoras and non-maquiladora firms with some foreign ownership. No spillover effects were found, consistent with the results of this type of analysis on productivity growth (Chapter 5). 3.33 When financial and modernization variables are included, we find that although current-year productivity (TFP) is positively related to entry, profitability is not. The share of assets that are imported is positively related to export entry (and perhaps to productivity as well). This may reflect modernization investments undertaken in anticipation of exporting. In addition, the probability of entry increases when inventory turnover in the previous year is low -- perhaps reflecting poor demand conditions in the domestic market. In some other variations of the model, investment in new assets also seems to drive entry into export markets - either as a cause or as an action taken in anticipation of entry. 47 Table 3.7: Probit Models of Entry Decision Dependent variable: enter (= 0 if remain non-exporter, = 1 if move from non-exporter to exporter) Model 1 Model 2 Model 3 (using national-level (using state-level (using state-level GDP) GDP and incl. GDP and incl. spillover variables) Financial variables) Explanatory Variables Coeff. Z-Stat. Coeff. Z-Stat. Coeff. Z-Stat. Constant term -8.083 2.435 ** -3.593 -4.290 ** -5.077 -4.494 Market Conditions Exchange rate 7.539 0.980 7.640 0.993 25.430 2.338 Interest rate 0.089 1.393 -0.001 -0.054 0.042 0.917 Interest rate variability 0.079 3.330 ** 0.027 3.682 ** 0.018 1.866 GDP (a) 0.004 2.755 ** -0.000 -1.708 * -0.002 -1.481 USGDP .000 2.756 ** 0.000 3.242 Plant Characteristics Medium (vs. large) -0.218 -5.361 ** -0.219 -5.386 ** -0.249 -4.903 Small/micro (vs. large) -0.480 -11.164 ** -0.480 -11.154 ** -0.445 -8.791 jv or foreign (vs. Mex.) 0.120 3.169 ** 0.118 3.110 ** 0.211 3.876 Other Plant Characteristics TFP 0.068 4.400 ** 0.071 4.549 ** 0.063 1.979 TFP (t-1) 0.001 0.053 Iso9000 cert/init. 1994 0.078 1.719 ** 0.077 1.697 * in-firm training 1994 0.008 0.181 0.006 0.152 Workplace processes 0.107 2.778 ** 0.107 2.760 ** 1994 Labor qualifications 1994 .001 1.081 0.001 1.098 Spillovers Maquilalforeign X share -0.015 -0.239 Financial Variables Profitability 0.001 0.385 Profitability (t-1) 0.006 0.999 Inventory turnover (t-1) -0.005 -3.693 New investment (t- 1) -0.009 -0.615 Imported asset share (t- 1) 0.231 3.493 Internal financing share 0.000 0.200 Int. financing share (t-1) -0.000 -1.385 Log likelihood -4846 -4844 -2689 No. observations 15121 15121 7147 * = significant at 5% level; ** = significant at 1% level. (a) Models 2 and 3 use real GDP at the state level. Industry dummies were included in all models, but the results are not shown here. 48 3. Export Volume 3.34 Finally, we investigate the determinants of the quantity of firm-level exports. The model is: (XQ)it = c + y(XQ)jt, + E+ ( (S)i i i k where: XQit = quantity of plant i exports in year t Pt = set of j market variables in year t Sit = set of k plant characteristics for plant i in year t 3.35 Export volume is measured as the value of exports in millions of pesos divided by an export price index at the two-digit level. The export volume model uses the same market variables and plants characteristics as the export status and export entry regressions. A Tobit model is used, since the observations on the dependent variable are left-censored at zero. The results of the model are shown in Table 3.8. 3.36 The results are similar to those in the export status and export entry regressions. Obviously, plant size is important -- larger changes in export volume are positively associated with larger scale plants -- and other plant characteristics, such as total factor productivity and in-firm training, are important determinants of export volume. With respect to macroeconomic conditions, the real exchange rate and U.S. GDP are also important and in the expected direction. In other versions of the model (not shown here), the share of internal financing was negatively associated with export volume, again consistent with our graphical analysis. 49 Table 3.8: Tobit Models of Export Volume Dependent variable: XQ (quantity of exports of plant i) Explanatory Variables Coeff. t-Stat. Constant term -5402.783 -8.028 ** XQ (t-l) 1.139 59.035 ** Market Conditions Exchange rate 18397.640 2.992 ** Interest rate 41.865 1.617 GDP -0.767 -1.525 USGDP 0.502 5.977 * Plant Characteristics Large (vs. small/micro) 698.437 21.059 ** Medium (vs. small/micro) 391.158 12.927 ** jv or foreign (vs. Mex.) 259.403 9.230 * Other Plant Characteristics TFP 157.241 12.019 ** in-firm training 1994 80.486 2.232 ** Workplace processes 1994 29.083 0.965 Labor qualifications 1994 0.494 1.227 Log likelihood -75127 No. observations 22469 * = significant at 5% level; ** = significant at 1% level. Industry dummy variables were included in all models, but the results are not shown here. D. Conclusions 3.37 The case studies and qualitative surveys reviewed in this chapter suggest that firms are attracted to exporting because of the strength of foreign demand relative to demand in domestic markets, particularly during recessionary periods at home. Many are prepared to and plan to switch back to the domestic market when domestic markets recover. Nevertheless, our regression results support the hypothesis that there is some "export permanence", i.e., that current export status is influenced by previous export experience, perhaps because of the sunk costs of exporting. 3.38 The case studies and qualitative surveys also suggest that there is a period of modernization that precedes or accompanies entry into export markets, particularly with respect to investment in IS09000 certification and quality control systems, and to some extent with respect to worker training. We were able to test some of these hypotheses using the 1994 ENESTYC survey linked to the EIA panel. The regression results suggest 50 that high-performance workplace practices and labor skills are associated with exporting and initial export entry, and anecdotal evidence suggests that these are made consciously in anticipation of stiffer competition in external markets. The association between exporting and IS09000 and other quality management systems is less apparent, through the regression results may be influenced by collinearity among independent variables. New investment, in particular purchases of imported assets, are associated with entry into exporting. 3.39 There is a clear association between total factor productivity and exporting, as well as entry into export markets -- even controlling for plant size, ownership, and other plant characteristics. In the next chapter we examine the causality between exporting and productivity growth, to see if causality runs from productivity to exporting (i.e., more efficient firms tend to export) or from exporting to productivity (i.e., through learning-by- exporting effects). However, even if causality is found to run from productivity to exporting, it may be that the anticipation of exporting is still driving pre-export efficiency improvements. 3.40 Regressions to explain export status and the probability of entering export markets consistently show the influence of plant size and ownership. Controlling for macroeconomic conditions and some plant characteristics, larger plants and plants with some degree of foreign ownership are more likely to export than Mexican plants. 3.41 Enterprise surveys suggest that financing plays a key role, both in the firm's expansion and in its ability to export. Expansion and exporting may actually be constrained by a lack of availability of credit, or firms may choose to limit their growth to that which can be financed with internal resources, aware as they are of the potential for macroeconomic shocks. In either case, enterprise growth and export competitiveness tend to be limited by the firm's internal resources. For larger firms (particularly those with access to foreign financing), the cost of credit is seen as a greater constraint to exporting than the availability of credit. We have not been able to carefully test these hypotheses due to the lack of credit information in the EIA, but we do find that entry into exporting is associated with a higher import content of assets. 51 4. EXPORTING AND PERFORMANCE: CAUSE OR EFFECT? 4.1 Export promotion programs are often justified on the basis of the belief that exporting leads to improvements in technical efficiency. "Learning-by-exporting" is said to occur because exporters benefit from the technical expertise of their buyers, become aware of quality standards in foreign markets, and discover ways to reduce costs. This hypothesis seems to be supported by the observed efficiency gap between exporters and non-exporters. However, the observed association between exporting and efficiency may be due to causality in the other direction, as more efficient firms tend to be the ones that break into export markets. If the efficiency gap between exporters and non-exporters is due to self-selection rather than to learning-by-exporting, there is less justification for export promotion. 4.2 Previous empirical studies using micro data suggest that there is little evidence that causality runs from exporting experience to improvements in performance. A recent study for Colombia, Mexico, and Morocco38 found that plants that begin exporting tend to have relatively low average variable cost, and plants that cease exporting are becoming increasingly high-cost; similar patterns are evident when labor productivity is used as the performance measure. However, cost and productivity trajectories usually did not continue to change after plants began exporting. These patterns are similar to those found in the U.S. manufacturing sector.39 4.3 For some firms and during some phases of the business cycle, exporting may even follow a downturn in enterprise performance - for example, if the decision to export is a response to a downturn in a firm's domestic market. As discussed in the previous chapter, interviews with firms during the crisis suggested that many turned to exports as a "safety valve" when the 1995 recession caused a decline in domestic demand for their products. If that is the case, one would expect to see exporting preceded by a downturn in profitability and inventory turnover, that then recover after the firm begins to export. In addition, pre-export performance may differ from post-export performance as a result of enterprise restructuring and new investment undertaken to become more competitive abroad. 4.4 Finally, the entry into export markets may follow a period of modernization or restructuring by the plant, undertaken voluntarily to be able to compete abroad, or forced because of a downturn in performance. This behavior would be reflected in a pattern of increased investment and possibly greater use of imported machinery and equipment prior to exporting. 4.5 In this chapter we investigate the dynamic relationship between exporting and enterprise performance, using our panel database to track individual plants over time. 38 Clerides, Lach, and Tybout (1996). Estimates for Mexico in this study used an INEGI EIA panel database for the years 1984-90. 39 Bernard and Jensen (1995). 52 The relationship is analyzed graphically by comparing plants that begin to export, or stop exporting, with control groups of other plants. We look at a broad range of indicators -- labor productivity, total factor productivity, profitability, inventory turnover, new investment, and the import share of new assets and material inputs -- to investigate what seems to be driving changes in the export status of plants and what happens to them afterwards. The analysis sheds light on the direction of causality between exporting and improved efficiency. A. EIA 4.6 Using the 1993-98 EIA panel database, we define the following groups of plants: Non-exporters Plants that never export during the 1993-98 sample period Permanent Exporters Plants that export in all years 1993-98 Entrants Plants that move from non-exporter to exporter status once during 1993-98 Quitters Plants that move from exporter to non-exporter status once during 1993-98 Switchers Plants that enter or quit more than once during 1993-98 4.7 The graphs below compare average performance variables for two study groups (Entrants and Quitters) with weighted averages for three control groups: Non-exporters, Permanent Exporters, and Switchers. In some cases, we also look at Entrants and Quitters by cohort40 in order to understand whether generalization across cohorts is valid. To put all members of a study group into one time period corresponding to the occurrence of the "event" (beginning to export or ceasing to export), the year of transition is normalized to zero. For example,. if a plant starts to export in 1996, the normalized time variable for that plant takes on a value of 0 for year 1996. The years preceding the event (1993, 1994, 1995) take on time values (-3, -2, -1) and the years following the event (1997, 1998) take on time values (+1, +2). To construct control groups of Non- exporters, Permanent Exporters, and Switchers for purposes of comparison with the study group, we applied a weighting scheme to plants in the control groups that took into account the normalized and actual time values of the members of the study group.4' This means that there is no unique path of the performance variables for the control groups, since the weighting scheme depends on the study group in question. 40 A cohort is a set of firms that changed their status (entered or quit) in a particular year (e.g., the 1994 Entrants cohort, the 1995 Entrants cohort, etc.). 41 The weighting procedure works as follows: (a) averages are found for each of the three control groups in each year; (b) a weight is associated with each year k, Wkm, 1994-1998, in proportion to the number of plants in the study group with the relative time of m in this year41; (c) these weights are applied to obtain the final averages for each of the control groups at the relative time m (m: -5.. .