ISBN 92-2-111415-5
ISSN 1020-5322
First published 1999

Download the document in Adobe Acrobat's Portable Document Format (PDF)

Due to technical problems, not all tables are available - please contact the ILO for a hard copy if required

Contents

Foreword

This paper consists of chapters written by four separate authors: Bruce Campbell of the Canadian Centre for Policy Alternatives, Andrew Jackson of the Canadian Labour Congress, Mehrene Larudee of Williams College and Teresa Guttierrez Haces of the National Autonomous University of Mexico. Bruce Campbell coordinated its preparation. The four chapters assess the labour market effects of the signing and initial implementation of NAFTA, the North American Free Trade Agreement, from respectively an overall, Canadian, the United States and Mexican perspective. None is particularly happy with what they see, viewing the operation of NAFTA as globalization on their own doorstep and in many ways finding it wanting. Many in Canada fear that labour standards are being pushed down to the United States levels, or else that continuing devaluation of the Canadian dollar is required in order to compete, and failing either, that the United States corporations are withdrawing from Canadian plants. In the United States many fear that the threat of relocation of corporate activities to Mexico is enough to resist higher wage claims and other union activity. In Mexico the lower wage maquila, export-oriented sector, increasingly sets the tone for labour relations and wages overall. Furthermore the very initiation of NAFTA may have prompted investor expectations and subsequent capital inflows into Mexico to an extent that the authorities could not handle, leading to the 1994/95 peso crisis and devaluation. However, the authors also note that NAFTA has some protectionist elements in its local content rules which restrict the activities of third country enterprises. The concept of NAFTA, especially the economic ties between North America and Mexico was ambitious (and even closer Canada-US economic ties are also sensitive). Little or no new institutional machinery, only the relatively toothless side agreements especially on labour, accompanied it. NAFTA's creation was market driven and it seems it has not yet acquired consensual backing.

Gek-Boo Ng
Chief
Employment and Labour Market Policies Branch

I. CUFTA/NAFTA and North American labour markets: A comparative inquiry

1. Introduction

In its broadest sense, the North American Free Trade Agreement (NAFTA) which absorbed and replaced the Canada United States Free Trade Agreement (CUFTA), is an international treaty which serves as a vehicle for advancing and consolidating the neoliberal restructuring and integration of national economies and states in North America. It provides a legal framework which embeds the principles of market supremacy and international competitiveness. As the Canadian head of the OECD said in an interview with the economics editor of the Toronto Star: "these free trade agreements are designed to 'force adjustments on our societies.' Countries, he said, should push the pace of adjustment...by reducing social benefits that encourage the unemployed to turn down low paying jobs."⁽¹⁾

North American integration is a corporate-driven process. Trade and foreign investment growth amongst the three countries has far outstripped the growth of the three economies. Fifty large corporations (mainly US-owned) account for 70 per cent of United States-Canada trade. (Rugman 1995) This pattern is likely similar for overall NAFTA trade. FDI is even more concentrated. It should come as no surprise, therefore, that NAFTA and the accompanying restructuring of the state reflects the strategic interests of transnational capital.

NAFTA's role and its effects can only be fully understood within a broader policy and historical context. Moreover, any assessment of its labour market impacts must take into account the different economic, social, and political structures and institutions in each country and the strategies and actions of the main actors. These differences help to explain differences in labour market responses. We will attempt in our analysis to capture these complexities and contribute to an understanding of CUFTA/NAFTA's role and effect in the larger constellation of policies that have induced the "structural adjustment" of North American labour markets.

The thesis of this paper is that policy (national and international) is a major determinant of labour market outcomes. Neoliberal policies (of which CUFTA/NAFTA is a cornerstone) operate in a mutually reinforcing and cumulative way. These policies also interact in a mutually reinforcing and cumulative way with corporate strategies and with technological changes. Over time, these have had a generally adverse effect on employment and income conditions of a majority of working people and their families in all three NAFTA countries. It is our contention that this is not an unintended consequence of these policies. Underlying these policies are relations of power and its redistribution: from workers to corporations, from low and median income to high income earners, from wages to profits, from governments to the market.

What benchmark date should we use in assessing the Agreements impact. CUFTA negotiations began officially in 1986, although the Canadian government had publicly declared its intention to negotiate more than a year earlier. The deal was signed January 1, 1988 and came into effect on January 1, 1989. The neoliberal reign in Canada dates formally from the Mulroney government in 1984, although it had been gaining ascendance in policy circles for the previous ten years. The Mulroney government brought in significant privatization and deregulation measures during its first term of office - the deregulation of the energy, financial, transportation and foreign investment sectors, the privatization of public airline, oil, aircraft, telecommunications and satellite companies etc. These measures were then entrenched in CUFTA, which in turn provided the impetus for further restructuring - for example, in banking, telecommunications, rail transport - which continued under NAFTA and the subsequent Liberal government.

Full employment policies were supplanted by "natural rate of unemployment" policies. Monetary policy cut inflation in half (to under 5 per cent) by the mid-1980s. In the 1988-91 period the monetary screws were tightened even further, squeezing inflation down to under 2 per cent and jacking up unemployment into the 10-12 per cent range, four points above the United States rate. The federal corporate tax rate was cut in the late 1980s from 36 per cent to 28 per cent and the top federal income tax bracket was reduced from 35 per cent to 29 per cent.

The formal benchmark for neoliberal ascendancy in the United States dates from the Reagan presidency in 1980, although the consensus among policy makers had been building throughout the 1970s. Paul Volker preceded Reagan by a year as chair of Federal Reserve and had already initiated the era of extreme monetary austerity. It was Reagan who in his presidential campaign first floated the idea of a continental "common market" stretching from "the Yukon to the Yucatan." The Reagan Administration launched an aggressive drive to deregulate, privatize and downsize the United States civilian public sector, particularly social programmes like unemployment insurance and welfare while massively increasing military spending. Monetary austerity created a major recession and huge foreign capital inflow, both with worldwide reverberations. Together with tax cuts, these policies created a huge fiscal deficit and foreign debt build-up. Reagan also took a hard line against unions inspiring a wave of corporate and legislative attacks on organized labour. The Reagan government also attacked the Canadian government's interventionist orientation, especially its national energy and foreign investment review policies. The United States monetary policy in large part triggered the 1982 Mexican financial crisis, and then, along with the IMF put together a financial bailout package which pushed Mexico to alter its economic development path. The structural adjustment programme contained the basic neoliberal prescription:⁽²⁾ investment and trade liberalization, domestic deregulation and privatization, public sector cutbacks, and inflation controlling monetary policy and other forms of wage restraint.

Mexico's structural reforms began in 1983-84. the first stage culminated with GATT membership in 1986 followed by a dramatic opening of the Mexican economy - lowering of tariffs and non-tariff restrictions on trade, and loosening of foreign investment rules. The Salinas regime (1988-94) accelerated these policy changes and in 1990 entered into NAFTA negotiations with the United States (which Canada subsequently joined) to consolidate and lock-in the transformation of the Mexican economy. NAFTA was signed on December 17, 1992 and, after the negotiation of the labour and environment side-accords demanded by incoming the United States President Clinton, was ratified and came into effect January 1, 1994. Almost a year later, triggered by massive capital flight, Mexico was engulfed in a severe financial and economic crisis.

In summary, the key benchmarks in the history of neoliberal restructuring and integration (depending on the country) are: 1980, 1982, 1984, 1986, 1989 and 1994. The periods of economic recession for the United States were 1981-82 and 1991-92. Canada's slumps were roughly coincident with the United States, though deeper and longer during the 1990s. For Mexico the deepest periods of economic crisis were 1982-88 and 1995-96. The transformation in Mexico has been deeper and the crises much more severe and prolonged than the United States or Canada.

Reforms (including NAFTA) have been supported by a consensus of elites in all three countries, though they have met with widespread resistance amongst the general population. For example, it is unlikely that neither CUFTA in Canada nor NAFTA in the United States would have survived a referendum. The Salinas government sold NAFTA to the Mexican people as their ticket to first world-style prosperity, stifling critical political debate on its merits.

Once in place NAFTA serves as what Grinspun and Krecklewich (1994) call an external conditioning framework, limiting the range of political choices and enabling the implementation of unpopular measures. For example, major social programme cuts undertaken as part of the "war on the deficit" in Canada were rationalized with: "there is no alternative in this era of free trade and globalization." The choice of policy measures to deal with the peso crisis were similarly justified.

It should also be kept in mind that CUFTA/NAFTA is an agreement in progress. Many of its provisions are being phased in over time (10 years in the case of CUFTA and 15 years in the case of NAFTA.) It mandates a large number negotiations on a variety of areas from government procurement, to standards harmonization and common subsidies rules. It provides the legal architecture which locks in subsequent neoliberal restructuring by the NAFTA countries, and of course prevents backsliding.

2. Key provisions of NAFTA affecting labour markets

NAFTA like its predecessor the CUFTA, is a complex set of documents comprising a text of more than 1000 pages and even larger volumes of national implementing legislation, hundreds of pages of tariff schedules, statements of administrative intent, regulations, a long record of dispute panel rulings, and side-agreements on labour and environment.

NAFTA removes tariffs and other non-tariff barriers on all goods and services impeding governments ability to protect strategic or vulnerable sectors from import competition. It also prevents governments from granting conditional exemptions from tariffs or duty remissions to foreign transnationals as a way of strengthening domestic productive capacity and employment.

Although the two are interrelated, NAFTA's most important provisions apply to investment, not trade, liberalization. It entrenches a set of rules protecting private property rights of investors (corporations, banks, mutual funds etc.) and their investments. Virtually all types of ownership interests, financial or non-financial, direct or indirect, actual or potential, are covered. It liberalizes investment, enhancing its ability to operate free of non-commercial considerations. Codification of these provisions in treaty reduces the risk of a future government unilaterally imposing new conditions or regulations on their investments. The reduction of investment risks is central, enabling for example, transnational corporations to locate production more and more on the basis of cost considerations - labour, taxes, transport, infrastructure etc. free from "non-market" or political, impediments. It enables portfolio or money market managers to freely transfer assets and income into and out of member countries.

The very broad national treatment provisions of NAFTA oblige each member country to treat foreign investors exactly the same as it treats its own national investors regardless of their contribution to the national economy. They create the impetus for powerful alliances between foreign and domestically-owned businesses since any policy to regulate foreign capital will have to be applied equally to national capital. Their combined power to promote further deregulation and resist new regulation is greatly enhanced. Moreover, they remove important industrial policy tools such as subsidies to domestic high-tech firms with stronger linkages effects to the economy than foreign firms have.

It prevents governments from imposing a wide array of performance requirements (from local sourcing and product mandating to trade balancing and technology transfer), tools which have attempted to channel foreign investment to strengthen national industrial capacity and create jobs. It formally maintains, though in vestigial form, Canada's and Mexico's foreign investment screening mechanisms, designed to ensure, with the help of performance requirements, that potential benefits from foreign investment, including job creation, were in fact realized. (It also entrenches the United States "national security" foreign investment prohibitions.)

NAFTA prevents governments from regulating the outflow as well as the inflow of capital. It prevents governments from placing any restrictions on the transfer within the region or outside by any investor at market currency exchange rates, any kind of financial transfer including profits, dividends, royalties, fees, proceeds of sale of an investment, payments on loans to subsidiaries. It also prevents governments from restricting the transfer of physical assets including technological assets. There is a partial exception which allows governments experiencing balance of payment difficulties to impose limited trade and financial controls (excluding investment transfers such as profits and dividends), but only after consulting with the IMF and adopting only those measures which the IMF prescribes. Finally, NAFTA provides comprehensive intellectual property protection (patent, copyright, trademark etc.) for corporations' technology.

