InterAm Database

National Law Center for Inter-American Free Trade

North American Integration Three Years After NAFTA: A Framework for Tracking, Modeling and Internet Accessing the National and Regional Labor Market Impacts

by Raúl Hinojosa Ojeda, Curt Dowds, Robert McCleery,

Sherman Robinson, David Runsten, Craig Wolff, Goetz Wolff

Support for this research was provided by the U.S. Department of Labor, Bureau of International Labor Affairs, (funded under appropriation code K105-5-$-RWBT-73611-2599-NAO), the U.S. Department of Commerce, Economic Development Administration (Project No. 99-06-07403), and the U.S. Department of Housing and Urban Development (Grant No. 95-2754 to the NADBank Technical Assistance Consortium, including Southwest Voter Research Institute, National Council of La Raza, ADVANCE EDC., and the UCLA NAID Center).

This research is solely the responsibility of the North American Integration and Development Center at UCLA and does not necessarily reflect the views of our funding institutions. We wish to gratefully acknowledge the research assistance of Jose Arredondo.

North American Integration Three Years After NAFTA: A Framework for Tracking, Modeling and Internet Accessing the National and Regional Labor Market Impacts

Introduction

In the years and months preceding the U.S. Congressional vote on NAFTA in November 1993, a wide variety of methodological approaches were developed to generate predictions of what might be the employment and income impacts of accelerated North American economic integration. A lively debate ensued as to the most appropriate methodology for conceptualizing relevant dynamics (trade, capital, and migration flows) and accurately predicting impacts at the national, regional, and sectoral levels. These debates served to push forward the development of new tools and techniques of policy relevant research, focusing attention on accurately determining the dimensions of the possible short and long term impacts, both positive and negative, of trade liberalization and regional integration. Given the political importance that the Congressional vote on NAFTA took on, this type of data analysis has now become a crucial part of the future of the U.S. political economy of trade and regional integration policy formulation. Since the passage of NAFTA, however, comparatively little research attention has been devoted to trying to track the pattern of North American integration and to explain analytically the significance of the complex trends now empirically observable. This lack of rigorous attention is particularly distressing given that a number of related policy issues are still very much unresolved and are central to the future of U.S. trade policy and North American integration. These issues include North American monetary stability and migration flows, the expansion of regional integration in the Western Hemisphere, and the expansion of Pacific Rim trade and investment and its potential impact on U.S. labor markets.

In addition, the NAFTA legislation in the United States will require a full report to the Congress on the impact of the first three years of the argreement, as a basis for evaluating its continuation and possible modification. Despite the lack of ongoing research, there has nevertheless recently been much speculation in the press about the impact on U.S. employment due to NAFTA trade liberalization and the Mexican Peso crisis. The recent round of speculations, however, have been based on estimations that have tended to reproduce many of the same errors which were typically found during the NAFTA debate on both sides of the issue. It should come as no surprise that opinion polls indicate that the public in NAFTA countries is still very confused about the employment impacts of NAFTA and about which country has benefited more from the agreement. Most of the recent round of estimates, however, are still at best based on partial information from one or another source, extrapolated by very imprecise or faulty methodologies. Rarely have recent estimates clearly presented the data and methodologies which underlie their conclusions. The lack of rigorous attention is particularly unfortunate due to the fact that new institutions and policies, such as North American Development Bank (NADBANK) and the NAFTA-TAA program, were developed as a part of NAFTA to provide an unprecedented level of support for identifiable cases of adjustment needs. Research is needed to develop and evaluate regional and sectoral based criteria for accurately implementing these new institutions.

This report is the result of an ongoing project by the UCLA North American Integration and Development Center to provide research, technical assistance and Internet accessible information on North American integration. We seek to construct new approaches to tracking the dynamics of North American integration and to make these approaches easily accessible to policy makers and organizations of civil society in all three countries. Specifically we are making this study, its supporting data, and methological information, all available through the NAID Internet WWW site (http://naid.sppsr.ucla.edu) via interactive database search engines. We hope that this can serve as model for a needed bridge between programs such as the NADBANK and NAFTA-TAA program and the public at large.