-1,O, 1,.. .4). 53 1. Productivity 4.8 The productivity of a single production input (e.g., labor) is measured simply as the value of production, or total sales, per unit of that input (e.g., labor hours). This measure of productivity captures the contribution of other inputs (e.g., capital) to output, and thus is affected by the importance of those other factors in production. In contrast, the growth of Total Factor Productivity (TFP) is a measure of technical progress in determining the growth of output, after subtracting growth attributable to changes in the quantities of inputs used. Below we look at time paths for both measures of productivity. Productivity Measure How Calculated Labor Productivity Real value of production and other non- extraordinary income divided by the number of employee hours worked, adjusting for industry and size differences. Total Factor Productivity Residual from estimated regression of value added on labor and capital inputs, adjusting for industry differences. 4.9 We begin by estimating total factor productivity as a residual in an econometric production function model, in which the dependent variable is real value added and the independent variables are the real capital stock (fixed assets) and labor hours, all in natural logarithms. (A detailed description of how TFP was estimated is contained in Chapter 5.) We experimented with several variations: a pooled regression (all industries combined) with industry dummy variables capturing inter-industry differences; separate industry-by-industry regressions to allow production function parameters to vary by industry, with the residuals subsequently pooled; and a method of excluding outlier values of variables.42 The results of pooled versus industry-by-industry regressions were fairly consistent, and the estimates were found to be robust to outliers. To make the presentation simpler, below we show only the results using TFP estimates obtained from industry-by-industry regressions including outliers. 4.10 Based on our hypothesis that there is an efficiency gap between exporters and non-exporters, we would expect to see the average TFP of Permanent Exporters higher than that of Non-exporters (two of the three control groups). Moreover, if exporting causes increases in productivity through learning-by-exporting, we would expect to see the average TFP of Entrants increase after they enter (at normalized time t=0) relative to other firms. If causality runs the other way - from improved performance to exporting - we would expect to see Entrants' TFP rising before t=0, or at least see that it is higher than that of the control groups before t=0. 42An observation-specific index was created that showed the influence of each observation on the resulting coefficients in the regressions (DFITS, see Bollen and Jackman (1990) or p.190 of STATA6.0 reference). The standard cutoff of 2(k/n)'2 was applied, where k is the number of explanatory variables, including constant term, and n is the number of observations used in the regression. Observations that had the DFITS index higher then the cutoff were censored out, and regressions were re-run without these outliers. 54 4.11 The graphical analysis confirms the expected efficiency gap: the average TFP of Permanent Exporters always exceeds that of Non-exporters (Figure 4.1). The TFP path of Entrants suggests that the causality between exporting and TFP runs in both directions. In the second and third years following entry (t=2 and t=3), the TFP of Entrants rises. This suggests that there are learning effects, but that they occur after a lag. The learning effects are seen more clearly in the 1994 cohort than in later cohorts, since we are able to trace the performance of the 1994 cohort for several years after entry. Figure 4.1: Total Factor Productivity of Entrants TFP .41 |Permanent exporters .24 .2 - ___, ~~~~~~~ | EntrantsJ I A ~~~No-XpSwitcers -.2 -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time 4.12 Similar patterns are observed when we look at labor productivity instead of TFP (Figure 4.2). The labor productivity of Permanent Exporters is always higher than that of others. In the years prior to beginning to export, the labor productivity of Entrants increases. There is some further increase after entry, but labor productivity falls in the fourth year after entry. The fourth year in Figure 4.2, however, represents only the 1995 cohort of entrants. Thus, there is some indication of bidirectional causality when labor productivity is considered, but the learning-by-exporting effect is not strong. 55 Figure 4.2: Labor Productivity of Entrants Labor productivity (residuals) .4 rmanent exporters .2 0 -.2- f -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time 4.13 The TFP pattern of Quitters displays a tendency for their TFP to go down right after quitting (Figure 4.3), with no signs of a pre-exit decline. Quitters' pattern of labor productivity shows a clear downward trend in the three years prior to exit, and continues to deteriorate (Figure 4.4). In other words, the decision to stop exporting follows a steady downturn in performance, and the move back to the domestic market does not seem to improve performance. Figure 4.3: Total Factor Productivity of Quitters TFP .4 Permanexpor .2 Quitters [witchers 0 -.2 -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time 56 Figure 4.4: Labor Productivity of Quitters Labor productivity (residuals) .4] P nexporter .2 0 -.2 -Non-exlorQters| -.2 l l l ~~~ ~~ ~ ~ ~ ~ ~~~~~I . l l _5 -4 -3 -2 -1 o 1 2 3 4 Normalized time 2. Financial Performance 4.14 Parallel with the observed gap in efficiency between exporters and non-exporters, we would expect to see that Permnanent Exporters have better financial performance than Non-exporters. In addition, if the better-performing firms self-select into exporting, we would expect the financial performance of Entrants to be better than that of Non- exporters in the years before entering. On the other hand, if firms decide to try exporting when the domestic market weakens, a decline in financial performance before exporting could be observed. After starting to export, the financial performance of these firms should improve. The graphs show the patterns of the following financial performance indicators:43 Financial Ratio Numerator Denominator Cash-based profitability ordinary income + Sales (or non-operating income depreciation cost + if sales=O) change in inventories Inventory turnover Cost of sales Average inventories during the period 4.15 The differences in profitability between groups are small in most years (Figures 4.5 and 4.6), with the exception of Permanent Exporters who are noticeably less profitable. A possible explanation would be that Permanent Exporters are more mature and operate in more competitive markets than the other groups. Both Entry and Quitting lead to an increase in profitability after the event. The observed decline in profitability of Entrants in the years preceding entry may reflect restructuring prior to the launch of 43 We also experimented with another measure of profitability -- profitability before extraordinary items and income tax. The results were roughly similar to those using cash-based profitability. 57 export sales. The rise in profitability of Entrants observed before entry is due to the 1996 cohort, and thus is not representative. Figure 4.5: Profitability of Entrants Cash-based profitability (residuals) .4 .2 0 -.2 -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time Figure 4.6: Profitability of Quitters Cash-based profitability (residuals) 1+X j | ~~~~~~~~Quitters .5 0+ Pemaet exportrs -.5- -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time 4.16 Inventory turnover measures the value of sales per unit of inventory stock; a higher value can be interpreted as an improvement in financial performance. In all normalized time periods, Permanent Exporters have lowest inventory turnover (Figure 4.7). A possible explanation is that Permanent Exporters have a relatively high share of imported inputs, which are costly to re-stock. The behavior of Entrants suggests that they are accumulating inventories in the years before exporting; turnover improves in the year of entry and some consequent two years. Unlike the case of profitability, this result is consistent across cohorts. It is possible that the decision to export is driven by the 58 inventory problem: firms are searching for a new market to sell inventories that have accumulated as a result of a decline in domestic demand. Finally, Quitters show a similar pattern: an accumulation of inventories before leaving export markets, with a subsequent improvement in performance (Figure 4.8). Figure 4.7: Inventory Turnover of Entrants Inventory turnover (residuals) .4 .2 = - 2 | Switchers |~~~Pemanntex * ~~~~-- - r _ - -v -- r- H-- r -5 4 -3 -2 -1 0 1 2 3 4 Normalized time Figure 4.8: Inventory Turnover of Quitters Inventory tumover (residuals) Non-exporters .2 Switchers| -.2 -.4 T--- T r - P - -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time 3. Investment and Imported Inputs 4.17 For the same type of output, exporters may use a larger share of imported inputs than non-exporters because the former receive duty-free access to imported inputs via the PITEX, ALTEX, or temporary import regime. In some cases, the product specifications and quality required in foreign markets may lead exporters to have a higher import 59 content than non-exporters. In the years prior to entry into exporting, the share of imported inputs might be expected to rise, as firms change product design and quality in anticipation of competing abroad or as they modernize their production. Similarly, one would expect to see a rise in investment in the years prior to exporting if plants are modernizing their operations in anticipation of entry. Investment and Imported Numerator Denominator Inputs Share of imports in total fixed Total imported fixed assets Total fixed assets acquired assets purchased during the period during the period Share of imported primary Imported materials Total materials purchased materials in total purchased New investment Investment in fixed assets Total fixed assets at the beginning of the period, at cost of acquisition 4.18 Permanent Exporters tend to use the highest share of imported inputs (Figures 4.9 and 4. 10). There is a strong upward trend in import content for Entrants that starts before these plants begin to export, and the share of new assets even exceeds that for Permanent Exporters some 2-3 years after entry (Figures 4.11 and 4.12). Entrants also show a jump in new investment in the years before starting to export (Figure 4.13). This may indicate that Entrants are modernizing in anticipation of exporting. There is no significant trend for Quitters versus control groups, but as expected their new investment is the lowest of all groups (Figure 4.14). Figure 4.9: Imported Material Inputs of Entrants Share of imported materials in total used (residuals) .1- | elann xpon rters 0 -.05 | Switers - 05 - / | ~~~~~Non-exportersj -5 -4 -3 -2 -1 0 1 2 3 4 timeO 60 Figure 4.10: Imported Assets of Entrants Share of imports in new fixed assets (residuals) .05 o- |Permaneritexporters| 0 O~~~~~~~~~~~~~~~ '' / | ~~~~~~~~~Non-exporters| -.05 - l l l l l l l l l l -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time Figure 4.11: Imported Material Inputs of Quitters Share of imported materials in total used (residuals) .1X 05 -| -.05 I - . -5 -4 -3 -2 -1 0 4 Normalized time 61 Figure 4.12: Imported Assets of Quitters Share of imports In new fixed assets (residuals) .05 |Prrnnt exporters -.05 -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time Figure 4.13: New Investment of Entrants New investments relative to existing fixed assets (residuals) .4 .2 -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time 62 Figure 4.14: New Investment of Quitters New investment relative to existing fixed assets (residuals) l 0wthr -.5 -5 -4 -3 -2 -1 0 1 2 3 4 Normalized time B. Maquiladoras 4.19 By definition, all maquiladora plants are exporters, so there are no Non-Exporters or Switchers in the sample. We define the following groups of plants in the CIM panel database: Permanent Exporters Plants that are present for all years 1990-98 Entrants Plants that come into the sample some time during 1990-98 4.20 For Entrants, the year of entry is normalized to t=0; years subsequent to entry are given normalized time values (+1, +2, +3, etc.). Since we do not observe Entrants in the sample before they begin exporting, there are no negative values of the normalized time variable. Note that higher values of time (+7, +8, +9) would be observed only for plants that became maquiladoras early in the 1990-98 period. The graphs below compare Entrants in the year of entry and years after entry with Permanent Exporters in those same years. 4.21 As with the non-maquiladora plants, estimates of total factor productivity are derived as the residuals from production function regressions of value added on capital stock and labor.44 Since the CIM does not contain information on fixed assets, we used 44 We experimented with two definitions of value added: Method A: Value added = total sales - non-labor variable input costs Method B: Value added = profits + labor costs We also tried two types of regressions to generate TFP residuals: a pooled regression (all industries combined) along with industry dummy variables; and industry-by-industry regressions with subsequent pooling of residuals. The results were similar across Method A and B, and across pooled and industry-by- industry regressions. Here we report only the results from Method A and pooled regressions. 