NAFTA guarantees investors the right to prompt compensation at "fair market value" for expropriations or measures which are seen to be "tantamount to expropriation"- a vague term for measures which are seen in some way to impair commercial benefits, including any future benefits which might be expected. Claims under these provisions may be adjudicated through various dispute panels, including an international tribunal at which corporations can directly challenge government measures.

It limits the ability of public or state-owned enterprises to operate in ways that are inconsistent with commercial practice and in any way impair benefits expected by private investors of the other countries. This clearly affects the ability of public enterprises to pursue public policy goals that may override commercial goals. It also limits the ability of future governments to re-regulate or re-nationalize once they have been deregulated or privatized (for example as has occurred in airlines, telecommunications, electrical utilities, transportation and energy sectors.) Thus, NAFTA has a built-in bias in favour of the private over the public sector, one which compresses the public space, ratchet-like, locking in every subsequent privatization or deregulation. Finally, NAFTA provides the legal framework for greater private (foreign and hence domestic) penetration into traditionally public areas, notably health care and education.

NAFTA facilitates and accelerates the realization of a privatized, continentalized transportation/communication infrastructure. It facilitates the development of high speed transportation systems that greatly reduce time and costs to the major consumer markets, important factors in industrial location decisions. It contains provisions which facilitate border-crossing for trucks the conduit for almost three-quarters of NAFTA trade. (Although the United States has thus far delayed the implementation of key trucking provisions.) Various north-south highway and rail corridors from Mexico through to Canada are in the process of being constructed or expanded. NAFTA also provides the legal framework for a continentalized telecommunications infrastructure which Mexico and Canada have moved closer to reality in subsequent deregulation/privatization measures.

Finally, NAFTA enables investors to challenge directly (if they do not wish to go through a national government) governments measures at all levels through an international investor-state dispute tribunal which they claim violate their rights. This further reinforces the bias of private sector interests over public policy considerations, putting a chill on any policy or regulation that might be perceived as infringement on investor rights.

3. The North American Agreement on Labour Cooperation (NAALC)

The NAFTA labour side agreement was finalized in September 1993, nine months after the NAFTA was negotiated. It came into being in response to warnings by the United States critics that NAFTA would facilitate the Mexican government's low wage foreign investment-led development strategy - in part by repressing fundamental labour rights - and thereby accelerate the loss of the United States jobs and incomes. This position was supported by incoming the United States president Bill Clinton who, as candidate, expressed concerns that transnational corporations could take advantage of "their ability to move money, management and production away from a high wage country to a low wage country. We could also lose income because those companies which stay at home can use the threat of moving to depress wages as many do today...if you look at the experience of the maquiladora plants...there is certainly cause for concern." (Clinton 1992, cited Levinson 1996.)

The inclusion of an agreement on labour rights as part of NAFTA was indeed an historic precedent, with signatories for the first time in an international trade agreement, acknowledging the link between the exchange of goods and services and the people who produce them; and promising to enforce their own labour laws and promote the 12 labour rights identified.

However the NAALC itself the fell short of critics' expectations. Its prime weakness - inadequate enforcement - stands in stark contrast to the strong enforcement of protections accorded to investors [see Stanford et. al. 1993, Robinson 1993, Levinson 1996, Bolle 1997]. It dealt only with non-enforcement of existing national laws, failing to address problems caused by the absence of laws or regulations. Most important, it did not provide effective means for changing the Mexican record of not enforcing it own labour law, or of changing the propensity of the United States and Canadian governments to weaken their labour laws to attract investment. Neither the National Administrative Organizations (NAOs) housed in each country's labour department, nor the NAALC Council of (Labour) Ministers, nor the NAALC Secretariat, were given sufficient independence or investigatory power to function effectively. Finally, Canadian provinces, which have primary jurisdiction over labour law (90 per cent the work force) were not covered under the NAALC. Thus, only if a sufficient number of provinces agree (as determined by a complex threshold formula) will the NAALC review and enforcement apply to Canada. To date just three provinces have signed on making it only partially in force.

NAALC created three groups of labour rights. Group III rights - workplace health and safety, and illness protection, protections for children and youth, minimum wages-have penalties for violation determined by an arbitration panel only after a lengthy (more than two years) dispute process. The monetary penalties themselves, the equivalent of 0.007 per cent of the value of one year goods trade between the Parties, were seen by many as insufficient.

Group II rights - the prohibition of forced labour, minimum employment standards pertaining to overtime, employment discrimination, gender pay equity, workers compensation, protection of migrant workers - contain no penalties for violation, only review and consultation among the NAOs, the Secretariat and the Ministerial Council and evaluation by outside committee of experts (ECE).

Group I rights - the right to associate, to organize, to bargain collectively and the right to strike, are subject only to NAO review and ministerial consultation. No penalties are provided for violation of these fundamental rights. Allegations of violations cannot even be evaluated by a committee of outside experts. Thus NAALC provides a limited set of penalties for a limited group of rights.

To date (June 1997) there have been 9 complaints filed with the NAOs. Two were combined in a single submission and one was withdrawn before hearings were held. All but one have been filed with the United States NAO. All but one alleged non-enforcement violations in Mexico. All but one pertained to one labour right, the right to organize (i.e. intimidation and firing of workers attempting to join independent unions.) As such, no complaint has gone beyond the NAO review and ministerial consultation stage. Only the most recent complaint, involving pregnancy testing, has the possibility of going to the next stage, independent expert committee evaluation (ECE).

The single complaint filed with the Mexican NAO was against Sprint (February 1995) for closing down a San Francisco telemarketing facility that workers were attempting to organize. The Mexican NAO requested ministerial consultations. One of the outcomes of the consultation was a report by the NAALC Secretariat on Plant Closings and Labor Rights (June 1997) which found that anti-union tactics were widespread in the United States. Another was a United States National Labor Relations Board ruling that the closing was motivated by anti-union bias and an order to the company to rehire the affected workers and compensate them for lost wages. The case is now being appealed by Sprint in the United States courts.

The first two petitions (February 1994) were filed as one by the Teamsters against Honeywell and the United Electrical Workers (UE) against General Electric alleging that the United States companies had dismissed workers for trying to organize an independent union, and that the Mexican government had failed to enforce its laws protecting the right to organize. The United States NAO report did not recommend ministerial consultations concluding that it was not able to make a decision as to whether or not the Mexican government was enforcing its own laws in part because the Mexican government itself had not made a judgment on the allegation of the employees. Instead it recommended a series of trinational workshops to promote discussion regarding freedom of association and the right to organize.

In the Sony case, (August 1994) the International Labor Rights Fund along with three other US and Mexican groups alleged that the company fired workers who tried to organize an independent union at five Sony-owned maquiladora plants. An allegedly rigged election conducted by the official union, the CTM, in collusion with the company subsequently reconfirmed the government union despite worker protests. A second attempt to form an independent union also failed when the government's Conciliation and Arbitration Board, on which members of the government, the corporations and the CTM sit - all with a vested interest in keeping out independent unions - refused to register the union.

The United States National Administrative Office found in its report that Sony had in fact intimidated the workers for trying to organize an independent union and agreed that the workers were gravely hampered by the registration procedure in setting up an independent union. The NAO recommended consultations among the United States and Mexican Labour Ministers. Ensuing consultations resulted in workshops, conferences and reports on union registration.

The long standing practice of blocking the formation of independent unions through its conciliation and arbitration boards was not altered by the NAALC review. No workers have been rehired as a direct result of this complaint.

The Pesca Union complaint (June 1996) alleged that the union which represented the Mexican Fisheries Ministry workers was improperly deregistered in the wake of a government reorganization, and recognition was granted to a rival union. Complainants alleged that members of the arbitration board were in a conflict of interest, violating ILO Convention 87. Studies, reports and ministerial consultations ensued.

In October 1996 a complaint was launched after a Mexican arbitration board allegedly denied union registration to workers at Maxi-Switch (a computer keyboard maker) after the company fired 400 workers for trying to organize an affiliate to an independent union and break the "contract of protection" with the official government-sponsored union. However, the complaint was withdrawn after the board reversed its decision and registered the independent union, one of the complainants. Some have seen this outcome as a the result of the publicity brought to bear on Mexican labour practices through the NAALC process. However, the company subsequently closed and reopened under a different name with the government union (CTM) once again the registered union in the plant.

The most recent complaint (May 1997) was brought by three United States and Mexican human rights groups alleging "a pattern of widespread state tolerated sex discrimination against prospective and actual female workers in the maquiladora sector.." Specifically, it involved mandatory pregnancy testing and denial or withdrawal of employment to those who test positive, thereby avoiding payment of the legal three-month maternity benefit. The United States NAO, as of June 1997, was still reviewing the case.

Despite the dismissal of NAALC as ineffectual by many critics, others argue that their criticism stems from unrealistic expectations about what was achievable and how it could be used. Compa (1997) and Herzenberg (1997) conclude that while common enforceable norms would have been desirable, the reality of wide income disparities and the overwhelming domination of one NAFTA partner, made effective enforcement of domestic law the more practical goal. They emphasize the importance of NAALC as a forum for subjecting countries' labour law to the glare of public scrutiny, as well as its value in stimulating communication, information exchange and building solidarity among labour rights advocates particularly between Mexico and the United States. (Canadian activists have not as yet become as engaged in NAALC.) They emphasize that its potential capabilities have not been sufficiently tested thus far and point out that the Secretariat and other institutional structures are still in their formative stage.

Herzenberg argues that "labour advocates' inability to use the NAFTA side-agreement to reverse specific worker rights violations should be neither surprising nor a primary basis for judging [its] usefulness. Effective use of the side agreement must be understood as part of broader organizing and political mobilization to challenge the ideological dominance of neoliberalism and gain the power to replace it with an alternative development model." [p.6]

4. NAFTA mechanisms which affect employment, incomes and standards

NAFTA in essence codifies a shift in the balance of power in favour of capital and away from governments (the kind that intervene in markets) and labour. By weakening public policy instruments and labour's power at the bargaining table it increases the pressure to level down employment conditions, wages and standards. These changes are being observed globally and as such are key elements of the process of globalization. NAFTA is not a separate phenomenon, but rather the concrete expression of globalization on the North American continent. Thus, when the ILO Director General said several years ago that globalization is eroding government policy instruments "which have such a decisive impact on the level and quality of employment and on domestic policies for social progress." (ILO, 1994:90-94) he could just as easily have substituted NAFTA which is, in effect, continental globalization.

It should be reiterated, however, that national social, political and economic institutions and structures differ from nation to nation as do the strategies of key actors. Therefore, national responses to NAFTA pressures like the response to broader globalization pressures, can be also expected to differ. We identify the following mechanisms associated with NAFTA-style integration.

4.1 Competitive pressure among corporations:
The concurrent deregulation and integration of the continental economy driven by expanding trade and foreign investment, intensifies competition not only among transnational corporations themselves, but also among national companies in trade-sensitive industries. As national markets with different costs and regulatory structures come into closer contact with each other, the pressure to cut costs and restructure through mergers and takeovers, downsizing, closures, relocations etc., increases.

In Canada, the largest wave of corporate restructuring occurred during 1989-1993. Although comprehensive statistics on plant layoffs and closures are not available for the country as a whole, Ontario, which contains 40 per cent of Canada's manufacturing capacity reported between January 1989 and August 1993, 452 permanent closures of major manufacturing facilities. Almost half of these plant closures were by foreign-owned (mainly US) companies. Significantly, 65 per cent of all layoffs during this period were the result of permanent as opposed to temporary, closures, compared to the 1981-82 recession where only 25 per cent of the layoffs were the result of permanent closures.

Increased competition also intensifies the pressure amongst employers to demand worker concessions - wages, benefits, conditions of work - as well as tax, spending and regulatory concessions from governments, especially programmes such as unemployment insurance which strengthen the bargaining power of workers. Finally, it increases the pressure (to the extent that technology permits) to lower costs through production and work reorganization - increasing the use of part-time, temporary and contract workers, and out-sourcing to non-union firms in low wage jurisdictions.