Through the example of this study, we hope that future studies will also take advantage of the new Internet medium to make available both their data and methodological steps for the sake of transparency, reproducibility, and further discussion (See Appendix 5). In this report we begin by reviewing a number of recent approaches that have been used to estimate changes in the pattern of trade and integration and their potential impacts on employment gains and losses. These include the tracking of pre and post-NAFTA trade flows (Espinoza and Noyola, 1996), the use of the Hufbauer-Schott (1993) employment multiplier analysis (AFL-CIO, 1995) , the use of NAFTA-TAA self-reported data on plant closures and import penetration (Bolle, 1996), and trade regression models used by de Janvry (1996) and Gould (1996). Following this review, we present an alternative methodology for:

(1) tracking the transformation in pre- and post-NAFTA patterns of trade and investment;

(2) evaluating the relative importance of NAFTA tariff liberalization versus other macro-economic dynamics;

(3) estimating the direct employment impacts of pre- and post -NAFTA patterns of trade and investment at a national level and at a 450 sector plus level for all the over 3,000 counties in the United States; and

(4) estimating the direct employment impact and eligibility of communities for NAFTA-TAA training and NADBANK adjustment investment assistance. We present this analysis with a partial equilibrium methodology, describing each analytical step. We then suggest a research program that will be necessary to implement this analysis more fully with a dynamic general equilibrium methodology.

A major contribution of this report is the development of a new methodological approach for estimating the sectoral, regional and aggregate impacts on the U.S. labor market of increased international trade using Armington elasticities estimated by the USITC (Reinert and Shiells, 1992). This methodology allows us to estimate more accurately the actual sectoral complementarity or competitiveness of imports and exports between Mexico and the United States and to estimate employment impacts at a national and regional/sectoral level.

This report presents the results of a detailed macro, sectoral and regional tracking of the evolution of U.S.-Mexico trade and the application of our methodology to estimate sectoral and regional employment impacts. We further explore the viability of these results through a number of regional/sectoral case studies of the effect of integration on local economies, including the applicability of this alternative tracking approach to determine more accurately eligibility for NAFTA-TAA and NADBANK. The methodological approach presented here should be seen as approximating the relative order of magnitudes of NAFTA-related impacts, seeking to stimulate further research and the full implementation of the proposed general equilibrium approach, rather than presenting final or definitive numerical estimates. In particular, we concentrate here only on the direct employment effects of trade, arguing for the need to expand this new approach to the estimation of the indirect employment impacts of exports and imports, which should also be done in a general equilibrium context.

The key findings of this report indicate that:

(1) The overall pattern of U.S.-Mexico trade and investment began to change radically nearly a decade before NAFTA with Mexico's unilateral trade liberalization, ushering in a dramatic growth in the two-way trade of intermediate goods, and has not significantly changed since the implementation of NAFTA. The most significant change in the U.S.-Mexico trade relationship over the last few decades has been an explosion of exports and imports in the last ten years, driven almost entirely by an expansion of Mexican manufactured exports based on the processing of imported intermediate imports (Figure 3.11). As a result, Mexican imports have become predominantly linked to the demand for Mexican exports rather than to fluctuations in Mexican domestic demand. This new import-export dynamic is growing even faster than the recent rapid expansion of Mexican maquiladora exports as the strategy of manufacturing for exports is adopted by many other regions, sectors, and types of firms in the Mexican economy. The period since NAFTA has merely extended these trends. We show that U.S.- Mexico intra-industry trade, and especially cross-border production, has been increasingly integrated over the last ten years.

The rapid growth in Mexican manufactured exports to the United States has been accompanied by a closely related growth in U.S. exports of intermediate inputs in similar industries. Mexican growth of industrial intermediate goods is also increasing, further accelerating a trend towards two-way, trans-border production. While this growth of imported intermediate inputs has come primarily from the United States, Mexico is also increasingly importing intermediate inputs from other countries for the purposes of processing and re-export. The determination of the "country of origin" of a good and the true value of international trade balances is thus becoming more and more difficult to define, a point that has been made by a number of observers (e.g. Reich 1991, Shaiken 1991).