63 the amount of electric energy consumed as a proxy for capital stock in the production function. 4.22 Permanent maquiladoras have higher productivity than Entrants in all but one normalized time period (Figure 4.15). Only plants that entered in 1991 and stayed until 1998 improved their TFP beyond the level of Permanent maquiladoras. A similar pattern is observed when comparing labor productivity instead of TFP (Figure 4.16). The productivity of Entrants declines in the three years after entry, and then rises. The increase after three years could be due to learning, or alternatively to self selection as low-productivity maquiladoras exit the sample.4P The contrast with our findings using the EIA panel is interesting: non-maquiladora plants show increases in productivity immediately after beginning to export, whereas maquiladoras do not show productivity increases (or "learning" effects) in the several years following entry. Yet, the productivity of maquiladoras does seem to improve as they become more permanent - which may be a factor explaining their survival. Figure 4.15: Total Factor Productivity of Maquiladoras TFP .05 0 -.05 0 1 2 3 4 5 6 7 8 9 Normalized time 4.23 Other performance measures for maquiladoras are also shown in Figure 4.16. Looking at a measure of cost efficiency (average variable cost), we find that, even though their productivity is lower, Entrants are more cost-efficient than Permanent 46 maquiladoras. Entrants show increases in cost efficiency in the first two to three years following entry, but costs then rise. Further breakdowns by cohort suggest that Entrants 45 To look more closely at Entrants and a comparable group of Permanent maquiladoras, the sample was split into cohorts: those that started in 1991, those that started in 1992, and so on. While there was much diversity across cohorts, in general Entrants had lower productivity than Permanent maquiladoras. " When examined more closely in by-cohort breakdowns, it appears that it is mostly the post-1994 Entrants that have lower costs during the first few years after entry, and mostly the 1991 entrants that stay until 1998 that have lower costs in normalized time values t=4 through t=7. 64 are more cost-effective than Permanent maquiladoras mainly in the post- 1995 period. We also find that Entrants tend to use relatively more domestically-purchased material inputs than Permanent maquiladoras. In most cases, this situation persists for some time following entry, which may indicate a shift toward greater use of domestic inputs by maquiladoras in the 1990s. Regression analysis (not shown here) suggests that higher shares of imported inputs are associated with higher productivity. This subject is one that requires further investigation. Figure 4.16: Other Performance Variables for Maquiladoras .1 Labor Productivity 02 Average variable cost 0 L~ ~iII.01- -A Entrants 0 -.2 ____________________________-.01 ___________________________ 0 1 2 3 I * 7 8 9 0 1 3' 4' 5 ' 7 8 9 tImeO tlrmeO .04 - .05 I____ .02- Imported materials/ Total materials 0 Imported materials Sales -.02 -.04 ~~~~~~~~~~~~~~~~~~~~~~~-.05 . 01 11 21 3' 41 51 Itl 7{ 81 91 0 1' 2' 3' 4 ' 7 8 9 tm.O timeO C. Conclusions 4.24 To summarize, the graphical analysis suggests the following conclusions regarding the relationship between exporting and plant performance: * Non-exporters have lower productivity than exporters, but their profitability is nearly the same. Inventory turnover tends to be higher for non-exporters than for exporters, which may be explained by higher costs of re-stocking for exporters. * The productivity of non-maquiladora plants that start exporting usually grows substantially in the two to three years prior to entering the export market. Subsequently, their productivity continues to grow relative to that of permanent exporters and non-exporters for at least two to three years, which suggests that 65 learning-by-exporting effects do exist. This finding contrasts with earlier work by Roberts and Tybout (1996) that fails to find learning-by-exporting effects. * In contrast to non-maquiladoras, newly entering maquiladoras have lower productivity than maquiladoras who are already in operation. Their productivity improves as they become more permanent. This could reflect "learning" on the part of new maquiladoras, or a process of selection as inefficient maquiladoras exit. * One or two years prior to entering the export market, non-maquiladora plants seem to experience a downturn in their domestic operations and a possible modernization or restructuring in anticipation of entering export markets. Their profitability falls, new investment increases, and purchases of foreign machinery and equipment intensify. The downturn in domestic operations may also be seen in the slowdown in the inventory turnover that improves when plants start exporting. * As is the case with entrants, quitters from exporting seem to experience a downturn in their operations in the several periods before they cease exporting. This is revealed mainly in the rapid inventory accumulation prior to quitting, rather than in a decline in profitability. Re-orientation to the domestic market is associated with two polar effects: substantial, but short-term, gains in terms of profitability (perhaps due to the resolution of the inventory problem), and a gradual, longer-term decline in productivity. - New maquiladoras appear to be more cost-efficient than existing maquiladoras in the post-crisis period. They are also beginning to use a greater share of locally-purchased inputs. 66 5. EXPORTING, SUPPLIER DEVELOPMENT, AND PRODUCTIVITY A. Overview 5.1 In the previous chapter, we investigated graphically the relationship between exporting and enterprise performance by comparing exporting firms, before and after starting to export, with control groups of other firrns. That analysis, which was intended to examine the direction of causality between exporting and enterprise performance, provided evidence for causality running both ways. That is, firms that begin exporting tend initially to be already more productive than firms that do not export; however, the event of exporting in itself is associated with subsequent improved performance. In this chapter, we pursue this initial finding more formally within a regression framework. 5.2 We use a production function framework to measure the impact of exporting, as well as other factors including inter-firm linkages, on total factor productivity (TFP) growth. We do this using the 1993-1997 EIA panel and the EIA panel linked to the 1995 ENESTYC survey. This production function approach relates the plant's value-added to its inputs of capital and labor, with dummy indicator variables for two-digit industries and year dummy variables to control for macro-economic shocks affecting all plants in a given year. This allows us to compute a plant-specific time-varying residual, purged of shocks, which we interpret as TFP or technological progress. TFP is then related to a variety of productivity determinants, including exporting. Two export measures are used: whether a plant currently exports, and its experience exporting. We are interested in whether exporting leads immediately to a change in productivity levels, or whether productivity gains are acquired gradually over time through learning. 5.3 We also control for other factors to better isolate, and compare, the productivity impacts of exporting to the effects of variables such as supplier linkages, foreign ownership, and high-performance workplace practices such as training and quality control. These potential productivity factors are of interest in themselves. For example, smaller firms may benefit from the rising export orientation of the economy as indirect exporters by doing subcontract work for large exporters; alternatively, supplier linkages may be a route for technology transfer and assistance from buyers-large firms and multi-national companies-to smaller, Mexican-owned suppliers. The 1995 ENESTYC survey asked about worker training, quality control systems, and ISO-9000 certification and we use this information to estimate their contribution to TFP growth. B. Measuring TFP Growth 5.4 TFP is estimated as the residual in an econometric production function model, rather than by growth accounting. We use two variants of the production function: one is a common production function estimated for all sectors combined, with inter-industry differences in technology being captured by industry dummy variables; the second is a production function estimated separately by industrial sector, which allows production 67 function parameters to vary by industry. The analysis in this chapter relies primarily on TFP estimated from the first model specification. 5.5 The specification of the Cobb-Douglas production function, in natural logarithms, may be written as follows: ln(VA)i, = [30 + ,B,n(K)ijt + 021n(L) j, + P 3 E (IND) i + | 4 E (YEAR), + E; j, j-1 t-l where VA is value-added, K is capital assets, L refers to labor inputs (in hours), IND refers to dumnmy variables for j-l industries and YEAR to t-l years, for the individual plant i in time t. The error termn it is the normal regression error with mean zero that is, by assumption, uncorrelated with the dependent and explanatory variables. 5.6 The variables used in the production function model are defined as follows: Dependent variable: Log(value added, in real 1997 N$)47 Independent variables: Log(fixed assets, not deflated)48 Log(total hours worked in the year) 8 dummy variables for industrial sector (32,33,...39) 4 year dummy variables (1994, 1995,... 1997) The dummy variables for industrial sector 31 (food and beverages) and year 1993 are omitted to avoid perfect collinearity. Year dummy variables are included to control for macro factors, such as the early 1995 and 1997 shocks, which are assumed to affect all plants equally in a given year but that are not adequately captured by price indices used to deflate output. While the distribution of the residuals from this production function has a mean of zero for the pooled annual cross-sections, individual plants can have different time-trajectories of residuals, sit; the analysis that follows investigates the determinants of their TFP growth paths. C. Production Function and TFP Estimates 5.7 The results of the sector-wide production function, shown in Table 5.1, are consistent with a constant returns to scale production function since the estimated coefficients for capital (0.32) and labor (0.76) roughly sum to 1.49 These estimates are similar to those obtained for other countries, and are broadly consistent with capital and labor shares of income. 47 Value added measure uses the formula recommended by the UN Statistics Office for industrial statistics. See UN (1984) and updates. 48 In each EIA year, we use fixed assets reported at the beginning of that year. No deflator is available for fixed assets, and we assume that the flow of capital services is adequately reflected in its book value. 49 The models is estimated by ordinary least squares for 1993-1997, using the panel of 28,626 plant-year observations. 68 Table 5.1: Production Function Estimates Dependent Variable: Coefficient t-statistic Log(value-added) Constant 2.7861 80.63 Log(assets) 0.3254 75.22 Log(total hours) 0.7630 112.12 Industry dummy variables 32 Textiles, apparel -0.4442 -23.10 33 Wood products -0.6822 -21.82 34 Paper products -0.1664 -6.87 35 Chemicals 0.0961 5.24 36 Non-met. minerals -0.3790 -14.97 37 Basic metals 0.2016 5.19 38 Machinery & equip. -0.1162 -6.56 39 Other -0.3383 -6.29 Year dummy variables 1994 -0.0230 -1.34 1995 -0.2126 -12.12 1996 -0.2461 -13.87 1997 -0.3219 -17.86 R2 0.7080 Number of observations 28,626 Note: for pooled cross-sections and sectors, 1993-97. 5.8 After estimation, TFP for each plant-year observation is calculated from the residual obtained by subtracting the predicted value of log(value added) from its actual value. To aggregate plant-level TFP to the manufacturing sector as a whole, plant- specific TFP levels are weighted (in each year) by the plant's share of total value added in the manufacturing sector.50 Since the residuals are in logarithms, the change in the weighted log residuals from year t to year t+1 is an estimate of the annual rate of TFP growth from t to t+1. 5.9 Table 5.2 shows the resulting TFP growth rates for exporters and non-exporters, and for the manufacturing sector as a whole.51 By themselves, TFP levels are not very informative at this level of aggregation. TFP growth rates are more informative: for the manufacturing sector as a whole, TFP growth accelerated between 1993 and 1995-from an annualized growth rate of 0.6 percent to 13.8 percent. Subsequently, TFP growth rates declined, to 1.3 percent in 1995-96, and turned negative (-3.8 percent) in 1996-97. 50 Care is taken that only plants that have valid TFP measures are included in the calculation of the plant's value-added weight in total sectoral value added. 51 Note that the composition of plants in the exporting and non-exporting sub-samples changes over time as plants enter or exit from exporting. 