4.2 Immobile labour-mobile capital:
Labour's bargaining power is disadvantaged under NAFTA-style integration because with the exception of a few elite categories (business executives, entrepreneurs and certain professional and technical categories), workers are legally confined by national borders. Culturally, workers are bound to their communities and internal migration let alone immigration, occurs only in exceptional circumstances.

Capital on the other hand, inherently footloose, can move much more effortlessly under its new regime, or threaten to move if labour does not make concessions. There is much anecdotal reporting from both unions and employers and survey research providing evidence of this kind of pressure.

For example, a Wall Street Journal survey of 455 senior corporate executives (September 24 1992) taken just after the NAFTA was initialled, found that 25 per cent would use NAFTA to bargain down wages and 40 per cent would move at least part of their companies' production to Mexico as result of NAFTA.

A 1997 study for the North American Commission on Labour Cooperation (NAALC) by Cornell University researcher Kate Bronfenbrenner, found that "NAFTA has created a climate that has emboldened employers to more aggressively threaten to close or actually close their plants to avoid unionization." Her study of 500 organizing drives and 100 first contract campaigns in the United States found that in manufacturing and transportation sectors, 62 per cent of employers threatened to close and move all or part of their operations instead of negotiating with the union; 10 per cent explicitly threatened to relocate to Mexico and many more implied as much. Where such threats were made, companies were substantially more successful in keeping out unions than where they were not made. (see also Larudee, 1997, in this volume). Since unionized workplaces in the United States and Canada pay their employees 20 per cent to 30 per cent more on average than non-unionized workplaces, it is clear that this having a depressing effect on wages.

4.3 Competitive pressure on governments and regulatory structures:
NAFTA, by opening national economies intensifies pressure on hundreds of national and sub national governments to compete with each other to maintain and attract investment by increasing subsidies (most of which remain legal under NAFTA) and lowering regulations and standards. There are no common rules governing acceptable and unacceptable subsidies or limiting subsidy wars among governments, and only ineffective protections limiting competitive bidding down of labour and environmental regulations to the lowest common denominator. Policy levers such as performance requirements and tariffs which served as sticks to nudge capital to behave in accordance with public policy priorities and discourage capital outflows, have been removed and only carrots remain. Thus, the need to attract investment creates dual stresses: downward pressure on regulations and standards and increased pressure on existing fiscal resources.

This reality is not disputed in official circles. In fact at times it is used to justify specific measures to downgrade regulatory structures (whether or not objectively valid) contrary to commitments made by governments prior to NAFTA. NAFTA provides the external conditioning framework which allows policy-makers to claim there is no alternative in the new economic reality.

For example, a 1996 report by the Canadian government on key issues facing Canada to the year 2005 noted, "As we become more integrated with the United States the efficiency of our regulatory framework...will take on greater importance in corporate decision-making increasing pressure to harmonize with the United States...[For example] labour laws and policies in Canada tend to impose a higher regulatory cost on employers and reduce the flexibility and dynamism of labour markets in Canada relative to those in the United States according to the report, what specifically puts Canada at a disadvantage? "Minimum wages tend to be higher and hours of work and overtime regulations tend to be more restrictive, advance notice and severance rules tend to be more stringent and domestic labour laws are more conducive to the formation and retention of unions."

The social safety net is under the same pressure according to the government's report. "The basic affordability of the system and the benefits payment regime... has a direct consequence on competitiveness...By raising the cost of labour as a productive input such programmes can either drive jobs south or encourage further substitution of capital for labour."⁽³⁾

The government has already taken this advice to heart. Public spending cuts, which accelerated after 1989, moved into high gear under the Liberal government whose "war on the deficit" reduced federal programme spending by one-third in relation to GDP during 1995-97. Programme spending by all governments was cut from 40 per cent of GDP in 1992 to 35 per cent of GDP in 1996. Among the most prominent cuts have been those to unemployment insurance, most of which occurred under the current government - after 1993. As a result, the proportion of the unemployed eligible to collect unemployment insurance dropped from 89 per cent in July 1990 to 43 per cent as of July 1997 and is expected to fall to one-third once the reforms have been fully implemented (CCPA/CHOICES 1998). In an environment of high unemployment the impact of such measures on wage demands is obvious.

4.4 Pressure on fiscal capacity:
The competitive drive to attract and maintain investment increases the pressure to reduce taxes and increase subsidies to capital. The resulting fiscal pressure tends to crowd out social spending: from unemployment and disability insurance, welfare and pensions, to education, training and health care. The pressure to compress workers' wages and reduce their purchasing power also weakens the income and consumption tax bases. The enhanced ability of elite workers to move or threaten to move, either to work in other jurisdictions or to transfer their savings puts additional pressure on fiscal capacity.

Moreover, NAFTA, by enhancing the ability of transnational corporations to internalize their operations, increases the volume and relative importance of already high levels of intra-firm trade and accompanying practices of transfer pricing (where prices of imports and exports are set by managers to show minimum profits in high tax jurisdictions and maximum profits in low tax areas) and so too the pressure to lower corporate tax levels. (Vernon, 1994.)

4.5 NAFTA and Macro-Economic Policy:
NAFTA-style integration and macro-economic policies are bound together as mutually reinforcing pillars of neoliberal strategy. Locking in a regime that allows largely unregulated capital flows, the hallmark of NAFTA, limits macroeconomic policy control especially for the smaller partners. The fiscal pressures engendered by NAFTA have been outlined above as have the constraints on public spending.

In Canada, competitiveness imperatives under NAFTA also shape monetary policy priorities, especially the need for wage control. (The rationale for tighter monetary policy is that the "natural rate" of unemployment - the rate at which inflation is triggered - is higher than in the United States due to larger labour market impediments: more generous social benefits, stronger unions etc.) Competitiveness priorities in a NAFTA environment require disciplining labour through unemployment-induced monetary policy in the short term, and in the longer term, harmonizing the natural rate of unemployment with that in the United States through cutbacks to social benefits, weakening labour standards and other removing impediments to a competitive labour market.

Moreover, severe monetary tightening during 1988-91, by depressing domestic demand, forced business to turn to the more buoyant US economy thereby hastening the desired assimilation of the Canadian economy with its larger partner. The resulting rise in the Canadian dollar added to the pressure on import-competing industries to adjust to the new reality or close down.

Similarly, the interplay of macro-policy and NAFTA was evident during the 1994-95 Mexican financial crisis (see also Larudee in this volume). The Mexican growth and structural adjustment strategy depended on attracting sustained inflows of foreign investment. These flows had stalled in the late 1980s despite massive structural adjustment. NAFTA's role was to deepen and make permanent the liberalized trade and investment climate and thereby attract larger and sustained inflows of foreign capital. A central requirement of the strategy was to preserve and enhance its major competitive advantage: low cost labour. This was done with the help of several domestic policy tools for constraining wages outlined later.

The anticipation of NAFTA raised expectations among foreign investors eager to participate in a new era of Mexican growth. FDI inflows, the productive capital that the government was seeking, did increase in the pre and immediate post-NAFTA period (1989-94). However, three-quarters of the inflows were portfolio and other short-term capital eager to benefit from expected post-NAFTA boom, and lured by high interest rates on Mexican government bonds. Only 15 per cent of the inflow was invested in production, and much of that was invested in export production with weak linkages to the domestic economy. (cited Dillon,1997, 63). Moreover, NAFTA had removed most industrial policy tools to strengthen backward linkages from the export sector to the domestic economy.

Foreign mutual funds flooded into the previously closed Mexican stock market which was now flush with issues of newly privatized state enterprises, sending their values soaring. It also poured into Mexican government bonds which had been bolstered by a peso-dollar exchange rate guarantee. The surge of capital into Mexico prompted a huge increase in imports unmatched by exports, and with it a succession of current account deficits rising to $29 billion in 1994. It also kept the peso high and stable, a key element of the government's inflation control strategy. The high peso and growing external gap reflected the effect of growing import competition on an already precarious domestic sector.

NAFTA, while putting in place a legal framework conducive to the inflow of foreign capital, also removed any capacity to control capital outflow. And flow out it did, with growing magnitude in 1994, nervous in the wake of peasant uprisings and political assassinations, wary of the ballooning trade gap, and lured out by rising US interest rates.

The fragile underpinnings of the Mexican growth strategy quickly unravelled at the end of 1994. Within weeks the peso lost nearly half its value and $28 billion fled the country. The contraction of the economy was deeper than anything Mexico had experienced since the 1930s. (The consequences for Mexican employment and wages have been detailed by Gutierrez in this volume.) The peso crisis prompted a quick response by the United States Administration which put together with the help of the IMF, the World Bank and various central banks, an unprecedented $50 billion bailout package.

The crisis shattered what has been called the NAFTA effect, the widely held perception both inside and outside Mexico that the creation of NAFTA-type environments was the way for countries to embark on a sustained growth path, enabling the prolonged restructuring to finally bear fruit (see Morales 1997). Despite the collapse of the euphoria around NAFTA, the official line in Washington and Mexico City denied any connection between the financial crisis and NAFTA, except to say that without the Agreement the crisis would have been worse.

Both the United States and Mexican governments had much at stake in protecting the credibility of the NAFTA option. The extension of NAFTA-type arrangements throughout the hemisphere is the centerpiece of US foreign policy in the area, and a key strategic goal for corporate America. To acknowledge a connection would be to undermine the credibility of the NAFTA model as a viable solution to their economic problems.⁽⁴⁾ For its part, politically destabilized and ever more deeply indebted ($173 billion in 1995), the Mexican government, dependent on US political and financial support, also denied the link between the crisis and an Agreement for which it had gambled so much of its political credibility.

The huge financial support package assembled by the United States government stemmed the haemorrhage and prevented a default on its external debt payments. Observers, citing the benefits of NAFTA, noted that the collapse of Mexican imports from the United States (and Canada) during the crisis was proportionately much less than from non NAFTA countries. This is not surprising given an export sector that had come to rely so heavily on imported US inputs and as such, was greatly insulated from internal instability. Mexican exports to the United States and Canada grew far more than to non-NAFTA countries. The subsequent inflow of direct investment was also to be expected in light of plummeting labour costs and the new commitment to deregulate the banking, communications and petrochemical sectors imposed as a condition of the bailout.

5. Domestic mechanisms affecting employment, incomes and standards

These mechanisms interact with and reinforce the NAFTA mechanisms affecting employment, incomes and standards.

The mechanism of choice in Canada for disciplining wages has been high unemployment maintained primarily through tight monetary policy⁽⁵⁾. Formal wage and price controls were used in the 1970s to deal with the stagflation problems of that era. Since then, explicit wage controls along with the suspension of collective bargaining rights have been confined to the public sector.

Monetary austerity from the beginning of the 1980s steadily forced up unemployment into the 9-10 per cent range where it has remained throughout the 1990s. As Jackson argues in this project, extreme monetary contraction was deemed necessary by policy makers in Ottawa to force a compression of wage growth which had gotten out line with wage growth in the United States, and was deemed detrimental to Canada's long-term competitiveness in a free trade environment.

Public sector cuts through the mechanism of the "war on the deficit" have also played also a key role in disciplining wages, particularly during 1994-97 This has occurred primarily through the erosion of the non-wage income system (social wage), especially unemployment insurance, welfare and pensions. This increased dependence on wage income at time of profound restructuring and a harsh labour market, serves as an effective restraint on wage pressure in an environment where international competitiveness preoccupations reign supreme.

Canadian unions maintained their numbers throughout the 1980s and so far in the 1990s in contrast to their US counterparts whose numbers have fallen sharply. During 1988-1993, overall union density has been stable at 32.6 per cent of the labour force, although in the manufacturing sector unionization fell from 35 per cent to 33.4 per cent a period of rapid restructuring (see Jackson). The presence of union friendly social democratic governments in several provinces has helped. More recently, hostile governments have attacked labour legislation in several provinces making it more difficult to organize and to bargain effectively.