This trade and production transformation dates back to Mexico's unilateral trade liberalization during the late 1980s which reduced mean tariff rates from 25 percent to 10 percent (Fig. 3.5). This represents a much larger liberalization than NAFTA required from Mexico over the last two years or will require over the next 15 years. Mexico's unilateral and NAFTA tariff reductions will, of course, be much steeper than those required from the United States due to its relatively lower tariffs before NAFTA.

A typical defense of NAFTA (Espinoza and Noyola, 1996) has been that, unlike the crisis of the early 1980's, the recent crisis has not resulted in such a deep and prolonged decline in U.S. exports to Mexico. Our research substantiates this difference between the two Mexican crisis periods, but also points out that this difference cannot easily be attributed to NAFTA. Rather the diference in crisis dynamics is primarily due to the structural transformations in U.S.-Mexico trade which began years before NAFTA and which have increasingly tied Mexican imports to export production for the U.S. market and have reduced the relative importance of imports linked to the Mexican domestic market. Like the Mexican crisis of 1982, the fall in U.S. exports to Mexico has been very severe in consumer and capital goods. Yet the current crisis is characterized by only a momentary decline in U.S. intermediate goods exports, as well as a related higher growth of Mexican manufactured exports (Figure 3.11). This "lock-in" of continued growth in U.S. intermediate goods exports to Mexico for processing and re-export is the principal reason why the crisis of the mid 1990s is different than the crisis of the early 1980s. Even before NAFTA, the re-imposition of import tariffs was no longer a relevant or realistic option as it would have effectively eliminated the lifeline of support for continued exports.

2) The lowering of tariffs through NAFTA has not had a significant impact on the rate of growth of imports or exports between Mexico and the United States, or on the composition of trade between sectors recently liberalized by NAFTA and those sectors still awaiting liberalization. The impact of NAFTA tariff liberalization appears to have had only a moderate impact so far on the pattern of U.S.-Mexico trade. Trade growth in sectors having undergone tariff liberalization in the first years of NAFTA was only a few percentage points higher than the rapid growth in non-liberalized sectors (Tables. 3.8 and 3.9). However, our analysis indicates that those sectors liberalized so far are the less labor intensive sectors, and that many of the adjustment impacts are still to be felt in the future.

(3) Most previous studies on NAFTA have significantly over-estimated job displacements due to imports from Mexico and, to a lesser extent, the numbers of jobs supported by exports.

Predictions on employment impacts from overall trends in U.S.-Mexico trade have been exaggerated, particularly with respect to job losses due to import penetration, but also with respect to job gains based on export growth. The most widely quoted methodological approaches used to estimate U.S. job losses due to trade with Mexico are seriously flawed (Hufbauer and Schott [1993], AFL-CIO[1996]), grossly overestimating the loss of jobs due to Mexican imports and thus overestimating the job losses related to the current growth of the U.S. trade deficit with Mexico. The major problem with this approach is that it equates the number of jobs created by exports with the number of jobs displaced by imports. There are many reasons to explain why previous studies have overestimated job displacement related to a growth in imports, as well as having overestimated the number of job supported by export growth:

(a) The most important explanation for the overestimation of job displacement related to imports is simply that imports do not always replace domestic production; that imported goods are often not close substitutes for domestically produced goods. Rather imports can be consumed by distinct market niches which complement the demand for domestically produced goods, are themselves complementary inputs into U.S. domestic production, or are consumed by growing domestic demand without displacing domestic production. This market and product differentiation has long been understood by trade theorists since Paul Armington's seminal work in the late 1960's. The U.S. International Trade Commission has actually spent a considerable amount of time measuring elasticities of substitution between imported and domestic goods, the so-called Armington elasticities, at a very detailed sectoral level for U.S. trade with various countries around the world.

(b) In the case of U.S.-Mexico trade, overestimation of import job displacement can also be attributed to a lack of conceptual and empirical attention to the high degree of complementarity of U.S.-Mexico production and trade which has emerged in the last 10 years of economic integration. A growing share of U.S. imports from Mexico in fact contain a very large proportion of U.S. intermediate exports to Mexico which are processed and then re-exported to the United States Thus it would be incorrect to count among the U.S. jobs displaced due to imports those indirect jobs involved in intermediary production, the common method in employment multiplier analysis. This method would lead to an overestimation of U.S. job losses related to imports from Mexico that contain a high degree of U.S. intermediate production.