69 Table 5.2: TFP Levels and TFP Growth Rates Year Overall Sample Non-Exporters Exporters TFP level 93 0.744 0.799 0.686 94 0.750 0.770 0.732 95 0.888 0.855 0.909 96 0.901 0.799 0.953 97 0.863 0.782 0.896 TFP growth % 93-94 0.6 -2.9 4.6 94-95 13.8 8.5 17.7 95-96 1.3 -5.6 4.4 96-97 -3.8 -1.7 -5.7 Source: author's calculations based on production function estimates 5.10 Non-exporters fare less well over this period as compared to exporters.52 While time trends in TFP growth mirror each other in these two sub-samples, non-exporters as a whole only experienced positive TFP growth rates in one out of the four sub-periods. In contrast, the group of exporting plants experienced positive TFP growth rates in three out of the four sub-periods, and TFP growth did not turn negative until the last sub-period, 1996-97. Thus, it appears that exporters in Mexico experienced more rapid TFP growth than establishments that did not export. D. Analyzing Firm-Level TFP Growth 5.11 In this section, we turn from broad aggregates to a plant-level analysis of the effects on TFP growth of exporting activities, inter-firm linkages, and correlates of high- performance workplaces such as worker training and quality control. As noted in the introduction, both exporting and inter-firm linkages are hypothesized to be important conduits for acquisition of market-relevant information and technology, and for firrn- level learning and development of technical capabilities. Here, we test these hypotheses in a regression framework by relating proxies of exporting and inter-firm linkages to the plant-level TFP estimated previously. We are particularly interested in the learning effects of exporting and inter-firm linkages in enhancing the growth of TFP. In addition, we have information on different attributes of high-performance workplaces-use of quality-circles, ISO-9000 certification, and worker training. However, this information is only available for one point in time for the sub-sample of EIA plants included in the 1995 ENESTYC survey, and we include them as control variables. 52 The results using TFP estimated by two-digit industry are broadly consistent with the finding of a TFP gap between exporting and non-exporting firms. There are two groups of industries by TFP growth performance. The first group-food, textiles and apparel, wood products and furniture, and clay and cement products, has lower TFP growth rates and tends to be domestic-oriented; the second group-paper and printing, chemicals, metals, machinery and other industries-has higher rates of TFP growth, and is generally more export-oriented. 70 5.12 We draw inferences about the causal effects of exporting and inter-firm linkages on TFP growth by exploiting the panel nature of the EIA data. There are problems in ascertaining causality from contemporaneous correlations derived from simple tables or regressions using cross-section data. As we noted earlier, a contemporaneous positive relationship between (say) exporting and TFP growth in a given year could simply reflect the fact that firms which export are a highly selected (non-representative) group, with high-levels of productivity that arise not from exports but from unobserved productivity traits. They are more productive to start with than firms that do not export, and that is why they export. We address this problem by exploiting the panel nature of the EIA data, and the fact that some plants are observed to begin (and continue) exporting over the 1993-1997 period. This allows us to estimate panel regression models relating current TFP levels to both current and past exporting histories, and thus to draw causal inferences from exporting experience to productivity growth. These exporting (sub-contracting) histories are also useful in providing insights into the presence of "learning effects", that is, whether productivity gains are derived inimediately or only slowly over time with accumulated past experience exporting (or subcontracting). 5.13 We capture the TFP effects of event histories using several variables-a (0,1) indicator variable to capture the contemporaneous (current period) productivity effects of that event, and a quadratic (time and time-squared) specification to capture the time- varying effects of accumulated years of experience with that event. The table below shows an example of how these variables are constructed for a plant that begins exporting in year t- I and stops in year t=4. Export Experience Variables: Example Time t 0 1 2 3 4 Exporting? (0,1) 0 1 1 1 1 Years of export experience 0 1 2 3 3 Export experience squared 0 1 4 9 9 5.14 In the TFP panel regressions using the full EIA data, we include these event histories for exporting, subcontracting-in (processing raw materials from contractors as suppliers), and subcontracting-out (outsourcing work to suppliers). Attributes of high- performance workplaces - quality circles, ISO-9000 certification, and worker training -- do not vary over time, and are included in regressions on the linked EIA-ENESTYC sub- sample of plants as control variables. They provide a check on the robustness of the exporting and subcontracting results to the omission of unmeasured (at least in the full EIA panel) productivity attributes of plants. 5.15 Before turning to the regression results, it is useful to review trends in exporting and subcontracting over this period. These are shown by plant size and for all plants combined in Figure 5.4. As noted in previous chapters, there was an increase in the proportion of plants that exported over the 1993-1997 period, from about 25 percent to 40 percent. The second graph indicates that this was accompanied by a rise in the share of firms that subcontracting-in -- the proportion of plants as suppliers rose from 15 percent 71 to 18 percent -- so supplier development appears to have taken place. Interestingly, this happened despite a slight decline in outsourcing -- the proportion of plants that subcontracted-out to other plants fell from 35 percent to 33 percent (the third graph). Figure 5.4 Exporters (% of plants) 0.6 -.-~~~~~~~~~~~~~large7 o 0.5~~~~~~~F_~ x 0.4 medium E) 0.3 2 X __ | E W W P : small .E 0.2 -total L 0.1 .tta Y93 Y94 Y95 Y96 Y97 Year Subcontracting In (% of plants) 0 20 0 0.6 -a-medium o 0.114 small a) 0.12 ta 0.10 Y93 Y94 Y95 Y96 Y97 Year Subcontracting Out 2 (% of plants) 0.38 0 036 . I + large 21 0.34 _ medium o0 0.32 Small T 0.3 -Total ) t0.28 Y93 Y94 Y95 Y96 Y97 Year 72 5.16 Trends in these variables by plant size largely mirror aggregate trends, with one exception -- subcontracting-out by large plants. The third graph shows that a higher proportion of large plants outsource production to others over this period, rising from 32 percent to 35 percent. If large plants tended to benefit more in terms of exports from the currency devaluation, it appears that they may have passed on some of these benefits to others by outsourcing. 5.17 Information elicited in the 1995 ENESTYC survey suggests that outsourcing, and concomitant supplier development, appears to be associated with transfers of assistance from buyers to suppliers. Table 5.3 shows the percentage of suppliers reporting different types of assistance from their buyers.53 Over half of suppliers receive some kind of assistance from buyers, the most common being supplies of raw materials, followed by all kinds of technical assistance. Less common, but very important in light of the credit difficulties faced by small Mexican-owned firns in the mid-1990s, is help with financing. While suppliers that have some foreign capital are more likely to receive assistance from buyers-many of whom may be foreign-owned firms or maquiladoras-Table 5.3 shows that more export-oriented suppliers are more likely to receive all types of assistance, both financial and non-financial, from their buyers. Table 5.3: Assistance Received by Suppliers from Buyers (percent) Total Ownership Export Status Types of assistance received Some foreign 100% Do not Export from buyers capital Mexican export Any assistance 56.8 64.1 55.6 57.0 56.5 Supply of raw materials 39.9 43.4 39.3 39.1 41.7 Financing 13.9 17.0 13.4 12.8 16.7 Workertraining 13.6 22.6 12.1 11.6 18.5 Leasing of equipment 11.7 7.5 12.5 11.2 13.0 Technical assistance 22.7 35.8 20.4 20.9 26.8 Source: 1995 ENESTYC Survey, questions 32 and 33. E. TFP Effects of Exporting and Subcontracting 5.18 The TFP panel regressions were estimated using a random effects model which takes into account correlations in the error structure introduced by repeated observations of the same plants over time. The model is: TFPit = Po + PI Eventt + ,B2 EXpit + P3 Exp 2it+P4Xit +34Zi +Vj + Sit where TFP is the residual measure estimated from the first-stage regression, Event is the contemporaneous indicator variable for the event (exporting and subcontracting), Exp 53 We note that patterns of assistance provided suppliers-from the perspective of outsourcing firms-are broadly similar to those reported in Table 5.3. 73 and Exp2 is the quadratic specification of years of experience with that event, X is a vector of firm and industry attributes for the individual establishment i in time t, and Z1 is a vector of other high-perfornance firm attributes that are time-invariant. The error tenns, vi and &it, require some explanation: &it is the normal regression error that is, by assumption, uncorrelated with both the dependent and explanatory variables; v; is a time- invariant error term specific to the plant and, in the context of our discussion, positively correlated with the plant's level of productivity because of unmeasured plant attributes. This error term confounds causal interpretation of the TFP effects of an event --the PI , P2 and P3 coefficients, and we attempt to minimize the role of the unmeasured productivity attributes, vi , by including several Z1 control variables from ENESTYC.54 5.19 Table 5.4 reports the regression results using the full EIA sample of 28,625 plant- year observations which include 6,291 unique plants. Model 1 includes current export status only; Model 2 also includes a quadratic specification of years of experience in exporting. Both models include event histories for subcontracting-in and outsourcing, a vector of equity ownership by foreigners, indicator variables for region, plant size, industry and year, as well as missing value dummy variables when information on ownership and region are missing. In the interest of space, coefficient estimates for region, industry and firm size are not reported. 5.20 First, consider the TFP effects of exporting. In Model 1, current period exporting -- without any consideration of past exporting history -- is associated with an 8 percent rise in TFP, an estimate that is statistically significant at the I percent level. When exporting history is taken into account in Model 2, the TFP effect of current period exporting is reduced dramatically to 2 percent, and this variable is no longer significant. However, exporting experience is statistically significant (though squared experience term is not), suggesting that each year of accumulated exporting experience is positively (and linearly) associated with a 4.5 percent rise in TFP. It appears that learning through exporting is taking place -- while the immediate productivity gains from exporting are modest, the results show that sustained productivity gains accrue as experience accumulates.55 54 A fixed effects model may be necessary to eliminate the unmeasured productivity effects that remain after controlling for Z, and that are systematically correlated with the event variables of interest. While a fixed effects model will address many of these problems through first-differencing, it also eliminates the time-invariant firm and industry attributes of interest. We experimented and determined that the fixed effects coefficient estimates of interest were not very different from the random effects results. 55 How robust are the TFP effects of exporting history estimated using the random effects model? To test whether there are important biases from unmeasured productivity attributes correlated with exporting, we estimated a fixed effects (first-differenced) model. Instead of using TFP as the dependent variable, we estimated a fixed effects panel production function augmented to include export histories (both models are conceptually similar since the TFP regression model is derived from a first-stage production function), industry and time dummies. The estimated coefficients are very similar-0.031 for the contemporaneous export indicator (not significant), 0.052 for years of export experience (significant at the 1% level), and 0.004 for the experience-squared term (not significant). We interpret these results as evidence that biases in the random effects model from unobserved firm productivity attributes are minimal. 74 5.21 Of the subcontracting variables, only the subcontracting-in variables are statistically significant in both model specifications. Contrary to expectations, the estimated signs of the subcontracting-in variables are not consistent with learning effects from becoming a supplier. The contemporaneous effects (0.0468) are offset completely by the negative effects of years of experience (-0.0605), even when and the positive effects of experience-squared (0.0085) are taken into account. In fact, the productivity level of suppliers appear to be always less than that of plants that do not subcontract-in from others -- by between 4 and 5 percent -- a level that is unchanged with years of experience as a supplier. This is in marked contrast to Malaysia where the empirical evidence, using a similar model, is strong that supplier filrms learn (improve productivity) with years of experience as a supplier.56 Table 5.4: Determinants of Firm-Level TFP-Random Effects Model (Full EIA Sample) Dependent variable: TFP Model 1 Model 2 Explanatory variables: Coeff. t-stat Coeff. t-stat Constant term -0.3458 -4.347 -0.3252 -4.088 Exporting measures Export indicator variable 0.0805 6.319 0.0204 1.136 Years experience exporting 0.0451 2.764 Years exporting squared 0.0039 1.347 Subcontracting measures Subcontract-in indicator 0.0468 2.401 0.0521 2.675 Years subcontract-in -0.0605 -3.166 -0.0666 -3.491 Years subcontract-in squared 0.0085 2.314 0.0100 2.715 Subcontract-out indicator 0.0033 0.219 0.0044 0.295 Years subcontract-out 0.0254 1.677 0.0235 1.550 Years subcontract-out squared -0.0020 -0.720 -0.0023 -0.839 Ownership variables Foreign equity indicator 0.2329 7.380 0.2268 7.188 % US ownership 0.0046 7.916 0.0043 7.469 % Japanese ownership 0.0009 0.430 0.0006 0.276 % EU ownership 0.0053 7.398 0.0050 6.976 Ownership missing indicator 0.0823 1.120 0.1092 1.488 No. observations 28,625 28,625 No. unique plants 6,291 6,291 Overall R2 0.0583 0.0612 Note: the random-effects models included dummy variables for 8 regions, an indicator variable for missing region, and dummy variables for 2 firm sizes, 8 industries, and 4 years. 