In the United States, internal mechanisms for constraining wages and other labour costs go back to changes to the National Labour Relations Act in the early post-war period which weakened the ability of unions to organize and bargain collectively. Particularly damaging were exemptions from the rand formula permitting workers in the southern states to opt out of union membership. These so-called right to work states were characterized by low wages compared with northern states and very low levels of unionization. The existence of this union free zone has had a chilling effect on wages and labour rights nationally and has served as a testing ground for both governments and companies' employment relations nationally.

The US tariff exemptions which enabled the creation of the Mexican maquiladora programme in 1965 provided another mechanism for companies wanting to escape higher wages and hassle of a unionized work force. As we shall see in the next section, the maquiladora has evolved into a effective vehicle for disciplining labour in the high wage areas of North America as well as within Mexico itself.

Fractious relations between US employers and unions in an atmosphere of growing international competition in the 1970s and 1980s spawned aggressive union busting and union avoidance campaigns. They were assisted by the anti-union rhetoric and measures of the Reagan and Bush administrations and a weakly enforced National Labour Relations Act. The consequence was a further decline in union density (from 25 per cent in 1979 to 15.5 per cent in 1995), a key factor driving the decline of American wages and growing inequality since the late-1970s. (Mishel et. al. 1997:199.)

The deterioration of labour market conditions, notably the growth of non-standard work has also played a role in constraining wages and increasing inequality. For example, the growth of non-standard work, has greatly reduced the proportion of workers with adequate medical coverage. Medical care, unlike Canada where coverage is universal, is a workplace benefit in the United States. In 1993, 58 per cent of the United States population had employer-sponsored health insurance, down from 66 per cent in 1980. Health coverage for workers with a college degree dropped from 79 per cent to 73 per cent during 1979-93. Coverage for workers with less than high school dropped from 52 per cent to 36 per cent during the same period. One half of private sector wage and salary workers working less than 20 hours per week did not have a health plan; one-quarter working between 20 and 43 hours did not have a health plan. (NAALC, 1997) One-third of full-time workers in small companies (less than 100) did not have a health plan. Fear of losing health insurance has clearly contributed to the general employment insecurity and had a chilling effect on wage demands.

The United States minimum wage has, like Mexico's, consistently lagged behind inflation. The 1995 US minimum wage was 13 per cent below its 1984 level. (NAALC, 1997) Proportionately more women in all three countries earned minimum wage or less.

The United States unemployment insurance system and adjustment and training programmes were greatly cut back during the 1980s, as were welfare and anti-poverty programmes. (Recent modest improvements in the minimum wage and unemployment insurance, and the NAFTA adjustment programme have been overshadowed by further cuts to welfare programmes.) Changes in the tax system under Reagan greatly reduced the burden on the top 10 per cent and especially the top 1 per cent of income earners, weakening fiscal capacity and accentuating income inequality.

Monetary policy has also been an important device for disciplining wages. However, the more advanced decline of labour market institutions and social supports has made the need for monetary policy less compelling than in Canada.

Domestic mechanisms of wage control in Mexico (beyond the policy of apertura and NAFTA) start from the fact that the unions are part of the governing party, the PRI. The leadership of the CTM, the official union confederation, are also PRI politicians and members of the government. Thus, the annual tripartite solidarity agreements (Pactos Sociales) between government, business and labour which have been in place since 1987 have been a very effective vehicle for holding down wage increases below inflation and productivity growth. So has the national tripartite Minimum Wage Commission which, between 1982 and 1988, reduced the real minimum wage by over 70 per cent (US-Office of Technology Assessment 1992:81). The 1995 Mexican minimum wage was 51 per cent below its 1984 level (NAALC, 1997). Real average hourly wages in 1994 had fallen 30 per cent below their 1980 levels, by mid-1996 they had dropped another 25 per cent (Sheiken, 1997).

Unions have been an important government policy tool for controlling worker demands. Furthermore, tripartite labour regulatory bodies have effectively blocked the formation of independent unions and thus more effective worker representation. Practices where companies sign contracts with union agents without the knowledge of workers to keep any union presence out of their workplaces, are commonplace.

The maquiladora has several additional devices for controlling wages besides those already mentioned including: collusion between employers and CTM officials to keep workplaces union-free; employer collusion to set wages; segmentation of skills such that the vast majority of workers are placed in the unskilled category; and a preference for young women who have tended to be more compliant and less militant in terms of workplace demands. (J. Carillo 1990, cited in Kopinak, 1996: 13-14). Lower maquila wages exercise a drag on wages in the rest of the manufacturing sector, and as maquila-like production expands it increasingly sets the tone for employment relations throughout the economy.

The lack of unemployment insurance in an atmosphere of a rapidly growing labour force and few formal sector jobs is a powerful deterrent to worker militancy. Moreover, policies of fiscal austerity and privatization have greatly cut back the Mexican system of social protection (medical care, housing, food subsidies education, pensions) since 1982, reducing both the proportion of workers covered and the extent of coverage.

As with the other NAFTA countries, nonstandard work has grown apace with competitiveness pressures and capital mobility, leaving fewer workers with access to benefits. The huge labour surplus has increasingly been pushed into the informal sector, particularly into self-employment. Although open unemployment has doubled since 1993, complementary measures show the real level of un/underemployment to be much higher (see Gutierrez in this volume).

In Mexico 34 per cent of the labour force received workplace benefits in 1995, down from 39 per cent in 1991, due mainly to the rise in non-standard and unpaid work and the decline of the public sector. In 1993 63 per cent of salaried workers received social security benefits down from 66 per cent in 1991. Among the self-employed, small employers etc., only 2 per cent received employment benefits.

6. The maquiladora

The maquiladora industrial programme was established in 1965 to spur industrialization in the Mexican border region with the United States, though it was soon expanded to include non-border areas as well. The programme took advantage of the United States tariff exemptions which allowed US components to be assembled abroad and re-exported back to the United States duty-free. It's importance as a deregulated export zone, stems not only from the profound effect it came to have on wages and jobs both within Mexico and north of the border, but also as a precursor of the approach taken in NAFTA. Ironically, a much different approach was taken in an agreement signed the same year, the Canada-US Auto Pact, a managed trade agreement which contained production and employment guarantees.

In the early years maquila plants did exclusively routine assembly using unsophisticated and obsolete equipment with a preference for hiring young women. During the 1970s maquila plants were granted exemptions from the rules limiting foreign ownership and exemptions from Mexican labour laws; for example, lengthening the probation period during which employer did not have to pay minimum wage, allowing dismissals without having to pay severance, etc. The maquiladora grew, if unevenly, throughout the 1970s experiencing declines in periods of recession in the United States. It was certainly not perceived as a potential replacement to Mexico's dominant form of import-substituting industrial development. By 1981 it employed 131,000 and generated a considerable amount of foreign exchange.

Under pressure from the IMF, The World Bank and the United States government to move to a deregulated, export-oriented model of economic development (post-1982) the maquiladora became a more important focus for Mexican policy makers. Electrical and electronic equipment, machinery and automotive sectors became priority areas for maquiladora production. The massive devaluation of the peso sharply reduced the cost of Mexican labour making it highly attractive to transnational capital.

While the domestic economy stagnated and businesses failed in large numbers throughout most of the 1980s, the maquiladora sector grew quickly. Also, a growing number of foreign-owned transnational plants, not originally set up under maquila rules, reoriented their production from the depressed domestic market toward exports, becoming, in effect, de facto maquiladoras. US auto companies, the main example, exported back to the United States under the reduced tariff provisions its general system of preferences (GSP).

We discussed earlier how the characteristics of employment relations in the maquiladora kept wages down and have increasingly come to shape employment relations throughout the economy. Since 1975, wages in the "maquilized" industry have been about one-half of wages in the non-maquila manufacturing sector. (Gambril 1994, cited in Kopinak, 1996:14.)

The expansion of more capital intensive production and more flexible work methods in the 1980s have been accompanied by a reduction in wages. Carrillo's study of auto plants (cited Kopinak, 150) showed that when a company converted to export production and transferred operations north to the border area, workers' wages decreased even as advanced technology was introduced. Workers in border plants received on average 60 per cent less than their colleagues making finished products near Mexico city. An unskilled worker in the "old" automotive sector made the same wage as a skilled maintenance mechanic at the border. Carrillo observed a homogenization of wages in all maquila sectors regardless of the technology used.

The position of women in the maquiladora labour market was not improved by the changes in the 1980s. Women were much more likely to have unskilled jobs. Few were able to gain entry into the new supervisory or technician jobs that became available. (Kopinak 1996: 183.)

By the end of the 1980s, exemptions which had been granted to the maquila sector in the previous decade, were extended to the entire economy. For example, foreign ownership limits were removed for all sectors except mining, petrochemicals auto parts and communications. The maquiladora sector which had once been the exception, had become the rule in terms of Mexican industrial strategy. It was the vehicle through which Mexico would become integrated into the global economy. The domestic component of value added in the non-maquiladora export sector dropped steadily from 90 per cent in 1980 to 39 per cent in 1994. (Dillon 1994:74.)

The 1989 government Decree on the Maquiladoras prepared the ground work for the NAFTA that was to follow, creating in microcosm the rules that would later be generalized to the whole economy. Firms that were not legally maquiladoras could shift to export production if they had idle capacity, creating part-time maquilas. The legislation was broadened to include agro-industry, mining, fishing and forestry. This blurring of the lines between the export and domestic production would further facilitate the maquilization of the economy.

NAFTA extended the maquila rules of duty free import and re-export (with duty only on the value-added-almost entirely labour) to the whole economy. By 2001, all duties will be removed. Companies that were able under maquila rules to import components and materials from third countries for maquila assembly and export, duty free, to the United States will no longer be able to do so unless they meet the NAFTA rules of origin, or content requirements. As a result many of these companies are shifting their production of components from other locations, mainly Asia, to North America, mainly Mexico. NAFTA increases in stages the ability of maquila companies to sell in the domestic market, removing all restrictions by 2001 when the name maquila will officially be dropped and become part of Mexico's domestic manufacturing industry. Non maquila companies will have virtually no protection and will either have to adapt their operations or fail in the new maquilized environment. Survival will likely entail entering into some kind of an alliance with foreign capital for access to technology and intermediate inputs.

Historically, the maquiladora has been delinked from the domestic Mexican economy, sourcing less than 2 per cent of its non-labour value-added from within the domestic economy. This applies as well to nationally-owned maquilas which are generally sub-contractors to transnationals, providing the labour, the premises and the administration while the transnational provides the capital, machinery, parts, components and supervision. (Kopinak, 1996, 188). There is evidence that under NAFTA the tie to the domestic economy has become weaker still, falling from its historic average of 1.73 per cent to 1.45 per cent ( Dillon, 1997:74.)

Despite the widely held perception of chronic labour shortage throughout the maquila zone, wages have in fact fallen steadily throughout the 1980s and 1990s. The reason according to Kopinak, (146) and others, is employers who collude to set the terms of employment, unions which don't adequately represent workers, and the government which determines what constitutes a fair day's pay through the minimum wage. This labour market control to keep wages low was pioneered in the maquiladora and is key to the maquilization of the Mexican economy.

Perceived labour shortages have been the result of employers unwillingness to pay higher wages. The absence of effective unions to bargain collectively on their behalf has according to Kopinak (1996, 194-95), led workers to pursue individual and household strategies to maximize their income notably by having more members working. The maquila labour market is heavily tilted against workers. High turnover rates are indicative of workers' frustration and the price employers are willing to pay to keep wages from rising.

The 1994-95 crisis brought another collapse in wages prompting a new wave of expansion and a new migration north to the border region and across to the United States. The jump in illegal arrests prompted renewed toughening of enforcement measures by US officials and measures such as proposition 187 in California to discourage immigration by denying any form of state assistance.