(c) By the same token, a significant part of U.S. exports to Mexico are in fact not really final exports, but are used as intermediates in production for re-export back to the United States This has led to overstating U.S. job "gains" as defined as additional U.S. jobs supported by the penetration of external markets. Instead, these exports are merely domestic intermediate production where final production has been moved offshore. However, complementary joint production with Mexico sustains a larger and larger number of U.S. jobs which may have been lost entirely if production were displaced by imports from third countries.

(d) Another reason why imports from Mexico do not displace U.S. production on a one-to-one basis is some incidence of trade diversion from other countries, thus displacing third country production. This is partly due to NAFTA's relative lowering of tariffs within North America, but more so due to Mexico's recent large devaluation. Some Mexican exports are also the result of the transfer of production from third countries to Mexico, goods which were previously exported directly to the United States anyway, such as televisions or garments.

(e) A final reason to suspect an overestimation of job creation due to U.S. exports to Mexico is that our research indicates that U.S. data do not fully record trans-shipments of intermediate and final goods from third countries through the United States to Mexico. It is nevertheless important to point out that most pre- and post-NAFTA estimates of both job gains and losses, from a wide range of methodological approaches, are in fact very small and not proportional at all to the intense public fear or grand boasting that characterized the NAFTA debate.

(4) Using a newly developed partial equilibrium methodology for estimating direct employment dynamics related to trade, we find that:

(a) the overall positive or negative employment impacts of U.S.-Mexico trade have not been significantly affected by the liberalization of tariffs due to NAFTA; (b) the net employment impacts due to NAFTA tariff liberalization have been slightly positive, representing a very small share of new jobs being supported by exports to (or put at risk of displacement through imports from) Mexico;

(c) the overall net U.S. employment impacts since NAFTA implementation (1994-1996) have also been slightly positive, even taking into account the large impact of the peso crisis of 1995;

(d) negative employment impacts related to imports have actually declined when comparing pre- and post-NAFTA periods, due primarily to the steep annual fall in labor intensity of U.S. manufacturing; and

(e) the most important negative impact on employment has been the decline in U.S. exports due to the Mexican peso crisis, not the liberalization of tariffs due to NAFTA.

Using the methodological approach presented here allows one to estimate the relative magnitudes of direct employment impacts due to trends in pre- and post-NAFTA trade, comparing NAFTA liberalized sectors versus not liberalized sectors. The overall amount of jobs being supported by exports to Mexico, or jobs being put at risk by imports from Mexico, has not significantly changed when comparing the pre-NAFTA period (1990-1993) to the post-NAFTA period (1993-1994).

Employment impacts specifically from NAFTA (attributable to the elimination of tariffs) represent only a small subset of impacts from overall changes in U.S-Mexico trade and investment flows. Much more important in determining employment impacts are the overall levels of U.S. and Mexican growth rates and the impact of the Mexican peso crisis. We find that in 1994, there was a growth of about 42,000 U.S. jobs directly supported by exports to Mexico, while import growth resulted in a maximum of about 18,000 direct jobs placed at risk of displacement (see Table 4.11). During the 1995 peso crisis and the subsequent decline in U.S. exports to Mexico, we find that there was a reduction of about 33,000 jobs directly supported by exports to Mexico.

In 1995, import growth from Mexico put an additional 13,000 jobs at risk of displacement. In 1996, it is estimated that U.S. job growth supported by exports will be again be positive (over 26,000), while the number of jobs placed at risk through imports will continue to decline to about 7,000.

Even these employment numbers should be considered as large estimates for the direct effects because of a number of factors which bias our methodology to overestimate the job-displacing effects of imports in a growing economy. In other words, the impact on U.S. trade-related employment during the first three years after NAFTA is estimated to be, at the very least, a near zero net impact, and more likely, a moderately positive number.