56 The Malaysia findings are in a forthcoming report by Hong Tan (1999) In that study, the author found a negative contemporaneous effect of being a supplier, a positive effect with experience, and a negative experience-squared effect. The implied learning effects with years of supplier experience ranging from 3 to 11 percent (varying by domestic and foreign ownership) for the random effects model, and from 5 to 13 percent for the fixed effects model. 75 5.22 Finally, the results in Table 5.4 indicate that plants with some foreign ownership have higher TFP than purely domestic Mexican firms. The productivity differential, which is statistically significant, is about 23 percent. The country of foreign ownership also seems to matter. Conditional upon the presence of some foreign equity, plants with a greater share of U.S. capital have much higher TFP levels; in contrast, a higher share of Japanese capital is not associated with higher TFP. 5.23 Table 5.5 reports the corresponding results for a similar exercise using the linked EIA-ENESTYC sub-sample of plants, this time including control variables for attributes of high-performance workplaces. In Model 2, the export history variables mirror those reported in Table 5.4 -- they suggest that there is a strong linear learning effect with years of experience, averaging about 4.4 percent per year. As before, statistical significance is only attained for the subcontracting-in variables, never for the outsourcing variables. The principal difference from the earlier results is that there now appears to be evidence of learning from supplier experience. 5.24 To see this, we report below calculations of the time profile of TFP gains with experience as a supplier. These experience profiles suggest that there is an initial dip in TFP levels from -0.4 percent to -1.3 percent when firms first become suppliers. However, supplier firms' TFP levels turn positive by the fourth year, and by the sixth year-the observed range in our data-their TFP levels are about 9 percent higher than that of non-supplier firms. Time Profile of Productivity Gains from Supplier Experience Time t 1 2 3 4 5 6 Impact on TFP Supplier? (0,1) .0399 .0399 .0399 .0399 .0399 .0399 Years of supplier experience -.0439 -.0878 -.1317 -.1756 -.2195 -.2634 Supplier experience squared .0087 .0348 .0783 .1392 .2175 .3132 Net Impact on TFP -.004 -.013 -.013 .004 .038 .090 Source: author's calculations based on Table 5.5. 5.25 What explains this change? One possible explanation was the presence of unmeasured productivity attributes, correlated with the subcontracting variables, that were not controlled for in the regressions using the full EIA data. This potential source of bias was addressed here in the linked EIA-ENESTYC sub-sample of plants through the inclusion of control variables for previously unmeasured attributes of firns, several of which had statistically significant (at the 1 percent level) effects on productivity. 5.26 Table 5.4 suggests that several attributes of high-performance workplaces are important determinants of productivity.57 First, note that concern for product and process quality, as manifested in the use of quality-circle practices, is associated with a 14 percent increase in TFP levels. However, controlling for quality control practices, ISO-9000 certification did not have a statistically significant impact on TFP levels; however, this 5 See the review of the literature in OECD (1996). 76 certification may be more important for signaling the quality and export credentials of a firm to foreign buyers, , as suggested in Chapter 3, than as a determinant of productivity. Second, worker training, in-house formal training in particular, also turns out to be an important determinant of firm-level TFP, though externally-based training was not statistically significant.58 The presence of an in-house formnal training program is associated with a 7.4 percent rise in TFP levels; increasing the proportion of workforce trained also raises TFP levels in the plant. Many firms invest in training and quality control practices on their own, but for some firms, SMEs in particular, there may be a role for effective government programs that deliver training, quality control and technical assistance to enhance SME productivity.59 5.27 To summarize, the analysis in this chapter indicated that: * Exporting firms have TFP levels that are, on average, 8 percent higher than that of otherwise similar non-exporters; * There is evidence that firms learn, and improve productivity, through experience exporting, and that each year of exporting experience is associated with an increase in TFP of over 4 percent; v There is a rising trend in supplier development over time, and evidence that subcontracting is accompanied not only by raw material supplies from buyers, but also by When adequate controls for unmeasured productivity attributes are included, the regression analyses indicate that firms learn and improve their productivity with years of experience as a supplier; * Firm-level productivity is improved by investments in worker training and implementation of quality control practices, and there may be scope for government programs to assist SMEs in upgrading worker skills and improving quality. 58 These results are consistent with those reported by Tan and Batra (1995). For a range of developing countries, they find evidence that fornal in-service training has a positive impact on firm-level productivity, and that the productivity effects are larger for in-house training as compared to external training. One of the countries in their sample was Mexico, and the Mexico analysis was based on the 1992 ENESTYC survey. 5 Examples of government programs to deliver training and technical assistance to SMEs include the Ministry of Labor's CIMO program, and similar prograrns run by SECOFI, NAFIN, and CONACyT to build SME competitiveness. 77 Table 5.5: Determinants of Firm-Level TFP (Linked EIA-ENESTYC Sample) Dependent variable: TFP Model I Model 2 Explanatory variables: Coeff. t-stat Coeff. t-stat Constant term -0.3487 -8.531 -0.3554 -8.630 Export measures Export indicator -0.0013 -0.059 -0.0016 -0.072 Years experience exporting 0.0450 2.305 0.0439 2.240 Years exporting2 0.0021 0.620 0.0022 0.659 Subcontracting measures Subcontract-in indicator 0.0399 1.682 Years subcontract-in -0.0439 -1.919 Years subcontract-in squared . 0.0087 2.005 Subcontract-out indicator -0.0053 -0.288 Years subcontract-out 0.0300 1.625 Years subcontract-out squared -0.0037 -1.103 Number horizontal linkages 0.0149 1.560 Ownership variables Foreign equity indicator 0.1629 2.836 0.16664 2.908 % US ownership 0.0033 4.196 0.0033 4.156 % Japanese ownership -0.0002 -0.093 -0.0002 -0.075 % EU ownership 0.0047 5.215 0.0046 5.112 Ownership missing indicator 0.4780 1.208 0.4634 1.172 1995 ENESTYC Variables QC indicator 0.1381 4.890 0.1374 4.870 ISO-9000 indicator 0.0009 0.086 0.0011 0.113 In-house Training indicator 0.0747 2.653 0.0744 2.650 External Training indicator 0.0460 1.747 0.0433 1.646 Proportion workers trained 0.0462 2.668 0.0461 2.666 Industry dummy variables 32 Textiles, apparel 0.1247 2.868 0.1094 2.477 33 Wood products 0.2719 3.761 0.2679 3.710 34 Paper products 0.1132 2.131 0.1063 1.996 35 Chemicals -0.0124 -0.303 -0.0161 -0.393 36 Non-met. minerals 0.2227 3.784 0.2217 3.769 37 Basic metals 0.0151 0.192 0.0083 0.105 38 Machinery & Equip. -0.0178 -0.469 -0.0264 -0.689 39 Other 0.3027 2.348 0.2961 2.302 No. observations 15,273 15,273 No. unique plants 3,208 3,208 Overall R2 0.0771 0.0795 Note: random-effects models are estimated for 1993-97 and include dummy variables for 8 regions, an indicator variable for missing region, and 4 years, which are not reported here. 78 6. CONCLUSIONS AND POLICY IMPLICATIONS 6.1 Beginning in the mid- 1 980s, Mexico began a process of structural reform that centered on the liberalization of trade and foreign investment. The reforns began in 1986 with Mexico's entry into GATT, and continued with the replacement of quantitative restrictions with tariffs, a reduction in the levels and dispersion of tariffs, elimination of official prices, simplification of trade regulations, and easing of restrictions on foreign investment. In the early 1990s, Mexico joined the United States and Canada in the North American Free Trade Agreement, which virtually eliminates barriers to trade and investment among the three countries. 6.2 The objective of the liberalization of trade and investment was to achieve higher and more sustained rates of economic growth, by expanding the country's participation in dynamic external markets and facilitating access to high-quality inputs and technologies. Higher rates of productivity growth were expected in a more open economy, as resources were reallocated to more efficient uses and competition encouraged continued reductions in costs and improvements in product quality. It was hoped that the greater outward orientation and its associated efficiency gains would extend to a large share of productive sector firms -- small as well as large, traditional as well as modern, and those located in less-developed regions of the country -- thus benefiting a broad segment of the population. If increased export orientation reaches a broad range of enterprises, industries, and regions, it provides a vehicle to reduce the performance gap between the "two private sectors" in Mexico: the first dominated by large, technologically- sophisticated, and often foreign-owned firms on the one side, and the other characterized by smaller, less efficient, Mexican-owned firms that traditionally have been oriented toward the domestic market. 6.3 In fact, the openness of the Mexican economy increased dramatically over the past ten years. The sum of exports and imports as a proportion of GDP rose from 23 percent in 1985 to 42 percent in 1995; exports have grown an average of 16 percent per year in U.S. dollar terns; and private investment in the export sector grew faster than in the rest of the economy. Manufactured exports increased substantially, achieving growth rates similar to those of Hong Kong and Singapore, the two Asian tigers with the greatest export dynamism. Mexico has moved from being primarily dependent on oil for its export revenues, to being primarily an exporter of manufactured products. 6.4 Still, exports of manufactures remain highly concentrated across firms. The manufactured export sector is dominated by maquiladoras, other firms with foreign ownership, and large Mexican enterprises. Moreover, export concentration appears to have increased during the 1990s. Although the growth in export value and in the number of firmns that export increased during 1993-98 for all size and ownership categories, the growth in export volume and outward orientation was more pronounced for large-scale and foreign-owned firms than for smaller, Mexican-owned firms. The automobile industry dominates manufactured exports, and the entry or expansion of a single 79 automobile firm can have a large influence on the total value of manufactured exports and on the regional distribution of exports. 6.5 The economic crisis that began in December 1994 accelerated the growth of the export sector. The devaluation of the peso and the decline in domestic demand in 1995- 96 created strong incentives to increase exports, both among incumbent exporters and among previously non-exporting firms. Between 1993 and 1997, the share of plants in the EIA that exported some of their production rose from 26 percent to 42 percent, and included large numbers of SMEs and Mexican-owned firms as well as larger and foreign- owned firms. During the crisis, exporting acted as a "safety valve" for many firms whose domestic market had all but disappeared. 6.6 As the domestic economy recovered from the crisis in 1997-98 and the peso gained against the dollar, some exporters shifted back to the domestic market. A large number of these "exits" from exporting were SMEs and Mexican-owned firms, while larger and foreign-owned firms seemed to have a more permanent presence in external markets. It may be that the export orientation of smaller and Mexican-owned firms is determined more by short-term domestic demand conditions, while larger and foreign- owned firms are taking longer-term advantage of trade opportunities opened by NAFTA. The more transitory nature of exporting by SMEs and Mexican-owned firms may also be a result of the observed duality in access to financing during and after the crisis: large, foreign-owned, export-oriented firms were able to access international sources of finance, whereas smaller and less global firms suffered from the dramatic decline in domestic bank lending to the private sector. 6.7 Thus, while the shift toward a more open economy and the increase in manufactured exports have been impressive, greater outward orientation of the economy has not extended down as much as it could have to the "other" private sector in Mexico. While not denying the dramatic achievements of structural reforms, it is disappointing that they did not do more to reduce the duality of the Mexican private sector. As one Mexican government official put it during a recent interview, "We are victims of our own success." The export sector is successful, but because it is dominated by maquiladoras and other foreign firms, it tends to be an enclave that is not integrated with the domestic economy. The export sector clearly has been the engine of economic growth, but it has been less successful as a vehicle for equitable growth. 