The peso devaluation aided the maquila producers (who conduct their operations in US dollars) by lowering their labour costs. The enclave nature of their production insulated them from the internal turmoil of the high inflation, high interest rates and collapse of internal demand. It accelerated the foreign takeover of Mexican business, and the conversion of Mexican-owned business itself to maquiladora status.

Fuelled by the NAFTA and the peso collapse the maquiladora entered another boom period. Employment jumped by 407,000 or 75 per cent since NAFTA's implementation, from 542,074 in 1993 to 948,658 by October 1997 (INEGI). In 1981, the year before the Mexican debt crisis, there were 131,000 employed in the maquila sector. By 1988 employment had grown to 369,489. Since then another 579,000 workers have been added to the ranks of the maquiladora work force. In 1982, maquila employment made up 5 per cent of manufacturing employment. By 1995 it had surpassed non-maquiladora manufacturing employment. (see Gutierrez.)

7. Economic opening under CUFTA/ NAFTA

Both Mexico and Canada have undergone fundamental opening of their economies since 1990. Mexico's trade (exports plus imports) doubled as a proportion of GDP from 24 per cent to 48 per cent during 1990-95. Canada's trade rose from 58 per cent to 71 per cent of GDP during this period, see Table 1. The United States, whose trade openness was by far the lowest, saw a moderate increase in trade openness, from 21 per cent to 24 per cent of GDP during this period. (NAALC,1997.)

Both Canada and Mexico were already very trade-dependent on the United States prior to CUFTA/NAFTA, Nevertheless, the share of Canadian exports going to the United States rose sharply, from 71 per cent in 1989 to over 80 per cent in 1996. Imports from the United States, although less-concentrated, also rose significantly, from 64 per cent to 67 per cent in 1995 (IMF).The most open economy to begin with, Canada's exports of goods and services as share of GDP, rose steadily from 28 per cent in 1989 to 42 per cent in 1995 (NAALC,1997). Imports rose at a similar pace in relation to GDP, from 31 per cent to 42 per cent indicating a major reorientation of economic activity. Exports of goods and services grew twice as fast as overall GDP growth during the 1980s and five times as fast GDP during the first half of the 1990s. Foreign investment flows also increased sharply in the free trade era. Canada, the industrialized country with by far the highest level of foreign ownership, saw inflows of FDI rise from $US 4.2 billion per year during 1983-88 to $6.2 billion annually during 1989-96. Outward flows of Canadian-owned FDI also grew rapidly, from $4.3 billion per during 1983-88 to $5.6 billion per year during 1989-96. Inflows and outflows of portfolio capital also grew rapidly. Inflows rose from $11.5 billion per year during 1983-88 to $21.2 billion per year during 1989-96; and outflows rose from $1.5 billion per year to $6.7 billion per year between these two time periods.

Table 1. Macro-economic and trade indicators

Mexico's exports to the United States, like Canada's, became more concentrated, growing from 70.4 per cent of total exports in 1990 as NAFTA negotiations began, to 82.9 per cent in 1993 and continuing to rise to 83.7 per cent in 1995. Import concentration from the United States also rose, though from a smaller base and at a less rapid rate. Mexico's exports as share of GDP did not change much during the pre-NAFTA period 1989-93), fluctuating between 18 per cent and 19 per cent of GDP. Although, they had risen during the early years of the trade liberalization - from 15 per cent in 1985. (NAALC, 1997) However, exports rose sharply to 27 per cent of GDP in 1995, reflecting the structural jolt in the economy. Imports on the other hand, rose in the pre-NAFTA period - from 12 per cent to 18 per cent of GDP-reflecting the deepening of trade and investment liberalization and high peso monetary policy during these years. (Imports rose during the first stage of trade liberalization, from 10 per cent of GDP in 1985 to 12 per cent in 1989.) Imports fell in the wake of the 1995 crisis to 15 per cent, but remained substantially above their pre-1989 level, reflecting to the increased dependence of Mexican manufactured exports on imported inputs.

Comparing the growth of the Mexican economy overall to growth of the export sector, GDP in both the 1980s and 1990s grew at an average annual rate of just 1 per cent. (Growth picked up moderately in the 1990s until 1995 when it contracted by 6.2 per cent.) However, throughout the 1980s and 1990s annual exports of goods and services grew rapidly - 6.6 per cent and 6.8 per cent respectively. (World Bank, 1997.)

The inflow of foreign direct investment into Mexico also increased dramatically during the 1990s, the annual inflow rising from $US1.8 billion during 1983-88, to $6.2 billion per year during 1989-96. Portfolio inflows also increased massively, from a negative $-0.3 per year during 1983-88, to an average $7.7 billion per year during 1989-96. (Of course, these more volatile and speculative flows resulted in a sudden $10 billion outflow in 1995.)

Mexico used its oil wealth to borrow heavily in international markets in the late 1970s. In 1980, shortly before the debt crisis, it had $58 billion worth of external debt. Fifteen years later its external debt had soared to $166 billion and more than doubled from 31 per cent to 70 per cent of GDP. However Mexico was a much more export-oriented economy and the ratio of external debt to exports had declined from 232 per cent to 171 per cent. The ratio of debt servicing costs to exports fell in half, from 44 per cent to 24 per cent. (World Bank, 1997.)

The transformation of the Mexican economy driven by trade and investment liberalization stems from the policies implemented in the wake of the 1982 crisis, a decade before NAFTA came into effect. A key structural change associated with this economic opening was the growth in manufactured exports to the United States using imported inputs or intermediate goods from the United States⁽⁶⁾. While manufactured exports rose from 13 per cent of total exports in 1982 to 84 per cent in 1995, intermediate goods imports rose from 53 per cent to over 80 per cent of total imports in 1995. (Hinojosa et al., 1996: 39.)

The changes were driven by the crisis and the IMF-mandated structural adjustment programme, agreed to and implemented by Mexican policy makers. But they were also driven by the fact that US transnationals had decided to make Mexico an integral of their global competitiveness strategies, as a low labour cost export platform close to the United States market.

The change in the structural relationship between imports and exports was evident during the 1995 crisis when intermediate goods imports from the United States fell only momentarily before resuming their growth in line with manufactured exports. Consumer and capital goods imports from the United States the other hand experienced a deeper and prolonged compression after the peso collapse. In the import substitution era, intermediate goods imports used for domestic production fell to a much greater extent during the 1982 crisis. (Hinojosa, 1996, 38.)

Thus, the demand for Mexican exports (i.e. US demand) is now the key determinant of Mexican imports rather than changes in domestic demand. This, according to Hinojosa, is the main structural difference between two Mexican financial crises. Moreover, this new import-export relationship is growing even faster than the growth of maquila exports "as this strategy of manufacturing for export is adopted by many other regions, sectors and types of firms in the Mexican economy." (ibid, 9) seeking insulation from the demand weakness of the domestic Mexican economy.

In 1986 Mexico joined the GATT and by 1988 the average tariff had dropped to 10 per cent from 25 per cent⁽⁷⁾. At about the same time Mexican policy-makers began to deregulate financial and investment flows. During 1989-94 foreign capital inflows supported a high peso and expansion of imports accompanied by widening trade and current account deficits. NAFTA was seen by policy makers as a tool for enlarging and stabilizing these foreign capital inflows [Hinojosa, 1996:36]. However, this time the financial inflows were largely portfolio capital - equity and bonds rather than bank loans or direct investment. Greater openness has meant greater dependence on international financial markets.

The United States economy, the world's largest, was only moderately dependent on Canada its largest trade partner - 21.5 per cent at the beginning of CUFTA. By 1995 it had barely moved, to 21.6 per cent. Slightly lower dependence on Canadian imports, 18.2 per cent of total imports in 1989, rose slightly to 19.2 per cent in 1995. (IMF) US export dependence on Mexico, its third largest trade partner, also low, increased significantly from 6.9 per cent to 9.0 per cent of total exports during 1989-93, falling back to 7.8 per cent in 1995. US imports from Mexico as a share of its total imports rose steadily throughout the pre and post NAFTA periods - from 5.6 per cent to 6.8 per cent in 1993 to 8.1 per cent in 1995.

The United States economy is far less open than those of Mexico and Canada. There was, nevertheless, a steady overall trade opening, exports rising from 7.2 per cent to 12.4 per cent of GDP during 1984-94, falling back moderately to 11.5 per cent in 1995. Imports also rose steadily during this period, from a larger base of 10.3 per cent of GDP in 1984 to 14.4 per cent of GDP in 1994, falling back to 13.1 per cent in 1995. (NAALC,1997.)

Comparing US export growth to overall economic growth, exports grew during the 1980s at an annual rate of 5.2 per cent, substantially above the annual 3 per cent rate of GDP growth. During the first half of the 1990s the divergence widened - with annual growth slowing to 2.6 per cent and exports rising at an annual rate of 7.3 per cent. (World Bank, 1997).

There was also a large increase in inward FDI into the United States during 1989-96 ($49 billion per year) compared to the 1983-88 period ($34.4 billion per year). More striking, however, was the rise in outward flows of US-owned FDI, from $16.6 billion per year during 1983-88 to $57.2 billion per year during 1989-96. Both inflows and outflows of portfolio investment also increased rapidly. The former jumped from $ 56.8 billion to 139.1 billion per year, and outward flows soared, from $6.1 billion per year to $69.5 billion per year between these two time periods.

7.1 Trade and manufacturing employment
As economic liberalization and integration have proceeded under NAFTA, manufacturing sector employment has fallen in all three countries-not only in relation to the rest of the economy but also in absolute numbers. Canada has been hardest hit, with a contraction of 255,000 or 12.8 per cent during 1988-96 (table 2), more than three times the United States decline of 802,000 or 3.8 per cent during this period (see tables 2 and 3)⁽⁸⁾. The first phase 1988-93, saw the biggest shrinkage, 375,000 or 18.8 per cent in Canada and 1,763,000 or 8.3 per cent in the United States. During this stage there was a recession in both countries, although more severe and prolonged in Canada due to deeper restructuring and to a tighter monetary policy. In phase II (1993-96) there was a partial recovery of employment in most sectors in both countries.

In Mexico the manufacturing sector as a whole declined only slightly, from 1.41 million to 1.37 million during 1990-1995. But this masked a profound structural shift. Maquiladora sector employment grew by 433,000 or 117 per cent during 1988-96 while the non-maquila sector employment fell - per cent during the same period. It coincided with a rapid growth of manufactures exports, mainly to the United States. Manufactures grew from 18.6 per cent of total exports in 1985, the beginning of the apertura, to 37.5 per cent in 1988. By 1995, manufactured exports had grown to $80 billion from $38 billion in 1988, increasing its share to 83.7 per cent of the total. Intermediate imports grew at a similar pace, reflecting the high import content of exports.

In the transportation equipment sector (mainly automotive), which accounts for about 30 per cent of NAFTA trade Canadian employment was steady (-0.4 per cent), with some fluctuation, during 1988-96, while US employment fell by 300,000 or 12.7 per cent. The maquila auto sector grew by 86,000 or 116 per cent during this period. Electrical/electronics sector employment in Canada fell by 40,000 or 25 per cent during 1988-96, and in the United States it contracted by 133,000 or 6.5 per cent. Maquiladora employment in the electronics sector (the largest) grew by 94,600 or 98 per cent during this period. Textile and clothing employment in Canada contracted by 47,000 or 26 per cent during 1988-96, and by 323,000 or 17 per cent in the United States.