While growth in two-way exports in the first and third years after NAFTA did indeed produce increases in net U.S. employment, the second year after NAFTA produced a net negative impact. Yet it is important to point out that the collapse in the Mexican growth rate due to the peso crisis of 1995 was the single largest negative impact on U.S. jobs, causing a deep net reduction in the number of U.S. jobs being supported by exports to Mexico. The reduction in direct jobs being supported by exports to Mexico (33,000) was almost the same as the total number of jobs that were put a risk by rising Mexican imports in 1994, 1995 and 1996 combined (37,000), years which included the launching of NAFTA, the peso devaluation, and the rise in the U.S. trade deficit with Mexico. The third year after NAFTA (1996) is expected to produce a return to a net positive direct employment impact on the United States, which is still lower by about one-half from the net positive employment impacts related to the first year of the initiation of NAFTA (1994).

(5) Though unfortunately not central to the NAFTA debates, the real issue with respect to U.S. employment levels concerns the development of policies that can achieve sustainable growth in Mexico. While the "Mexican Rescue Package" appears to have been crucial in averting another deep collapse, it is by no means certain that sustainable growth has been assured. Of all Mexico's trading partners, the United States has the most to lose when Mexican economic growth suffers. The U.S. trade deficit with Mexico is larger than the Mexican overall trade surplus, since Mexico has maintained a trade deficit with Asia and the European Union.

As in 1982, the vast majority of recent job losses are due to the decline in exports with a much smaller amount due to the increase in competitive imports. As in the early 1980s, the principal danger to jobs was not the pattern of trade, but the inability to sustain a stable macro-economic environment between the United States and Mexico. The top priority is to implement policies to ensure that Mexican growth is sustainable. This will most likely require polices and institutions for a further coordination and support of monetary and macroeconomic targeting in Mexico and the United States.

(6) The NAFTA-TAA program is a good indicator for estimating employment losses due to plants moving to Mexico, but is less reliable as an indicator of employment loss due to import penetration.

The NAFTA-TAA certifications also do not provide a good indicator of the level or composition of direct foreign investment to Mexico.

DOL has certified 82,223 workers for NAFTA_TAA through August 1996, or an average of 2,569 workers per month for 32 months.

While our methodology produces estimates which compare closely with the number of displaced workers certified due to plants moving to Mexico (NAFTA-TAA Certification Category C1), our analysis indicates a likely strong undercount for direct employment loss due to imports (C3). Since an analysis of the indirect employment effects of imports will require data currently not available, we cannot yet estimate the reliability of secondary employment certifications (C6). (Table 4.1) (7)

The NAFTA-related institutions and policies represent a net plus. Existing U.S. Government programs to address employment adjustments related to NAFTA, such as NAFTA-TAA (Trade Ajustment Assistance) program and the NADBank CAIP (Community Adjustment and Investment Program), are found to be underestimating the employment impacts based on the methodologies they are currently adopting.

Yet the scope of these programs are nevertheless found to be "in the ball bark" with respect to the orders of magnitude of the adjustment challange. Existing U.S. government criteria for NAFTA-related programs must be refined to be made more responsive and inclusive of the types of adjustment needs which are currently going unmet.

The NAFTA-TAA program, in particular, is based on self-identification of job layoffs resulting either from plant re-locations or layoffs due to import penetration.

Our analysis of the program indicates that it should be considered as a more reliable indicator of relocation of large-scale, unionized plants, but is least reliable for tracking import penetration, particularly in sectors and regions with small firms.

Our alternative methodology for determining employment impacts due to import penetration suggests a 50 percent undercount by NAFTA-TAA in such sectors, and a 90 percent regional undercount.

To the extent that NADBANK relies solely on NAFTA-TAA figures, they will of course reproduce these shortcomings. We recommend that NADBANK and NAFTA-TAA look to methods for measuring employment impacts of import penetration as an explicit criterion.

In addition, both the NADBANK and the NAFTA-TAA certification process must be made more aggressive to encompass employment impacts on smaller firms. In both cases, selected sectors and regions demonstrating strong import penetration should be targeted for pre-certification and/or aggressive outreach efforts.