6.8 The report uses large panel databases on manufacturing plants to track export behavior and enterprise performance over time, in order to determine the factors explaining the decision to export and exactly what seems to drive improvements in firm- level productivity. There is a clear association between plant-level efficiency and exporting, even controlling for plant size, ownership, and industry. Regression results suggest that high-performance workplace practices and labor skills are linked with exporting and productivity growth, and that investments in quality and modernization are made in anticipation of entry into foreign markets. Qualitative survey evidence indicates that lack of access to financing is an important constraint for smaller and domestic 80 market-oriented firms, both as a limit to pre-entry investment as well as post-entry export expansion. 6.9 Among non-maquiladora plants, productivity grows substantially in the two to three years prior to entering export markets, and their productivity continues to grow relative to permanent exporters and non-exporters for at least two to three years after entry. These findings suggest that learning-by-exporting effects may exist. Regression results indicate that the productivity benefits of exporting grow as years of exporting experience accumulate -- in other words, that there are even greater gains when exporting is "permanent" rather than "transitory". Other important determinants of productivity include the use of quality-circle practices and worker training (in-house formal training in particular). 6.10 The report investigates whether productivity gains extend to indirect exporters (firms that provide intermediate inputs to direct exporters) and other suppliers. If so, the efficiency benefits of greater outward orientation can filter through to a larger share of the productive sector. Using the detailed information on firm-level training, technology, and quality management in the ENESTYC survey, we do find evidence of learning effects associated with becoming a supplier to other firms. Information in the ENESTYC survey shows that outsourcing brings transfers of assistance from buyers to suppliers in the form of raw materials, technical assistance, training, and sometimes financing. 6.11 The objective of this report was to provide a retrospective view of the performance of the export sector in the wake of structural reforms, and thus did not evaluate policy options in depth. Nevertheless, based on our analysis of performance to date, our understanding of the Mexican industrial sector, and lessons from other 60 countries , we can suggest areas that seem to be important, directions for future policy reform, and topics that need further investigation: * The pre-entry phase of preparing to compete in export markets is important. Building labor skills, investing in high-quality equipment and materials, reorganizing the production line, and establishing quality management systems raise productivity and give firms the "credentials" to participate in foreign markets. The Government should focus on extending the coverage and quality of its support for enterprise training, technology diffusion, and information. * Buyer-supplier relationships are an important channel for extending the productivity benefits of exporting to a wider group of Mexican firms. The Government has long recognized the need to strengthen the linkages between the export sector and the domestic economy, and has a number of supplier development programs to encourage large firms to purchase from or subcontract to smaller Mexican firms. In general the impact of these programs has been limited. They should be re-evaluated in light of international experience, and more effective ways of facilitating private-to-private transactions should be designed. 60 See, e.g., Levy, Berry, and Nugent (1999). 81 * While "transitory" exporters may be able to move in and out of foreign markets as a reaction to changes in domestic demand, it appears that the productivity benefits of exporting increase as export experience accumulates. We need to better understand what encourages or discourages "permanent" exporting behavior, particularly among SMEs and Mexican-owned firmns. * Lack of access to credit, which particularly affects SMEs and Mexican-owned firms, is a constraint to modernization and export entry as well as to export permanence and growth. Although it is beyond the scope of this report to propose specific measures, improving access to finance seems to be key to increasing export orientation across a broad spectrum of firmns. 82 BIBLIOGRAPHY Aitken, Brian, Gordon H. Hanson, and Ann E. Harrison (1997). "Spillovers, Foreign Investment, and Export Behavior." Journal of International Economics 43, pp. 103-32. Baldwin, Richard (1989). "Sunk Cost Hysteresis." NBER Working Paper No. 2911. Banco de Mexico (1997), Informe Annual 1996. Banco de Mexico (1999), "La orientacion exportadora de last empresas mexicanas", February (mimeo). Banco de Mexico/SECOFI (1999a), "Entorno de las empresas exportadoras mexicanas en el primer semestre de 1999", April (mimeo). Banco de Mexico/SECOFI (1999b), "Entorno de las empresas exportadoras mexicanas en el segundo semestre de 1999", November (mimeo). Bancomext (1998), "Encuesta sobre la problematica del sector exportador y el conocimiento y satisfaccion de los productos y servicios ofrecidos por Bancomext", June (mimeo). Bernard, Andrew B. and J. Bradford Jensen (1995). "Exceptional Exporter Performance: Cause, Effects, or Both?" Mimeo. Cervantes, Jesus A. (1996). "Cambio estructural en el sector externo de la economia mexicana." Banco de Mexico, Comercio Exterior, March. Clerides, Sofronis, Saul Lach, and James R. Tybout (1996). "Is 'Learning-by-Exporting' Important? Micro-Dynamic Evidence from Colombia, Mexico, and Morocco." NBER Working Paper 5715, August. Das, Sanghamitra; Mark Roberts; and James Tybout (2000). "Micro Foundations of Export Dynamics." Draft prepared under World Bank research project RPO 679- 20, April. Dixit, Avinash (1989). "Entry and Exit Decisions Under Uncertainty." Journal of Political Economy 97: 620-38. Dussel, Enrique, Clemente Ruiz Duran, and Michael J. Piore (1996). "Adjustments in Mexican Industries to the Opening of the Economy to Trade." MIT and Universidad Nacional Autonoma de Mexico (processed), May. 83 Krugman, Paul (1989). Exchange Rate Instability. Cambridge: MIT Press. Mercado, Alfonso and Maritza Sotomayor (1996). "El comercio de automotores entre Mexico y Canada." Comercio Exterior (46), July. Organization for Economic Cooperation and Development (1996). Technology, Productivity and Job Creation, Volume 2, Analytical Report, Paris. Ramirez, Jose Carlos and Kurt Unger (1997). "Las grandes industrias ante la restructuracion: una evaluacion de las estrategias competitivas de las empresas lideres en Mexico." Foro Internacional (37) April-June. Roberts, Mark J., Theresa A. Sullivan, and James R. Tybout (1995). "Microfoundations of Export Booms." Washington, D.C.: World Bank, International Economics Department. Roberts, Mark J. and James R. Tybout (1995). "An Empirical Model of Sunk Costs and the Decision to Export." Policy Research Paper No. 1436. Washington D.C.: The World Bank, March. Roberts, Mark J. and James R. Tybout (eds.) (1996). Industrial Evolution in Developing Countries: Micro Patterns of Turnover, Productivity, and Market Structure. New York: Oxford University Press for the World Bank. Roberts, Mark J. and James R. Tybout (1 997a). What Makes Exports Boom? Washington, D.C.: The World Bank. Tan, Hong and Geeta Batra (forthcoming 1999). Technology Development, Inter-Firm Linkages, and Productivity Growth in Malaysia, PSD Report to the Economic Planning Unit, Government of Malaysia. Tan, Hong and Geeta Batra, Enterprise Training in Developing Countries: Incidence, Productivity Effects, and Policy Implications, PSD Monograph, 1995. Tan, Hong (forthcoming 1999). Technology Development, Inter-Firm Linkages, and Productivity Growth in Malaysia. Tybout, James R. and M. Daniel Westbrook (1995). "Trade Liberalization and the Dimensions of Efficiency Change in Mexican Manufacturing Industries." Journal of International Economics 31 (1): pp. 53-78. World Bank (1988). "Mexico: The Economic Effects of the Trade Policy Reform." Report No. 7314-ME, June. World Bank (1994). "Mexico: Reform and Productivity Growth." Report No. 12605- ME, June 30. 84 World Bank (1998). "Mexico: Strengthening Enterprise Finance" Report No. 17733- ME, September. World Bank (1998). Mexico: Enhancing Factor Productivity Growth", (Country Economic Memorandum), Report No. 17392-ME. 85 Annex A: TRADE AND EXPORT ORIENTATION TABLES A.1 Balance of Payments: Current Account A.2 Balance of Payments: Capital Account A.3 Import Tariffs and Licensing Requirements A.4 Exports by Sector of Origin A.5 Mexico's Major Trading Partners A.6 Main Products Traded by Mexico A.7 Composition of Exports A.8 Firm Concentration of Exports A.9 Firm Concentration of Non-Maquila Exports A. 10 Export Status by Size of Plant A.11 Export Status by Industry A. 12 Export Status by Ownership A. 13 Entrants and Exits by Size A. 14 Decomposition of Export Growth A. 15 Export Transitions A. 16 Export Status of Plants (EIA) 86 Table A.1: BALANCE OF PAYMENTS CURRENT ACCOUNT (Billion dollars) 1991 1992 1993 1994 1995 1996 1997 p/ CURRENT ACCOUNT -14.6 -24.4 -23.4 -29.7 -1.6 -2.3 -7.4 REVENUES 58.1 61.7 67.8 78.4 97.0 115.5 131.5 Exports (merchandise) 42.7 46.2 51.9 60.9 79.5 96.0 110.4 Oil exports 8.2 8.3 7.4 7.4 8.4 11.7 11.3 Non-oil exports 34.5 37.9 44.5 53.4 71.1 84.3 99.1 Agriculture 2.4 2.1 2.5 2.7 4.0 3.6 3.8 Mining 0.5 0.4 0.3 0.4 0.5 0.4 0.5 Manufacture 31.6 35.4 41.7 50.4 66.6 80.3 94.8 In-bond 15.8 18.7 21.9 26.3 31.1 36.9 45.2 industries Other 15.8 16.7 19.8 24.1 35.5 43.4 49.6 Non-factor services 8.8 9.2 9.4 10.3 9.7 10.8 11.3 Tourists 4.3 4.5 4.6 4.9 4.7 5.3 5.7 One-day visitors 1.6 1.6 1.6 1.5 1.5 1.6 1.8 Other 2.8 3.1 3.3 3.9 3.5 3.8 3.8 Factor Services 3.6 2.9 2.8 3.4 3.8 4.2 4.6 Interest 2.9 2.2 2.0 2.7 3.0 3.3 3.8 Other 0.7 0.7 0.7 0.7 0.8 0.8 0.8 Transfers 3.0 3.4 3.7 3.8 4.0 4.6 5.3 Workers remittances 2.7 3.1 3.3 3.5 3.7 4.2 4.9 Other 0.4 0.3 0.3 0.3 0.3 0.3 0.3 EXPENDITURES 72.7 86.1 91.2 108.0 98.6 117.8 139.0 Imports (merchandise) 50.0 62.1 65.4 79.3 72.5 89.5 109.8 Consumer goods 5.8 7.7 7.8 9.5 5.3 6.7 9.3 Intermediate goods 35.5 42.8 46.5 56.5 58.4 71.9 85.4 In-bond industries 11.8 13.9 16.4 20.5 26.2 30.5 36.3 Other 23.8 28.9 30.0 36.0 32.2 41.4 49.0 Capital goods 8.6 11.6 11.1 13.3 8.7 10.9 15.1 Non-factorservices 10.5 11.5 11.5 12.3 9.0 10.2 11.8 Insurance and freight 1.8 2.1 2.2 2.6 2.0 2.5 3.3 Tourists 2.1 2.5 2.4 2.4 1.2 1.5 1.8 One-day visitors 3.7 3.6 3.1 2.9 1.9 1.9 2.1 Other 3.0 3.3 3.8 4.3 3.9 4.3 4.6 Factor services 12.2 12.5 14.2 16.4 17.1 18.1 17.3 Financial 11.8 12.0 13.7 15.7 16.6 17.7 16.8 Interest 9.2 9.6 10.9 11.8 13.6 13.4 12.4 Public Sector 7.3 7.7 8.3 8.0 9.2 8.7 7.5 Domestic currency 0.8 1.8 2.9 2.4 1.9 0.9 0.5 Foreign currency 6.5 5.9 5.4 5.6 7.4 7.8 7.0 Private Sector 1.9 1.9 2.6 3.8 4.3 4.7 5.0 Domestic currency 0.0 0.0 0.0 0.0 0.0 0.0 0.0 Foreign currency 1.9 1.9 2.6 3.8 4.3 4.7 5.0 Remitted profits 1.1 1.3 1.3 1.4 1.3 1.6 1.9 Reinvested profits 1.4 1.0 1.4 2.4 1.6 2.6 2.2 Commissions 0.1 0.1 0.1 0.1 0.2 0.2 0.3 Non-financial 0.4 0.5 0.5 0.7 0.5 0.4 0.5 Transfers * * * * * * * p/ Preliminary SOURCE: Bank of Mexico 87 Table A.2: BALANCE OF PAYMENTS PART 2: CAPITAL ACCOUNT (Billion dollars) 1991 1992 1993 1994 1995 1996 1997 p/ CAPITAL ACCOUNT 24.5 26.4 32.5 14.6 15.4 4.1 15.4 LIABILITIES 25.5 20.9 36.1 20.3 22.8 10.4 8.7 Loans and deposits 8.0 -1.6 2.8 1.1 23.0 -12.2 -8.8 Public Sector 0.1 -3.5 -2.2 -0.4 11.5 -8.9 -6.0 Development banks 1.7 1.2 0.2 1.3 1.0 -1.2 -1.0 Non-bank public sector -1.6 -4.7 -2.4 -1.7 10.5 -7.7 -5.0 Bank of Mexico -0.2 -0.5 -1.2 -1.2 13.3 -3.5 -3.5 Commercial banks 5.8 0.3 3.3 1.5 -5.0 -1.7 -2.0 Non-bank private sector 2.4 2.1 2.8 1.2 3.1 2.0 2.7 Total foreign investment 17.5 22.4 33.3 19.2 -0.2 22.6 17.5 Direct investment 4.8 4.4 4.4 11.0 9.5 9.2 12.5 New investment 3.4 3.0 3.0 5.7 6.8 5.5 9.3 Reinvestment 1.4 1.0 1.4 2.4 1.6 2.6 2.2 Inter-company accounts 1/ -0.1 0.4 -0.1 2.9 1.1 1.1 1.0 Portfolio investment 12.8 18.0 28.9 8.2 -9.7 13.4 5.0 Stock market 6.3 4.8 10.7 4.1 0.5 2.8 3.2 Domestic currency securities 3.4 8.1 7.4 -2.2 -13.9 0.9 0.6 Public sector 3.4 8.1 7.0 -1.9 -13.8 0.9 0.5 Private sector 0.0 0.0 0.4 -0.3 -0.1 * 0.1 Foreign currency securities 3.0 5.1 10.8 6.3 3.6 9.7 1.2 Public sector 1.7 1.6 4.9 4.0 3.0 8.9 -1.7 Private sector 1.3 3.6 5.9 2.3 0.6 0.8 2.9 ASSETS -1.0 5.6 -3.6 -5.7 -7.4 -6.3 6.7 In foreign banks 0.9 2.2 -1.3 -3.7 -3.2 -6.1 4.9 Credits to non-residents * 0.1 -0.3 * -0.3 -0.6 -0.1 Extemal debt guarantees -0.6 1.2 -0.6 -0.6 -0.7 0.5 -0.7 Other -1.3 2.1 -1.5 -1.3 -3.3 -0.2 2.7 ERRORS AND OMISSIONS -2.2 -1.0 -3.1 -3.3 -4.2 * 2.5 Value adjustments and purchases of gold and silver -0.3 * * * * * * CHANGE IN INTERNATIONAL RESERVES 2/ 7.4 1.0 6.0 -18.4 9.6 1.8 10.5 1/ Statistics on foreign investment include temporary imports of fixed assets by in-bond industries as of 1995. 2/ Gross international reserves minus intemational liabilities of the central bank with maturities under 6 months. */ Less than 50 million dollars p/ Preliminary SOURCE: Bank of Mexico 88 Table A.3: IMPORT TARIFFS AND LICENSING REQUIREMENTS 1990 1991 1992 1993 1994 1995 1996 1997 p/ STRUCTURE OF IMPORT TARIFFS (Percentage of total tariff items) Total tariff items 1/ 100.0 100 100 100 100 100 100 100 Exempt 2.5 2.4 2.4 3.5 9.6 9.7 13.9 14.4 Upto 10O%tariff 49.7 49.7 49.7 48.9 42.5 42.5 38.7 38.3 Over 10-20% tariff 47.7 47.9 47.8 47.5 47.1 42.9 42.4 42.2 Over 20% tariff .. .. .. .. 