Table 2. Canada: Manufacturing employment

Table 3. United States: Manufacturing employment

Both Canada and Mexico (post 1994) have seen their merchandise trade surplus with the United States increase under free trade, though their current account surplus is much smaller because of the large net outflow of interest and dividend payments to the United States. The relationship between changes in trade balances, however, and employment changes is not straightforward. An improvement in trade balances does not necessarily mean an increase in employment. Employment also depends on a variety of interrelated factors all of which are affected by integration under NAFTA. These include: changes in trade and investment with non-NAFTA countries; macroeconomic policy (fiscal and monetary) and performance (including output, income and aggregate demand); nature and changes in the structure of wages (eg., the displacement of high income jobs and replacement with lower income jobs, or the decline in union bargaining power); cross-border movement of capital (eg. production facilities); the level and changes in import content of exports (affected by, for example, the ownership structure); changes in the labour intensity of exports and imports including its sectoral composition; level and changes in intra-firm trade (and accompanying transfer pricing).⁽⁹⁾

By way of example, we looked at the changes in trade balances between Canada and the United States in the 21 major manufacturing categories (two-digit SIC codes) during 1998-96 and compared these to Canadian employment changes in these sectors [table 5]. Bear in mind that the export share of manufacturing output in Canada rose from 40 per cent to 60 per cent during this period.

8. Comparative portrait of NAFTA economies and labour markets⁽¹⁰⁾

The United States was the world's largest economy in 1995, 12 times larger than the Canadian economy, and 27 times larger than the Mexican economy. Canada's GDP per capita was four-fifth's of the United States level. Mexico trailed far behind at just over one-quarter of the United States level and just over one-third of Canada's per capita GDP (NAALC 1997).⁽¹¹⁾

The Mexican economy grew hardly at all during the 1980s and mid-1990s, and with population growing rapidly, GDP per capita fell 14 per cent during the 1980s and another 6 per cent during the first half of the 1990s. This is a staggering collapse of living standards, especially when compared with the three previous decades when growth averaged close to 7 per cent per year and per capita GDP grew almost 4 per cent per year.

The Canadian economy slowed considerably in the 1980s compared with the post-war golden age. Nevertheless, with a slower growing population, GDP per person still grew 24 per cent. The situation changed markedly in the free trade era as the economy stagnated and GDP per capita actually declined 2.8 per cent during 1990-95. [Personal income per capita declined even more rapidly - 5.1 per cent during 1989-96.] The United States economy maintained a steady if slower growth path in the 1980s declining somewhat during the 1990s. Consequently, per capita GDP grew 20 per cent during the 1980-90 and 6.2 per cent during 1990-95. That growth, as we shall see later, has been more unevenly distributed.

8.1 Employment and unemployment
In 1995 the United States accounted for 73 per cent of the North American labour force, with Mexico holding 19 per cent and Canada 8 per cent (table 4). The Mexican labour force grew 48 per cent, almost three times as fast as the United States and Canada labour force during 1984-95, the result of higher population growth in previous decades as well as the faster entry of women into the labour force.

The United States and Canada had similar labour force participation in 1995 (66.7 per cent and 64.8 per cent respectively) while Mexico's was lower (59.6 per cent) due to the lower Mexican female participation rate (37 per cent) compared to the Canadian (58 per cent) and US (59 per cent) participation rates. Mexico had higher a male participation rate (84 per cent) than either Canada (73 per cent) or the United States (75 per cent) and the male participation rate increased during 1984-95 compared to the United States and Canada whose rates dropped.

Mexico had proportionally almost twice as many youth (15-24 years) participating in the labour force (29 per cent) as either the United States (14 per cent) or Canada (17 per cent) in 1995. The United States had the highest participation rate of older workers (55-64 years) in the work force: 57 per cent in 1995, up from 54 per cent in 1984. Mexico's rate was 53 per cent, up from 41 per cent in 1984. Canada had the lowest older worker participation at 48 per cent, down from 52 per cent in 1984. This reflects the greater need to work longer due to insufficient retirement income in the United States and especially in Mexico. The Canadian rate reflects higher and growing early retirement driven in part by fewer job opportunities.

Mexico, has a large informal employment sector (estimated at 38 per cent of total employment in 1988). The informal sector is a subsistence sector characterized by very low earnings, no benefits and a high degree of precariousness. It includes household domestic work, employees, employers and piece workers in establishments with fewer than five workers, (excluding formal sector self-employed as well as professional and unpaid workers). It has grown steadily as job opportunities in the formal economy have been unable to absorb the rapidly growing labour force. The need to survive makes unemployment an unaffordable luxury for most.

The high proportion of employment in the primary sector also distinguishes the Mexican economy from the US and Canadian economies: 24 per cent compared to 3 per cent and 4 per cent respectively. The service sector in Mexico is proportionately smaller, accounting for 54 per cent of employment, compared with 74 per cent in both Canada and the United States. The vast majority of employment growth in all three countries during 1984-95 was in the services sector - 96 per cent in Canada, 98 per cent in the United States, 91 per cent in Mexico (1991-95). Manufacturing accounted for similar share of employment in all countries in 1995: 15.3 per cent in Canada, 15.8 per cent in Mexico, and 16.4 per cent in the United States.

Non standard work - part-time, temporary, self employment and multiple job holding  - has grown steadily in all three countries during 1984-95 despite some cyclical fluctuation. The proportion of part-time workers has grown in all three countries; up in Canada from 14.8 per cent to 18.6 per cent of the employed work force during 1984-95; up in the United States from 17.6 per cent to 18.5 per cent; and up in Mexico from 25.6 per cent to 26.6 per cent (1991-95). The involuntary part-time work force grew fastest in Canada and actually fell in the United States, reflecting the lower US unemployment rate.

Table 4. Labour Market Indicators

In Canada and the United States, the share of part-time female workers has remained the same but it has grown among men. In Mexico the female share of part-time workers has grown. A greater share of women hold part-time jobs involuntarily in all three countries, although the difference is most pronounced in Canada.

Self-employment is the main form of non-standard work in Mexico at 26 per cent of the labour force, down slightly from 1991. In Canada the figure is 9 per cent, and in the United States, 6.6. per cent. In all countries, most of this work is in the service sector although in Mexico also has a significant share in agriculture. There is a small but growing share of multiple job holders in all three countries reaching 5.9 per cent in the United States, 4.9 per cent per cent in Canada, and 6.9 per cent in Mexico in 1995. Temporary work is hardest to compare because of data limitations and methodological differences. The share of temporary workers in Canada grew from 3.5 per cent to 9.0 per cent during 1984-94. In Mexico temporary workers made up 4.7 per cent of the work force in 1995. In the United States temporary workers made up 2.2 per cent of employment in 1995 (although the figure rises to 4.9 per cent using a broader definition of temporary employment. The United States has the highest incidence of short-tenure jobs; In 1991, 29 per cent of all jobs lasted less than one year, compared with 22 per cent in Canada and 25 per cent in Mexico (1993).

Education levels of workers in all three countries have grown. Higher educated, higher skilled workers in all three countries tend to have more stable, higher paying jobs. Less skilled, less educated workers have higher unemployment and underemployment rates, and less access to training. In 1994, 43 per cent of the employed work force in the United States and Canada, and 81 per cent in Mexico, had high school education or less. The proportion of workers with at least some post-secondary education was 57 per cent in Canada, up from 38 per cent in 1984; and 56 per cent in the United States, up from 44 per cent in 1984.

The official unemployment rate in Canada has been high and despite some cyclical fluctuation has grown throughout 1980s and 1990s averaging 10.2 per cent during 1990-95, up from 9.3 per cent during the 1980s. US unemployment declined through the 1990s and currently stands at 5.3 per cent. Unemployment averaged 6.3 per cent during the first half of the decade lower than the 1980s. The gap between Canadian and US unemployment which tracked the United States rate very closely from 1950 to 1980, opened up a two point gap after the 1980s recession. It widened again to 4 points in the 1990s. Mexico's open unemployment has been low throughout averaging 2.8 per cent during 1990-95, but doubled in 1995 and 1996.

Unemployment rates were higher for less educated workers in both the United States and Canada, although in Mexico unemployment for workers with less than 6 years of education was less than for those with some high school or post secondary education. This probably reflects the unaffordability of unemployment.

In all three countries young workers have had much higher unemployment rates, (more than double) than the general rate and a much higher share of jobs lasting less than a year. During the 1990s Canadian and US, unemployment rates for men were higher than for women. In Mexico they were slightly higher for women.

8.2 Earnings, productivity, and income distribution
During 1990-96 hourly earnings (wages and benefits) of production workers rose 1 per cent in Canada annually while declining 0.8 per cent annually in the United States and 2.5 per cent annually in Mexico. By contrast, labour productivity during 1990-96, grew 2.2 per cent per year in Canada, 3 per cent in the United States, and 7.5 per cent annually in Mexico. During 1985-90, hourly earnings of production workers in Canada did not grow at all, fell 11.7 per cent per year in Mexico, and barely increased at the rate of 0.3 per cent per year in the United States. Productivity on the other hand, grew 0.8 per cent annually in Canada, 2.8 per cent in the United States and 4.9 per cent per year in Mexico.

In US dollars, Canadian average weekly earnings for all employees rose until 1991, closing the gap with the United States average earnings. Thereafter they fell because of the Canadian dollar decline, widening the gap again to 13 per cent below the United States level. In US dollar terms, Mexico's average weekly earnings for wage and salaried workers in the private formal sector were in 1984 14 per cent of US levels and 15 per cent of Canadian levels. In 1995 they were even further behind at 13 per cent of US levels.

Throughout the region higher earnings were associated with higher levels of education. And gender differences in earnings narrowed at higher levels of education. In 1994, Canadian median weekly earnings of full-time male workers with 8 years of education or less, were 43 per cent less than those of workers with a university degree. The gender wage gap narrowed from 64 per cent of male earnings in the first category to 79 per cent of male earnings in the latter category.

In 1995, median earnings for US workers with less than high school were 55 per cent less than of those with a university degree. Women in the first category had earnings 25 per cent less than men in that category. The gap narrowed to 15 per cent in the high education category. Earnings of high education workers in both the United States and Canada grew at a higher rate than those at the low end. This was more marked in the United States where earnings at the low end grew 2.3 per cent per annum during 1984-91 whereas at the high end they grew 4.6 per cent per annum.

During the 1980s and 1990s income inequality after taxes and transfers increased in both Mexico and the United States; and it remained stable in Canada until the early 1990s, but has since shown signs of widening largely as a result of cuts to social transfers.

The gap between the top and the bottom quintile in Canada narrowed slightly but increased in relation to the middle three quintiles. In both the United States and Mexico the top quintile increased relative to the bottom 4 quintiles. In Canada (1993) the top quintile received 41 per cent of household income. Average income in the top quintile was 7.3 times the average in the lowest quintile. In Mexico (1994) the top quintile received 54.2 per cent of aggregate household income. The average income in the top quintile was 13 times the average income in the bottom quintile. In the United States (1994) the top quintile received 47 per cent of total income and the average income in the top quintile was 13 times the average income in the bottom quintile.

In both the United States and Canada, taxes and transfers increased the share of income for the bottom three quintiles and reduced it for the top quintile, with transfers in both countries having a greater impact in lowering inequality than taxes. Inequality in Canada (1993) as measured by the gini coefficient was reduced 30 per cent by taxes and transfers - taxes reducing inequality by 9.5 per cent and transfers by 22 per cent). In the United States, inequality was reduced 15.6 per cent - 4.5 per cent from taxes and 11 per cent from transfers.

During 1984-94, real average household income in Canada was unchanged (though it dropped 4.8 per cent during 1989-95). In Mexico it grew 2.9 per cent per annum because despite the decrease in earnings, the number of earners per household rose. The ratio of the average income in the top quintile to the bottom quintile in Mexico increased from 10 times to 13 times. Average US household income increased 9.5 per cent during 1984-94 and the average income gap between the top quintile and the bottom quintile rose from 12 times to 13 times.

If real wages were either flat or declining throughout the 1980s and 1990s and GDP and productivity were growing, one group that was benefiting in all three countries was senior corporate executives. We track the relationship between CEO compensation (a proxy for senior executives as a group) and earnings of average workers.