We also make recommendations based on our sectoral and regional research which indicates significant problems in the ways in which data are used to track imports into the United States. More accurate reporting and monitoring of data methods is needed in a number of areas, including Armington elasticities, employment multipliers, and the trans-shipment of traded intermediate inputs.

In general, jobs gain/loss accounting methodologies should not be used to evaluate the relative benefits of trade.

These methodologies should, however, be central to our understanding of the adjustment costs of the impacts of trade. In evaluating the relative benefits of trade, it is much more important to focus on understanding the long-term dynamic impacts on productivity growth and overall welfare gains, not merely on the short-term employment effects.

Accurately identifying employment displacement risks, however, is very important to assist workers and communities take adequate steps to prepare for a positive adjustment.

Failure to identify and address adjustment risk will inevitably generate exaggerated political opposition to trade liberalization, in some cases based on ignorance and fear, and in some cases based on the legitimate defense of uncompensated individual costs which are incurred on behalf of the overall societal welfare.

The report proceeds in section two with a critical review of the pre- and post-NAFTA methodological approaches, indicating which have been useful for predicting or explaining recent trends. In section three, we present results from an alternative methodological approach to aggregate and sectoral tracking of the impact of NAFTA on U.S.-Mexico trade. In section four, we present the results of alternative approaches for estimating U.S.national and regional employment and income impacts of increased North American integration.

In section five, we present sectoral and regional case studies to complement our more macro analysis. In section six we present some conclusions and policy recommendations.

The appendices to the report present information on our modeling methodology and data base construction, as well as a description of the availability of this data for on-line querying through the NAID Center's NAFTA Tracking Internet Application.

We present this data as a contribution of the method by which we think discussion on North American integration can be best served: with clear presentation of all data sources and methodological steps.

I. Introduction 7

II. Review of Post-NAFTA Methodological Approaches 16

II.1 Explaining the Post NAFTA Pattern of Trade 16

II.2 Empirical Accounting of Employment Impacts of Post NAFTA Trade and Investment Flows: NAFTA-TAA 19

II.3 Estimating the Employment Impacts of Post NAFTA Pattern of Trade: The Hufbauer and Schott Approach 22

II.4 Towards an Alternative Methodological Approach to Tracking Post-NAFTA Impact: The NAID Armington 26

III. Evaluating the Impact of NAFTA on Trade: Results from an Alternative Methodological Approach to Aggregate and Sectoral Tracking 33

III.1 Explaining the Post-NAFTA Pattern of Trade and Investment 33

III.2 Pre and Post NAFTA Sectoral Structure of U.S.-Mexico Trade 43

IV. Estimating the National and Local Employment and Community Adjustment Impacts after NAFTA 52

IV.1 NAFTA-TAA: An Empirical Examination 53

IV.2 Employment Results of the NAID-Armington Direct Employment Impact Estimation Method 57

IV.3 NADBANK CAIP Criteria 64

IV.4 Comparative Description of Three Measures 69

V. Regional/Sectoral Case Studies 73

V.1 The Los Angeles Apparel Industry Case Study 73

V.2 Watsonville, California, Frozen Vegetable Industry and Naples, Florida: Fresh Tomato Growing and Packing 77

VI. Conclusion and Policy Implications 80

VII. Bibliography 120

Appendix 1: The NAID-Armington Estimation Model

Appendix 2: Data Base Construction for the NAID-Armington Estimation Model

Appendix 3: U.S. Trade DATA for an Analysis of Binational Integration

Appendix 4: NAID Center Application for NAFTA-TAA Certification Watsonville, CA

Appendix 5: Unified Framework for Tracking, Modeling, and Internet Data Accessing

Appendix 6: Western Hemispheric Trade and Employment results using the NAID-Armington Direct Employment Impact Estimation Method

Dr. Raul Hinojosa-Ojeda Research Director North American Integration and Development Center

Public Policy Building

Suite 2381

School of Public Policy and Social Research University of California Los Angeles

Los Angeles, CA 90095

310-825-8956 (Direct) 310-206-4609

(NAID Center) 310-825-8574 (FAX) http://naid.sppsr.ucla.edu