0.7 4.9 4.9 5.0 Prohibited 0.1 0.1 0.1 0.1 0.1 0.1 0.2 0.2 IMPORT LICENSING REQUIREMENTS (Number of tariff categories) Total 11,834 11,821 11,820 11,841 11,082 11,081 11,184 11,231 Unrestricted items 11,607 11,604 11,605 11,632 10,916 10,911 10,997 11,139 Consumer goods 1,491 1,491 1,492 1,504 1,629 1,744 1,604 1,635 Intermediate goods 7,951 7,948 7,948 7,961 7,317 7,441 7,358 7,478 Capital goods 2,165 2,165 2,165 2,167 1,970 1,726 2,035 2,026 Controlled items 210 200 198 192 149 153 170 75 Consumer goods 53 55 53 47 17 10 19 16 Intermediate goods 138 126 126 126 111 115 117 29 Capital goods 19 19 19 19 21 28 34 30 Prohibited items 17 17 17 17 17 17 17 17 (In percent of imports) Unrestricted items 2/ 86.4 91.1 89.5 78.4 89.4 92.9 93.1 91.8 Consumer goods 12.5 11.2 11.0 13.0 16.1 16.6 10.5 1 1.1 Intermediate goods 52.5 56.1 55.0 43.7 47.8 52.8 57.8 54.6 Capital goods 21.4 23.8 23.5 21.7 25.5 23.5 24.8 26.0 Controlled items 2/ 13.6 8.9 10.5 21.6 10.6 7.2 6.8 8.2 Consumer goods 6.9 4.3 4.9 3.9 3.1 0.1 4.5 5.6 Intermediate goods 5.8 3.9 4.6 16.8 5.9 5.4 1.3 0.9 Capital goods 0.9 0.7 1.0 0.9 1.6 1.7 1.0 1.7 1/ Figures may not add-up due to rounding off. 2/ Weighted by value of imports p/ Preliminary SOURCE: SECOFI 89 Table A.4: EXPORTS BY SECTOR OF ORIGIN 1980 1985 1990 1997 p/ TOTAL (Billion dollars) 1/ 15.5 21.7 26.8 65.3 (Percentages of total) 2/ 1. Agriculture and forestry 5.5 6.4 5.4 5.2 lI. Livestock and fisheries 1 1.6 0.7 0.6 III. Extractive industries 63.8 35.5 18.9 16.6 a. Oil and natural gas 61.4 33.2 18.2 15.9 b. Other extractive industries 2.4 2.3 0.8 0.7 IV. Manufacturing industries 29.7 55.4 74.7 77.3 a. Food, beverages and tobacco 3.5 4.1 4.5 4.6 b. Textiles and leather products 0.9 2.4 4.5 5.6 c. Timber products 0.3 0.6 0.8 1.0 d. Paper and printing 0.4 0.8 0.7 0.8 e. Oil by-products 6.2 3.3 1.1 1.0 f. Petrochemical products 0.5 1.1 0.4 0.4 g. Chemical products 3.1 6.3 6.1 5.9 h. Plastic and rubber products 0.2 0.5 1.0 1.1 i. Non-metallic mineral products 1.4 2.0 2.1 2.2 j. Iron and steel 1.1 3.4 4.4 4.7 k. Mining and metallurgy 1.9 3.6 2.7 2.4 1. Transportation equipment and machinery parts 9.8 27.0 45.6 47.0 m. Other industries 0.3 0.5 0.5 0.6 V. Non-classified products and other 0.0 1.0 0.3 0.2 I/ Excludes in-bond industries. 2/ Figures may not add-up due to rounding off. p/ Preliminary SOURCE: Bank of Mexico. 90 Table A.5: MEXICO'S MAJOR TRADING PARTNERS (In percentage) Exports 1/ Imports COUNTRIES 1994 1995 1996 1997 p/ 1994 1995 1996 1997 p/ TOTAL 2/ 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 AMERICA 92.3 92.1 92.8 93.5 74.9 78.8 79.9 79.0 North America 87.3 85.8 86.2 87.3 71.1 76.3 77.4 76.5 United States 84.9 83.3 83.9 85.4 69.0 74.4 75.5 74.7 Canada 2.4 2.5 2.3 2.0 2.0 1.9 1.9 1.8 Central America 1.1 1.2 1.2 1.4 0.2 0.1 0.2 0.2 South America 2.7 3.7 3.6 3.5 3.3 2.0 1.9 2.1 Argentina 0.4 0.4 0.5 0.5 0.4 0.3 0.3 0.2 Brazil 0.6 1.0 0.9 0.6 1.5 0.8 0.8 0.8 Chile 0.3 0.6 0.7 0.8 0.3 0.2 0.2 0.3 Colombia 0.5 0.6 0.5 0.5 0.2 0.1 0.1 0.1 Peru 0.2 0.2 0.2 0.2 0.3 0.1 0.1 0.1 Venezuela 0.3 0.5 0.4 0.6 0.4 0.3 0.3 0.4 Other 0.4 0.4 0.4 0.3 0.2 0.2 0.1 0.1 Antilles 1.2 1.5 1.7 1.4 0.3 0.4 0.3 0.3 EUROPE 4.9 5.0 4.2 4.0 12.3 10.0 9.3 9.8 European Union 4.6 4.2 3.7 3.6 10.8 9.3 8.6 9.0 France 0.9 0.6 0.4 0.4 1.9 1.4 1.1 1.1 Germany 0.6 0.6 0.7 0.7 3.9 3.7 3.5 3.6 Italy 0.1 0.2 0.1 0.2 1.3 1.1 1.1 1.2 Spain 1.4 1.0 0.9 0.9 1.7 1.0 0.7 0.9 United Kingdom 0.4 0.6 0.6 0.6 0.9 0.7 0.8 0.8 OtherE.U. 1.1 1.1 0.9 0.9 1.1 1.5 1.4 1.5 Other 0.4 0.8 0.5 0.4 1.5 0.7 0.7 0.8 ASIA 2.5 2.6 2.9 2.2 12.2 10.7 10.1 10.5 Japan 1.6 1.2 1.5 1.0 6.0 5.5 4.6 3.9 Korea 0.1 0.1 0.2 0.1 1.5 1.3 1.3 1.7 People's Republic of China 0.1 0.1 * * 0.6 0.7 0.8 1.1 Taiwan * 0.1 * * 1.3 1.0 1.0 1.0 Other 0.7 1.1 1.1 1.0 2.8 2.2 2.3 2.7 REST OF THE WORLD 0.3 0.2 0.2 0.2 0.6 0.5 0.7 0.7 1/ Includes in-bond industries. 2/ Figures may not add-up due to rounding off. */ Less than 0.05 percent. p/ Preliminary. SOURCE: Bank of Mexico 91 Table A.6: MAIN PROI)UCTS TRADED BY MEXICO EXPORTS IMPORTS 1985 1990 1997 p/ 1985 1990 1997 p/ TOTAL (Billion dollars) 1/ 21.7 29.8 65.3 TOTAL (Billion dollars) 1/ 14.5 31.3 73.5 (Percentage of total) 2/ (Percentage of total) 2/ Crude Oil 61.4 33.2 15.8 Automobile replacements 2.0 2.0 9.4 Automobiles 0.5 9.3 15.1 Computers 2.2 2.8 7.3 Trucks 0.1 0.1 6.1 Automobile motors 1.0 0.6 3.0 Automobile motors and spare Cereals 3.8 2.6 1.3 parts of motors 5.0 6.0 3.8 Spare parts for electric Computers 0.3 1.3 3.6 installations 2.1 2.1 2.7 Insulated cables Seeds, cotton and soybean 2.0 0.9 2.0 for electricity 0.1 0.6 2.6 Measurement instruments 1.6 0.9 1.4 Tomatoes & vegetables 1.7 3.2 2.2 Non-specified industrial Iron & steel manufacturers 0.8 1.8 2.6 machinery 0.8 0.9 1.5 Spare parts for machinery 0.3 1.1 3.2 Synthetic resin articles 0.6 0.7 1.3 Iron (in bars and pigs) 0.2 1.2 2.1 Metal working machinery 1.4 1.1 1.2 Textile articles 0.2 0.6 2.5 Steel tubes & plates 1.6 1.7 1.2 Spare parts for automobiles 1.1 1.6 1.6 Medicines 1.5 1.1 1.2 Coffee 2.3 1.2 1.3 Mixtures for industrial use 1.6 1.3 1.4 Copper and silver (in bars) 1.2 1.8 1.0 Resins 0.7 0.8 1.2 Fresh fruit 0.4 0.9 0.9 Prepared paper and cardboard 0.6 0.9 1.0 Beer & tequila 0.5 0.9 1.1 Pumps 1.7 0.9 1.0 Plastic materials & Fuel oil and gasoline 1.1 2.3 1.4 synthetic resins 0.3 0.9 0.9 Radio and telegraphic devices 1.4 1.9 1.1 Textile man-made fibers 0.4 0.6 0.9 Edible oils and fats 0.9 1.3 0.8 Glass & glassware 0.7 1.0 0.9 Radios and TVs 0.9 1.8 1.4 Frozen shrimp 1.5 0.8 0.7 Meat, fresh & frozen 0.7 1.0 1.1 Policarboxylic acid 0.7 0.8 0.4 Generators, transformers Oil by-products 2.8 1.4 1.0 & electric motors 1.0 0.5 0.7 Cattle 0.9 1.3 0.3 Cellulose pulp 1.3 1.1 0.6 Milk, skim & powder 0.7 1.8 0.5 Elevator machinery 0.7 0.7 0.6 Gas 2.2 0.3 0.3 Machinery for agriculture 1.2 0.3 0.2 Shipyard machinery 1.7 0.2 * Automobile assembly material 11.4 12.8 * Other 16.4 28.3 29.4 Other 49.9 52.5 53.2 1/ Excludes in-bond industries. 2/ Figures may not add-up due to rounding off. * Less than 0.05 percent. p/ Preliminary SOURCE: Bank of Mexico 92 Table A7: COMPOSITION OF EXPORTS (NS million) 1993 1994 1995 1996 1997 31 Food, bev., tobacco 3,397.8 4,319.9 9,529.8 13,105.2 18,695.8 32 Textiles, garments, leather 1,394.6 1,432.1 4,313.0 6,869.9 8,889.3 33 Wood, wood products 499.5 416.8 626.9 1,079.7 1,452.2 34 Paper,paperproducts 350.9 434.9 1,664.1 1,211.4 1,471.3 35 Chemical products 7,191.4 9,632.9 24,457.2 28,772.6 31,676.6 36 Non-metallic minerals 1,101.5 1,614.8 3,487.0 5,135.1 6,017.0 37 Basic metal industries 4,269.8 4,921.6 19,993.0 20,567.6 23,727.6 38 Metal products, machinery 30,952.6 40,387.1 100,226.4 168,395.0 200,569.0 39 Other mfg. 93.6 132.9 320.0 436.5 583.0 Total 49,251.8 63,293.0 164,617.4 245,573.1 293,081.7 Composition of Exports (Percentage of total) 1993 1994 1995 1996 1997 31 Food, bev., tobacco 6.9% 6.8% 5.8% 5.3% 6.4% 32 Textiles, garments, leather 2.8% 2.3% 2.6% 2.8% 3.0% 33 Wood, wood products 1.0% 0.7% 0.4% 0.4% 0.5% 34 Paper, paper products 0.7% 0.7% 1.0% 0.5% 0.5% 35 Chemical products 14.6% 15.2% 14.9% 11.7% 10.8% 36 Non-metallic minerals 2.2% 2.6% 2.1% 2.1% 2.1% 37 Basic metal industries 8.7% 7.8% 12.1% 8.4% 8.1% 38 Metal products, machinery 62.8% 63.8% 60.9% 68.6% 68.4% 39 Other mfg. 0.2% 0.2% 0.2% 0.2% 0.2% Total 100.0% 100.0% 100.0% 100.0% 100.0% SOURCE: EIA 93 Table A.8.1: FIRM CONCENTRATION OF EXPORTS (Number of firms) 1/ 1992 1993 1994 1995 1996 1997 Value of Exports (US$000) >= 18 14,328 13,801 14,141 18,475 19,069 17,593 > 25 12,890 12,536 12,873 16,842 17,353 16,244 > 50 10,135 9,978 10,327 13,463 14,030 13,424 > 100 7,624 7,712 8,125 10,441 11,006 10,738 > 500 3,643 3,816 4,097 5,129 5,701 6,005 > 1,000 2,590 2,740 2,915 3,667 4,121 4,463 18-50 4,193 3,823 3,814 5,012 5,039 4,169 50-100 2,511 2,266 2,202 3,022 3,024 2,686 100-500 3,981 3,896 4,028 5,312 5,305 4,733 500-1,000 1,053 1,076 1,182 1,462 1,580 1,542 1,000-5,000 1,628 1,724 1,751 2,235 2,453 2,610 5,000-10,000 396 397 452 532 660 697 10,000-25,000 311 324 359 455 491 583 25,000-50,000 138 147 187 218 246 253 100,000-250,000 63 86 88 131 166 190 250,000-500,000 38 43 52 59 66 89 500,000-1,000,000 11 11 14 21 25 21 1,000,000-2,000,000 2 4 8 10 7 13 > 2,000,000 3 4 3 3 3 3 0 0 1 3 4 4 Total 14,328 13,801 14,141 18,475 19,069 17,593 1/ All exporting firms (RFCs), including maquiladoras, with exports greater than US$1500 per month. SOURCE: Bank of Mexico. 94 Table A.8.2: FIRM CONCENTRATION OF EXPORTS (percentage of firrns) 1/ 1992 1993 1994 1995 1996 1997 Value of Exports (US$ooo) >= 18 100.0 100.0 100.0 100.0 100.0 100.0 >25 90.0 90.8 91.0 91.2 91.0 92.3 > 50 70.7 72.3 73.0 72.9 73.6 76.3 > 100 53.2 55.9 57.5 56.5 57.7 61.0 > 500 25.4 27.7 29.0 27.8 29.9 34.1 > 1,000 18.1 19.9 20.6 19.8 21.6 25.4 18-50 29.3 27.7 27.0 27.1 26.4 23.7 50-100 17.5 16.4 15.6 16.4 15.9 15.3 100-500 27.8 28.2 28.5 28.8 27.8 26.9 500-1,000 7.3 7.8 8.4 7.9 8.3 8.8 1,000-5,000 11.4 12.5 12.4 12.1 12.9 14.8 5,000-10,000 2.8 2.9 3.2 2.9 3.5 4.0 10,000-25,000 2.2 2.3 2.5 2.5 2.6 3.3 25,000-50,000 1.0 1.1 1.3 1.2 1.3 1.4 100,000-250,000 0.4 0.6 0.6 0.7 0.9 1.1 250,000-500,000 0.3 0.3 0.4 0.3 0.3 0.5 500,000-1,000,000 0.1 0.1 0.1 0.1 0.1 0.1 1,000,000-2,000,000 0.0 0.0 0.1 0.1 0.0 0.1 > 2,000,000 0.0 0.0 0.0 0.0 0.0 0.0 Total 100.0 100.0 100.0 100.0 100.0 100.0 1/ All exporting firms (RFCs), including maquiladoras, with exports greater than US$1500 per month. SOURCE: Bank of Mexico. 95 Table A.9: FIRM CONCENTRATION OF NON-MAQUILA EXPORTS (US$ million) (Percentage Distribution) Jan.-Aug. J 1991 1992 1993 1994 1994 199 1991 1992 1993 1994 19 26,854 27,516 30,033 34,613 22,153 31,680 100.0 100.0 100.0 100.0 10 EX 18,538 19,130 22,306 27,162 17,310 25,951 69.0 69.5 74.3 78.5 7 8,316 8,386 7,728 7,451 4,843 5,729 31.0 30.5 25.7 21.5 2 rs 4,925 5,249 6,414 7,708 4,731 7,312 18.3 19.1 21.4 22.3 2 ns 150 159 278 311 237 195 0.6 0.6 0.9 0.9 aining Firms in Order of Export Value 1,342 1,353 1,497 2,406 1,517 1,759 10.0 9.9 9.6 12.6 1 0 870 863 983 1,261 861 1,114 6.5 6.3 6.3 6.6 5 610 638 773 823 559 830 4.5 4.6 5.0 4.3 0 457 472 556 632 402 619 3.4 3.4 3.6 3.3 0 679 726 801 918 626 932 5.0 5.3 5.1 4.8 0 474 516 647 721 492 716 3.5 3.8 4.1 3.8 0 370 438 532 585 408 593 2.7 3.2 3.4 3.1 0 308 343 443 479 334 489 2.3 2.5 2.8 2.5 0 268 294 389 439 283 424 2.0 2.1 2.5 2.3 0 240 264 318 380 242 384 1.8 1.9 2.0 2,0 0 215 232 270 344 220 352 1.6 1.7 1.7 1.8 00 192 209 243 308 202 313 1.4 1.5 1.6 1.6 10 170 195 222 281 190 286 1.3 1.4 1.4 1.5 20 160 182 202 254 174 262 1.2 1.3 1.3 1.3 30 153 169 185 235 155 244 1.1 1.2 1.2 1.2 40 140 156 176 222 143 225 1.0 1.1 1.1 1.2 50 127 144 162 208 135 206 0.9 1.0 1.0 1.1 60 118 133 151 194 125 192 0.9 1.0 1.0 1.0 70 111 128 140 183 119 180 0.8 0.9 0.9 1.0 80 105 121 131 170 114 170 0.8 0.9 0.8 0.9 90 102 116 125 162 108 160 0.8 0.8 0.8 0.8 00 99 III 121 151 102 152 0.7 0.8 0.8 0.8 10 94 107 115 145 96 141 0.7 0.8 0.7 0.8 20 90 103 110 137 90 133 0.7 0.8 0.7 0.7 30 86 99 105 131 86 125 0.6 0.7 0.7 0.7 40 84 93 101 124 83 118 0.6 0.7 0.6 0.6 50 81 90 96 118 80 114 0.6 0.7 0.6 0.6 60 78 85 93 112 76 110 0.6 0.6 0.6 0.6 70 76 81 90 107 72 107 0.6 0.6 0.6 0.6 80 74 79 88 102 69 105 0.5 0.6 0.6 0.5 rms 7,973 8,540 9,865 12,332 8,163 11,555 59.2 62.2 63.2 64.4 6 ms 5,491 5,185 5,748 6,809 4,179 6,890 40.8 37.8 36.8 35.6 3 13,464 13,725 15,613 19,141 12,342 18,445 100.0 100.0 100.0 100.0 1o nk of Mexico, from customs data. 96 Table A.10: EXPORT STATUS BY SIZE OF PLANT Year Micro/Small Medium Large of Survey (Number) (%/) (Number) (%) (Number) (%) 1993 534 16.6 553 29.1 643 42.6 1994 502 16.0 544 29.3 669 45.8 1995 707 22.1 662 39.4 719 53.4 1996 722 24.6 716 43.7 827 59.0 1997 682 26.3 763 46.7 945 63.0 Total 3,147 20.9 3,238 37.2 3,803 52.7 SOURCE: EIA. Table A.11: EXPORT STATUS BY INDUSTRY (Number of plants) 1993 1994 1995 1996 1997 Total 31 Food, bev., tobacco 230 237 262 283 303 1,315 32 Textiles, garments, leather 240 229 353 394 413 1,629 33 Wood, wood products 33 28 58 72 76 267 34 Paper, paper products 106 92 99 116 120 533 35 Chemical products 430 424 477 520 552 2,403 36 Non-metallic minerals 85 96 107 109 112 509 37 Basic metal industries 50 62 80 82 80 354 38 Metal products, machinery 528 527 633 670 721 3,079 39 Other mfg. 28 28 31 30 32 149 Total 1,730 1,723 2,100 2,276 2,409 10,238 SOURCE: EIA. 97 Table A.12: EXPORT STATUS BY OWNERSHIP CATEGORY Year of Domestic 1/ Joint Ventures 2/ Foreign 3/ Missing Total Survey Number % Number % Number % 1993 1,128 22.5 148 51.0 449 34.8 5 1,730 1994 1,152 23.1 157 54.1 413 34.2 1 1,723 1995 1,481 30.4 171 59.2 444 40.1 4 2,100 1996 1,618 34.6 166 58.9 470 46.6 22 2,276 1997 1,676 37.6 184 67.9 494 52.8 55 2,409 Total 7,055 29.4 982 69.1 2,715 48.9 145 10,897 1/ Domestic = 100% Mexican ownership. 2/ Joint Ventures = 51-99% Mexican ownership. 