The gap in the United States between CEO compensation and average worker wages widened from 60 times in 1978 to 122 times in 1989 to 173 times in 1995 (Pearl, Meyer and Partners Inc. And Wall St. Journal, cited Mishel et. al. 1997:227). During 1989-95 average hourly wages grew 0.7 per cent while executive compensation (inflation adjusted) increased almost 13 times faster (38 per cent). [Mishel 1997: 134 and 227] During the 1980's executive compensation grew at a slightly faster rate but annual wages increases were almost twice as fast as in the 1990s. (ibid.)

The inequality in the average compensation of CEOs of US companies compared to average wages is much higher than in Canada. According to our calculations, the CEO-average wage gap in 1994 was 36 times in Canada. Comparisons over time are not possible since only recently have Canadian CEOs been obliged by law to reveal their compensation. Nevertheless, executive compensation surveys indicate that average increases in the 1990s have been at least as fast and probably faster than their US counterparts. For example, a KPMG survey found average executive compensation increased 32 per cent during 1992-95 (cited Financial Post. 25.9.96) Meanwhile while real average hourly wages increased only 1.4 per cent during the entire period 1989-95. (Jackson). Thus, it is safe to assume that the gap between CEO and average earnings, although much lower, is growing at least as fast as the United States gap.

Even less information is available on Mexican CEOs. A 1994 Towers Perrin Inc. international comparison of CEO compensation showed Mexico ranked fourth (the US was at the top). It moved into first place when adjusted for taxes, social security contributions and cost of living. Given that average formal sector compensation in Mexico is currently 13 per cent of the US average and that it has been falling in real terms, it is safe to say that the Mexican CEO-average earnings gap, regardless of the methodology used, is far higher than the United States differential.

9. TNC restructuring under CUFTA/NAFTA

The bulk of trade in the NAFTA region is conducted by a small number of large, mostly transnational, corporations. According to Rugman (1995) just 50 large corporations, most of them US-owned, account for 70 per cent of US- Canada trade. According to Heredia (cited Grieder, 1997: 273), 300 corporations account for 85 per cent of its exports and imports although we suspect that it is now more concentrated.⁽¹²⁾

There is a small number of Canadian-owned transnationals with investments in all three countries. Large Mexican companies are national in the sense that while they may export to the United States or Canada they have little direct investments in either country. There is also a growing number of East Asian and European transnationals operating in all three countries.

US transnationals at the beginning of the 1990s owned two-thirds of the stock of FDI in both Canada and Mexico (Eden,1994:195). Five years later they had maintained their share of FDI in Canada ($82 billion), though in Mexico their share ($18 billion) had dropped to 60 per cent of the Mexican FDI stock, reflecting the influx of Asian and European transnationals. (SECOFE cited US ITC, 1997, 3-37.)

Canadian FDI in the United States has grown rapidly in last 15 years accounting for 8.2 per cent of the United States FDI stock in 1995 (ITC,1997,3-53). Canadian-owned transnationals' direct investment in Mexico, though small, is also growing. Its share of Mexico's FDI stock grew from 1.4 per cent to 2.5 per cent of the total stock in 1995 (see Guttierrez).

Transnational corporations pursue strategies which combine company-specific advantages with country-specific advantages. Transnationals make direct foreign investments in plant, equipment etc. to gain access to the domestic market, and, in a NAFTA-type setting, both to rationalize production to gain from scale economies and specialization and to exploit cost differences within the region - labour, energy, taxes, government incentives, interest rate, infrastructure, transport etc.

Before Mexico deregulated its foreign investment rules in the 1980s, only the maquiladora sector was open to majority foreign ownership and thus to the significant cross-border integration of production. TNCs entered Mexico in minority-owned arrangements, licensing or contract relationships with Mexican firms for the purpose of gaining access to the domestic market. With trade and investment deregulation, direct investment in majority-owned affiliates and trade between affiliate and parent (intra firm) has grown in tandem. TNC subcontracting or out-sourcing to Mexican firms has also grown.

About 45 per cent of US-Canada trade is intra firm, though the ratio is substantially higher in manufacturing. Mexico-US intra-firm exports are now 55 per cent, up from 44 per cent of total exports prior to NAFTA (Heredia, cited Grieder op. cit., 1997:273), with intra-firm manufacturing trade close to 100 per cent.

In pre-CUFTA Canada, part of the manufacturing sector (one-half of which is foreign-owned) was organized on a continent-wide basis (excluding Mexico). The automotive sector, operating under the rules of the 1965 Canada-US Auto Pact (a managed trade arrangement) was the most important integrated sector with almost all of its trade being intra-firm. Another part of manufacturing (roughly one-third) was so-called branch-plant operations-affiliates producing almost exclusively for the domestic market. CUFTA triggered a major restructuring in this sector with some branch plants converting to continental production and many just closing down and consolidating production south in the United States or Mexico, leaving just warehouse and distribution facilities in Canada. Also, there is much evidence of affiliates moving back administrative functions and even R&D to the parent company.

Cost considerations are key determinants of company location decisions. According to a 1996 KPMG Inc. study comparing business costs in 13 Canadian and 10 United States cities across seven manufacturing sectors, labour costs are by far the most important cost factor, accounting for an average 65 per cent of location-sensitive costs and between 19 per cent and 32 per cent of total operating costs depending on the sector. [3,16.]⁽¹³⁾

TNCs simultaneously pursue both high innovation and low-cost strategies. Investment in technology, the life-blood of the company, is concentrated close to the parent and costs are spread out across many markets as possible to recoup the heavy investment cost. This technology has become highly transferable and its ownership when protected as it is under NAFTA, can be diffused in accordance with country-specific or region-specific cost advantages. The Mexican combination of high productivity, low wages has become very attractive for transnationals.

Let us now look at changes over time in manufacturing labour costs among the three countries as they might be viewed through eyes of transnational corporate investors making continental or global investment location decisions. (US Bureau of Labour Statistics, June 1997). Hourly compensation trends (in US dollars) for production workers in manufacturing between 1980 and 1996 show that Canadian costs rose from 88 per cent of US costs to exceed US costs during the early 1990s, settling back to 94 per cent in 1996 as the Canadian dollar fell, the same ratio as it was in 1975.⁽¹⁴⁾

In 1980, Mexican manufacturing labour costs were 22 per cent of US costs and foreign direct investment outside the maquiladora was limited by law. Costs in Asian exporting countries such as the Republic of Korea, Hong Kong, Taiwan, China and Singapore were one-half or less of Mexican labour costs. Five years later in the wake of the 1982 crisis and the early stages of investment and trade deregulation, labour costs had fallen to 12 per cent of US costs, roughly on par with the Asian exporters. Mexican labour costs rose slightly from 1986 to 1994, then again fell sharply in the wake of the 1995 financial crisis. In 1996, with investment and trade deregulation consolidated in NAFTA, Mexican hourly manufacturing labour costs were just 8 per cent of US costs and 8.5 per cent of Canadian costs. Mexican manufacturing labour costs had fallen to between one-quarter and one-sixth of its low-cost Asian competitors. The declining Mexican labour costs throughout this period correlate very highly with the rise of TNC investment in the maquiladora and related export sectors. Recalling that maquiladora wages have been roughly one half of wages in the domestic manufacturing sector, the Mexican advantage becomes even more pronounced.⁽¹⁵⁾

Labour productivity in Canadian manufacturing grew at a lower rate (0.8 per cent annually) than compensation growth (4.6 per cent) during 1984-91, resulting in a yearly rise in unit labour costs of 3.8 per cent. Between 1991-95 the situation reversed with labour productivity increasing at 2.8 per cent per year, above annual compensation growth of 1.6 per cent, resulting in a decline in unit labour costs of 1.2 per cent per year. (NAALC, 1997.)

In Mexico, labour productivity at large manufacturing firms grew at 6.3 per cent annually during 1986-94, while compensation costs grew at 4 per cent per year resulting in a drop in unit labour costs of 2 per cent annually. In the United States, labour productivity grew at 2.7 per cent annually during 1984-91 slightly less than compensation costs. During 1991-95 labour productivity increased to an average 3.4 per cent annually, above the annual growth of compensation of 2.3 per cent reducing unit labour costs by 1.1 per cent per year.

The common perception is that large wage differentials between Mexico and its NAFTA partners reflects the fact that productivity in Mexico is much lower. While this is true for the economy as a whole, it is not true for the manufacturing export sector.

Harley Shaiken, author of several important studies of work organization and technology transfer in the Mexican auto and electronics sector, provides the following examples of Mexican ability to handle the most technologically sophisticated manufacturing operations (Shaiken, 1993.)

• A Big Three auto engine plant in Mexico achieved productivity levels at 85 per cent of US performance in less than two years and in some machining lines surpassed US performance. Quality exceeded its US counterpart in 4 of 6 years for which data are available, by 32 per cent in the most recent year. The company is investing in a $500 million expansion with the capacity to produce 500,000 engines a year.
• The Ford assembly plant (Hermosillo) had the fifth highest quality rating out of 46 assembly plants in north America.
• The Sony television assembly plant (Tijuana) has won Sony's world-wide quality award for several years running and its productivity is identical to its counterpart in San Diego.
• Samsung (Tijuana) opened in 1996 a picture tube, television and VCR assembly plant which, by 2000, will produce 3 million televisions per year, mainly for export to the United States and Canada. (Shaiken 1997:3).

The decline of Mexican manufacturing wages for the last 17 years has occurred while manufacturing productivity grew steadily. During 1980-92, average manufacturing labour productivity grew 41 per cent (INEGI cited Shaiken, 1993) and another 18 per cent to 1995 (INEGI, cited International Metal Workers Federation, 1996:25). Meanwhile, real hourly wages fell 30 per cent during 1980-94 then tumbled another 25 per cent by mid-1996 (Shaiken,1997: 3). During 1980-96, average manufacturing labour costs fell 64 per cent against US and almost as much against Canadian labour costs. (BLS, 1997 op cit.).

Corporate surveys confirm, not only that NAFTA was the major factor influencing corporate strategies, but also the attractiveness of Mexico as an investment location. For example, a 1992 survey by the American Management Association of 200 of its member companies with investments in all three countries (three-quarters in manufacturing) found that they planned to expand employment in their Mexican operations by on average 13.7 per cent over the next three years, by 3.3 per cent in their US operations, and by 0.7 per cent in their Canadian affiliates. (Companies with large Canadian subsidiaries planned to reduce their Canadian work force by 2 per cent). The survey also found that manufacturing TNCs planned to increase investment spending in Mexico over the next three years by an average of 15.5 per cent, by 3.7 per cent in the United States and by less than 1 per cent in Canada.

The assessment of US financial analyst Louis Nevaer, exemplifies the importance of NAFTA to US corporate interests. NAFTA opened great opportunities for US business and prevented any possibility of a return of Mexico's state-led inward looking development model, "NAFTA is the surrender of economic sovereignty implying political sovereignty as well". (Nevaer, 1996, xv)... "The speed with which corporate America is extending into all areas of the Mexican market astounds in its completeness." [13] According to Navaer, "the tumultuous first year of NAFTA with its armed insurrection, political assassinations, and financial collapse were to be expected, "inevitable consequences of transformation and integration of the Mexican economy with those of the United States and Canada."

10. North American restructuring: The record of 11 TNCs⁽¹⁶⁾

In the next sections we have compiled statistical profiles for a sample of 11 transnationals in order to obtain a more focussed picture of corporate restructuring during the "free trade" era (1990-96). These companies are among the small group of companies that account for the large majority of North American manufactures trade. We were surprised to find how little of information on individual corporations is publicly accessible and how difficult it is to piece together even very basic profiles. The ability of transnational corporations to mask their activities is yet another manifestation of the enormous power they wield.