3/ Foreign = 0-50% Mexican ownership. SOURCE:---- Table A.13: ENTRANTS AND EXITS BY SIZE Entrants Exits Year Micro/ Medium Large Total Micro/ Medium Large Total Small Small 1994 128 104 114 346 153 115 71 339 1995 255 199 127 581 88 57 42 187 1996 178 130 103 411 112 60 41 213 1997 149 124 112 385 110 70 54 234 Total 710 557 456 1,723 463 302 208 973 SOURCE: EIA. Table A.14: DECOMPOSITION OF EXPORT GROWTH Year Incumbents Entrants Exits Total N$m. % N$m. % N$m. % N$m. % 1994 13,411.2 97.4 1,789.9 13.0 -1,433.0 -10.4 13,768.2 100.0 1995 96,349.3 95.1 5,965.2 5.9 -992.0 -1.0 101,322.5 100.0 1996 48,371.4 70.9 20,543.2 30.1 -669.7 -1.0 68,244.9 100.0 1997 41,353.4 90.8 5,090.9 11.2 -881.2 -1.9 45,563.0 100.0 Total 199,485.3 87.2 33,389.2 14.6 -3,975.9 -1.7 228,898.7 100.0 SOURCE: EIA. 98 Table A.15: EXPORT TRANSITIONS (Number of plants) Year Missing Incumbents Entrants Exits Non-exporters ao 1993 6,620 0 0 0 0 6,620 1994 17 1,374 346 339 4,417 6,493 1995 28 1,513 581 187 3,970 6,279 1996 43 1,845 411 213 3,514 6,026 1997 68 1,999 385 234 3,101 5,787 Total 6,776 6,731 1,723 973 15,002 31,205 SOURCE: EIA. Table A.16: EXPORT STATUS OF PLANTS IN EIA Year Total number of Exporters Non-Exporters Exports of Survey Plants (Number) (%) (Number) (%) (N$bil.) 1993 6,620 1,730 26.1 4,890 73.9 49.3 1994 6,493 1,723 26.5 4,770 73.5 63.3 1995 6,279 2,100 33.4 4,179 66.6 164.6 1996 6,026 2,276 37.8 3,750 62.2 245.6 1997 5,787 2,409 41.6 3,378 58.4 293.1 Total 31,205 10,238 32.8 20,967 67.2 815.9 SOURCE: EIA. 99 Annex B: DATABASES A. INEGI EIA Panel Database INEGI conducts an annual survey of manufacturing establishments called the Encuesta Industrial Anual (EIA). The sample frame includes about 6500 establishments (plants) that account for about 80 percent of production in each six-digit industry group. Because the sample framne is chosen in this way, it includes all of the largest plants in the population and a significant share of medium-scale'plants, but a smaller share of small- scale plants; there are not many microenterprises in the sample. Table B. 1 shows the distribution of plants by two-digit industrial group. Table B.1: Number of Plants in ETA by Industry 1993 1994 1995 1996 1997 1998 Total 31 Food,bev.,tobacco 1181 1179 1175 1168 1147 1133 6983 32 Textiles, garments, leath. 1298 1299 1287 1279 1220 1164 7547 33 Wood, wood products 282 282 278 273 252 233 1600 34 Paper, paperproducts 518 518 514 498 488 478 3014 35 Chemical products 1285 1289 1269 1238 1209 1187 7477 36 Non-metallic minerals 535 531 527 520 490 470 3073 37 Basic metal industries 180 180 179 175 167 163 1044 38 Metalproducts,mach. 1511 1506 1483 1463 1397 1338 8698 39 Othermfg. 72 72 71 70 68 65 418 TOTAL 6862 6856 6783 6684 6438 6231 39854 Note: 1998 is constructed using monthly surveys. It is important to note that the EIA does not include plants that operate under the special maguila regime. The latter are covered by a separate census (described below) conducted by a different INEGI department. The two surveys are not strictly comparable in terms of industrial classifications, regional definitions, and the definition of value added. Each plant in the EIA is assigned a unique 10-digit identification number that consists of the plant's 6-digit industry code followed by a 4-digit code identifying the specific plant in that industry. The 6-digit industry for each plant corresponds to the product with the largest share of the total sales. Most of the plants in the EIA retain their identification number over time, but if a plant changes its production mix such that it is classified under a different 6-digit industry, its identification number would change. In order to maintain the same identification number over time for a plant that is "physically" 100 the same, we made adjustments in identification numbers for those plants that had changed industry classifications (see below). The design of the EIA was changed in 1993: the sample frame was expanded from about 3500 to 6500 plants, the industry classification system was changed, and therefore the firm identification numbers changed. New variables (such as the export variable) were added to the questionnaire. The pre- 1993 ElAs and the 1993-present EIAs have not been linked, so for this report we used the 1993-1997 EIAs only. INEGI's monthly survey of manufacturing plants (the Encuesta Industrial Mensual, or EIM) uses the same sample frame as the EIA (and the same plant identification numbers) but contains fewer variables. In particular, the EIM does not contain information on the plant's capital stock, so the usual method of estimating total factor productivity is not possible. However, for the variables common to the EIA and EIM, we aggregated the 1998 EIM to extend the EIA panel through 1998.61 To adjust plant identification numbers for plants that had changed their 6-digit industry classification during 1993-98, we constructed a re-classification dataset containing the old and new identification numbers using information on reclassifications maintained by INEGI. Since the same plant could experience several re-classifications during 1993-98, we applied an iterative procedure to re-assign identification numbers. In 23 cases we found inconsistencies in the re-classification information: plants (identification codes) that were supposed to leave the sample were still present, creating the danger of double-counting. For these cases we removed plants with both codes from the entire sample, losing 180 plant-year observations (0.45 percent of the original sample). Re-classification had a substantial impact on the dataset: 8.4 percent of the plants were reclassified. These would have distorted our analysis of entry and exit patterns had the reclassification not been done. Table B.2: Results of Plant Re-Classification Year ID number ID number Total unchanged changed 1993 6293 541 6834 1994 6285 541 6826 1995 6210 544 6754 1996 6100 553 6653 1997 5824 584 6408 1998 5610 587 6197 Total 36322 3350 39672 61 Note that since we have questions about the quality of this data, we exclude 1998 from much of the analysis in this first draft. 101 We did not balance the panel because we wanted to be able to observe plant entry and exit over time. A total of 6043 plants are present in the dataset for all six years. INEGI provided information on the reasons for leaving the sample for most of the plants that left the sample. Among the reasons were transition to maquiladora status (only five cases during 1993-98), bankruptcy, and temporary suspension of operations. Three other adjustments were made to the data. First, there were some cases in which a plant registered a positive value of production but zero employees. Most of these were plants that subcontracted out their production to other firms. Since our "size" variable is defined according to the number of employees, these plants would be mistakenly classified as microenterprises. For these cases, INEGI re-assigned a size category based on the plant's value of sales and an annually-adjusted sales-based definition of enterprise size used by NAFIN. Second, the regional classification of plants in the EIA is based on the location of the plant reporting data to INEGI. For multi-plant firms, this could mean that a plant's location is reported as the firm's headquarters location. INEGI used other information on the physical location of plants to adjust the regional variable in these cases. Finally, some plants reported no production, sales, or other income, even though they showed positive values for other variables. These plants were taken out of consideration in the data analysis. B. INEGI Maquiladora Panel Database INEGI collects monthly data on plants operating under the maquiladora regime. The plants are identified by SECOFI, which registers firms under this regime. Note that in contrast to the EIA, the Censo Industrial Maquiladora (CIM) covers all plants in the maquiladora population, instead of a sample. The number of maquiladoras has grown from 1,882 in 1990 to 3,308 in 1998. Table B.3: CIM Incumbents, Entrants, and Exits Year Continued Entrants Exits Total from previous year 1990 --1882-- 1882 1991 1791 268 (91) 2059 1992 2006 243 (53) 2249 1993 2069 287 (180) 2356 1994 2140 251 (216) 2391 1995 2025 402 (366) 2427 1996 2237 485 (190) 2722 1997 2548 432 (174) 2980 1998 2827 481 (153) 3308 102 The industry classification in the CIM is different from that used in the EIA; the CIM disaggregates automobiles/autoparts and electronics. The number of plants in each 2-digit CIM industry is shown in Table B.4. Table B.4: Number of Plants in CIM by Industry 1990 1991 1992 1993 1994 1995 1996 1997 1998 Total I Food products 54 58 63 64 67 65 76 84 89 620 2 Garments 301 355 418 450 485 559 696 835 984 5083 3 Shoes & leather products 51 57 63 67 61 62 61 65 63 550 4 Fumiture 246 274 303 325 319 297 338 350 370 2822 5 Chemical products 87 109 126 135 126 111 116 121 142 1073 6 Transportation equipment 168 174 180 185 186 182 195 211 224 1705 7 Mach. & equip. (excl. electronic) 40 45 50 48 46 44 43 42 43 401 8 Electronics assembly 117 117 121 122 125 128 133 143 145 1151 9 Electronics materials, accessories 403 406 427 449 448 416 452 466 507 3974 10 Toys and sports equipment 34 36 42 44 52 47 53 56 64 428 11 Other manufacturing industries 292 323 340 345 359 390 420 443 483 3395 12 Services 89 105 116 122 117 126 139 164 194 1172 Total 1882 2059 2249 2356 2391 2427 2722 2980 3308 22374 There are fewer variables in the CIM compared to the EIA and EIM, and due to the importance of the maquila sector as a source of employment, the CIM focuses on employment and wage data. "Value added" is defined as domestic value added, i.e., total sales less imported inputs. Despite the fact that maquiladoras are allowed to sell up to 75 percent of their production in the domestic market, the CIM assumes that all maquiladora sales are for export. In practice, however, this may be closer to the truth: anecdotal evidence suggests that few maquiladoras sell to the domestic market. In fact, some maquiladora production is exported to the U.S. and re-imported back into Mexico. Maquiladora plants are concentrated in the North and Pacific-North regions near the U.S. border, and to some extent in the Gulf region. Annual data for 1990-98 was constructed from the monthly CIM, aggregating monthly values for each plant. There were some differences in variables between the 1990-96 CIM and the 1997-98 CIM: the latter disaggregates employment and wages by gender, so we re-aggregated these for 1997-98. The unique plant identification number is a combination of the codes for state, municipality, two-digit industry, and a 4-digit plant identification number. Therefore, changes in the plant's location or industry can mistakenly change its identification number. Although we were not able to re-assign identification numbers to these plants, using separate SECOFI information on the identity 103 of maquiladoras we found that there were only about 54 "suspicious" cases during all of 1990-98. Table B.5: Maquiladora Plants by Region 1990 1991 1992 1993 1994 1995 1996 1997 1998 Total North 613 668 709 738 696 733 833 942 951 6883 Pacific-North 887 937 1018 1095 1114 1037 1149 1214 1365 9816 Pacific-Centr. 47 54 59 45 52 63 73 78 97 568 Pacific-South 2 2 3 4 4 5 7 8 9 44 Central-North 9 9 14 27 44 61 77 94 123 458 Central 25 34 45 48 49 62 76 84 101 524 Central-South 32 49 59 62 76 95 125 144 199 841 Gulf 248 279 310 309 323 329 329 352 378 2857 Southeast 19 27 32 28 33 42 53 64 85 383 Total 1882 2059 2249 2356 2391 2427 2722 2980 3308 22374 C. INEGI ENESTYC Database In two years, 1992 (covering calendar year 1991) and 1995 (covering calendar year 1994), INEGI has fielded a special survey called the Encuesta Nacional de Empleo, Salarios, Tecnologia, y Capacitacion (ENESTYC). The questionnaire was designed jointly with the Ministry of Labor (STPS) was designed to measure the impact of the opening of the economy on employment, wages, training, and new forms of labor contracting. Designed as a module attached to the EIA, ENESTYC gathers additional information on: * training (workers trained, trainers, course content, results) and health and safety issues (preventative actions, risks, and accidents); * technology (R&D expenditure, purchases, transfer, and origin of technology, quality control, and IS09000 certification); * labor (occupational structure, subcontracting, part-time work, unionization, and absenteeism; and * internal plant organization and production processes. In both 1992 and 1995, the sample frame was expanded beyond that of the EIA to cover more microenterprises and SMEs. The data has been used by STPS and the Bank to evaluate the impact of the Bank-financed CIMO training project on enterprise performance. 104 INEGI has linked the 1992 and 1995 surveys into a panel database. In addition, for this report we were able to link the plants in the 1995 survey with those in the 1994 EIA. This allowed us to use ENESTYC data on training, quality control, and technology use in calendar year 1994 with EIA-based estimates of total factor productivity and exporting behavior. Table B.6: Number of Plants in ENESTYC 1995, by Industry Large Medium Small Micro Total 31 Food, bev., tobacco 423 310 215 178 1126 32 Textiles, clothing, leather 201 297 247 91 836 33 Wood & wood products 30 80 78 59 247 34 Paper & paper products 102 159 85 33 379 35 Chemical products 227 332 190 73 822 36 Non-metallic minerals 71 88 100 76 335 37 Basic metals industries 41 41 39 17 138 38 Metal products & mach. 365 413 360 149 1287 39 Other mfg. Industries 19 26 17 10 72 Total 1479 1746 1331 686 5242 D. Banco de Mexico Database The Departamento de Medicion of Banco de Mexico maintains a database on international trade transactions that contains information received from Customs. It has data on individual import and export transactions by firm and 6-digit product, aggregated to monthly totals. Firms are identified by their tax identification number (the Registro Federal de Causantes, or RFC) - note that this means that the observation is the firm rather than the plant. The RFC also contains information on the month and year the firm began operations. Customs data is classified according to the international trade classification system, which is different from the industrial classification system used by INEGI. A concordance is available, but it is only approximate because it is based on a previous version of the trade classification system. Because it is derived from transactions through Customs, the database covers the universe of (legally) importing and exporting firns. Firms are identified as trading under the maquiladora, PITEX/ALTEX, duty drawback regimes. Information on the origin of imports and the destination of exports are also included. There are some problems of data reliability due to keypunch errors which cannot be independently checked. Although errors occur for a relatively high proportion of firms, their share of exports is small.