The statistical portraits include: employment, revenues, assets, profits, average wages, and CEO compensation - broken down where possible into the United States, Canada, Mexican maquiladora, and worldwide operations. They were put together using various sources: 10 K reports to the United States Securities and Exchange Commission, reports to the Ontario Securities Commission, newspaper reports and company annual reports,. The data on company activities in the maquiladora were obtained from Solunet Inc.(El Paso, TX.), a company which monitors corporate activity in the border region, covering the majority though not all, of their operations in Mexico Our corporate profiles provide a useful if incomplete picture (see corporate tables) of corporate restructuring under CUFTA/NAFTA. Ten companies are US-owned. One, telecommunications transnational, Northern Telecom, is Canadian-owned.⁽¹⁷⁾ Most of the companies are in the two most important maquila sectors-automotive and electrical/electronic. The auto sector companies are: the Big Three - General Motors, Ford and Chrysler, and parts makers United Technologies and Allied Signal. From the electronics sector are General Electric, IBM, Northern Telecom and Rockwell. Chemical giant Dupont, and chemical/ automotive/consumer products conglomerate 3M, complete our sample.

10.1 Employment
Although the trend varies somewhat from company to company there are clear patterns. Canada tended to be hardest hit in terms of employment, the bulk of that coming during the first, or CUFTA phase (1990-93) compared to the second, or NAFTA phase (1994-96). (In many cases, the size of the Canadian work force has become smaller than the maquiladora work force) Usually, the Canadian cuts were deeper than the United States cuts which in turn were usually less severe than worldwide cuts. Even where companies were growing worldwide their Canadian operations were cut in most cases. There were, however, notable exceptions such as Dupont, where the Canadian operations were less severely cut than the United States operations.

Only one company (IBM) expanded its Canadian work force. Only three expanded their US work force. Without exception, maquila employment expanded greatly reflecting the rapid rise in maquila employment described earlier, although there is some evidence of cyclical fluctuations.

General Electric cut its Canadian work force by more than 3000 or 35 per cent during the entire six year period, compared to 15 per cent in the United States (and 20 per cent worldwide). It expanded its maquiladora work force by 25 per cent, surpassing its Canadian work force by more than 2000 in 1996.

General Motors cut its Canadian work force by almost 10,000 or 25 per cent, most of it in the NAFTA phase (1993-96) contrary to the usual trend, and by 10 per cent (49,000) in the United States. (This compares with 14 per cent reduction in its world wide work force.) It compressed its maquila work force by 28 per cent in phase one, but then undertook a massive 170 per cent expansion in the NAFTA phase. Employment since 1990 rose by more than 21,000 reaching 43,000 by 1996. (Remember that maquila employment represents only a portion of total Mexican employment. GM has a total of 70,000 employees in Mexico.)

Ford's Canadian work force fell 22 per cent in phase one (1990-93), but recovered much of that loss in the NAFTA phase, with employment for the whole period still down 8 per cent, a loss of 2200 jobs. Ford's US work force grew steadily, increasing almost 34,000 (22 per cent) during 1990-96. Worldwide employment grew 12 per cent during this period. Ford's maquila work force more than doubled, from 11,000 to 23,000 during the period, most of the increase coming after 1993.

Chrysler's Canadian work force remained stable at around 14,000 while its maquila work force almost tripled, reaching 14,000 in 1996. The United States work force grew 10 per cent during the NAFTA phase. Employment worldwide remained stable during the period under examination.

Allied Signal cut its Canadian work force by 24 per cent during phase one recovering slightly during the NAFTA phase, but still down overall by 21 per cent. US employment figures were not available but the company cut its global work force by 28 per cent. Maquiladora employment jumped almost six-fold to 5950 in 1996, more than twice the size of its Canadian workforce. The bulk of the increase came in the NAFTA phase.

United Technologies cut its US work force steadily throughout the period by 29 per cent. This compares with a global employment cutback of only 6 per cent. Its Canadian subsidiary Pratt-Whitney shrunk by 15 per cent although in the NAFTA phase it increased its work force significantly. The company almost doubled its maquila employment to 14,200 in 1996, with the bulk of the increase in the NAFTA phase.

IBM, despite cutting its Canadian employment by almost one-third during phase one, expanded its work force by 5 per cent for the entire period. US employment increased steadily throughout the entire period, by more than 20 per cent. Worldwide employment fell during the period by 30 per cent. Though starting from a small base, IBM tripled its maquila employment to 1500, the large majority of it in the NAFTA phase.

Rockwell cut its Canadian operation by 13 per cent during the entire free trade period though in the first phase employment was cut 26 per cent with one half of the loss being made up in the second phase. Not sufficient information was available to determine the employment trend of the United States parent, but worldwide employment dropped 5 per cent (1991-95). Maquila employment more than doubled, though from a small base, with most of that coming in the NAFTA phase.3M reduced its Canadian work force by 11.3 per cent during 1990-96 while its maquiladora work force expanded by 169 per cent over the same period. US employment figures for 3M are not available, but the company's world-wide work force increased 6.5 per cent during 1992-96.

Dupont cut its Canadian work force steadily throughout the free trade period, by 18 per cent. The United States parent cut 40 per cent of its work force (1992-96). Worldwide staff cuts were almost as severe at 32 per cent during 1990-96. Maquiladora employment expanded fourfold though from a very small base, mostly in the NAFTA phase.

Canadian-owned Northern Telecom expanded its worldwide workforce 18 per cent to 67,584 (1991-96). Major restructuring in the early 1990s saw cuts of 4000 to its Canadian workforce but it has since expanded employment to 21,000 in 1996. Canadian sales were 14 per cent of worldwide sales in 1995 while US sales were 54 per cent. Both have been falling as a share of the total as European and Asian sales experienced large gains. Northern's maquila plant tripled its work force to 1500 1996. Northern's domestic Mexican workforce has also grown on the strength of a $330 million contract to upgrade the telephone system.

10.2 Wages/productivity/profit
Using sales per employee as a surrogate for labour productivity and using available data we are able to provide an approximate picture of these relationships in Canada and the United States.⁽¹⁸⁾ We then looked at how those productivity gains were distributed among workers, CEOs (the surrogate for senior executives) and worldwide profits. For the US-owned companies the comparison is with average US wages in the sector. For Northern Telecom the comparison is with the average Canadian sector wage.

The findings confirm that this sample of transnationals has had major productivity gains during the 1990s. However the lion's share of these gains has been appropriated by profits and more dramatically, by senior executive compensation. Average worker wages have been modest, in some cases barely keeping pace with inflation. These trends reflect broader industry trends (though more accentuated) and the growing inequality in both countries.

General Electric increased the productivity of its Canadian operation by 112 per cent and that of its US operation by 37 per cent. Comparing gains in CEO compensation with nominal wage gains for the average production worker in the sector (SIC 36), CEO compensation rose more than 500 per cent from 1991 to $32.6 million in 1996, while worker wages in the sector increased a modest 16 per cent to $26,284 in 1996, roughly the same as their Canadian counterparts. Thus the CEO-average wage ratio widened from 224 times to 1240 times. World-wide profits which averaged $3.9 billion during 1991-93, jumped to an average $6.3 billion during 1994-96. [We do not have senior executive compensation for the Canadian subsidiary of GE because like the other Canadian subsidiaries of the companies in our sample, their compensation is well down the ladder of executive compensation. Their compensation does not register on lists of highest paid Canadian CEOs.]

The labour productivity of GM's Canadian operation increased 88 per cent while productivity of its US parent rose 51 per cent. CEO compensation rose 247 per cent to $5.9 million. Average wages in the United States auto sector (SIC 371) increased 24 per cent. Wages in the Canadian auto sector (SIC 323) increased 33 per cent. The ratio of US CEO compensation to the average US sectoral wage widened from 50 times in 1991 to 143 times in 1996. Meanwhile world-wide losses averaging $8.5 billion turned into $4.6 billion annual profits during 1994-96.

Productivity at Ford Canada jumped 88 per cent while productivity at the United States parent rose 54 per cent. CEO compensation rose 645 per cent while sectoral wages rose 24 per cent (33 per cent in the Canadian sector). The CEO-average wage ratio widened from 34 times to 198 times during 1991-96. World wide profits which were negative, $-2.4 billion during 1991-93, jumped to an average $4.6 billion during 1994-96.

At Chrysler Canada productivity jumped 93 per cent, while at the United States parent productivity rose 28 per cent (1993-96). CEO compensation rose more than 1000 per cent to $12.6 million during 1991-96 while average US sectoral wages increased 24 per cent, (33 per cent at their Canadian counterparts), with a consequent widening of the gap from 33 times to 305 times. Worldwide profit at Chrysler averaged $1.8 during 1991-93, increasing to $3.1 billion during 1994-96.

Productivity at Allied Signal's Canadian affiliate increased 28 per cent. While US figures are not available, worldwide productivity jumped 52 per cent. CEO compensation rose 550 per cent to $13 million, while average sectoral wages increased 24 per cent, widening the gap from 59 times to 319 times. The information available on profits was not sufficient to make a comparison.

Productivity at Pratt Whitney Canada increased 35 per cent and 32 per cent at its parent United Technologies. CEO compensation soared 415 per cent to $6.7 million in 1996, while average sectoral wages increased 22 per cent. Global profits increased from an average $270 million during 1991-93 to $750 million during 1994-96.

IBM's Canadian productivity level rose 14 per cent during 1991-96 (although during 1990-91 it jumped 74 per cent). It is not possible to calculate a figure for the United States parent, but worldwide productivity jumped 68 per cent. CEO compensation jumped fourfold to $10.5 million while average sectoral wages increased 16 per cent (15 per cent in Canada) resulting in a widening of the CEO-average wage gap from 80 times to 343 times. Meanwhile world-wide profits which averaged a negative $-5.4 billion during 1991-93, rose to $2.2 billion during 1994-96.

Rockwell's Canada's productivity rose 75 per cent, while worldwide productivity rose 28 per cent. CEO compensation increased a relatively modest 133 per cent compared to the average sectoral wage increase of 23 per cent, widening the CEO-average wage gap from 44 times to 89 times. Global profits averaged $549 million during 1991-93, rising to $649 million during 1994-96.

The productivity of 3M's Canadian affiliate rose 38 per cent while productivity world-wide rose 24 per cent (1996-96). CEO compensation increased 100 per cent while sectoral average wages in the United States (and Canada) increased 14 per cent, widening the CEO-wage discrepancy from 74 to 131 times. Global profits averaged $1.24 billion during 1992-93 and $1.27 billion during 1994-96.

Dupont Canada's labour productivity jumped 63 per cent during 1991-96 while increasing 55 per cent at its US parent. CEO compensation increased 75 per cent while the average sectoral wage increased just 16 per cent (14 per cent in Canada), widening the gap from 52 times to 78 times. Global profit averaged $771 million during 1992-93 jumping to $3.2 billion during 1994-96.

Northern Telecom's world wide sales revenue grew 57 per cent, much faster than its labour force, reflecting a 33 per cent gain in labour productivity. Profits averaged $185 million during 1991-93, including a major loss of $878 million in the latter year. Profits recovered thereafter, averaging $501 million during 1994-96. Average wages increased 16 per cent the both US and the Canadian electronics sectors. We do not have figures on CEO compensation before 1993, but increases are much more modest than for the US-owned companies in our sample. Consequently the CEO-average wage gap is smaller, fluctuating between 54 and 108 times during 1993-96.

Appendix

For technical reasons, the Corporate statistical profiles are unavailable - please request a paper copy of this document from the ILO.

Table 5. Canada-US: Canada-Mexico exports and imports by manufacturing sector, 1988-96

II. Impact of the Canada-US Free Trade Agreement [FTA] and the North American Free Trade Agreement [NAFTA] on Canadian labour markets and labour and social standards

1. Introduction

Economic integration and labour and social standards.

Prevailing economic orthodoxy holds that liberalization of trade and investment may create short-term transitional problems for some groups of workers, sectors and communities, but has long term positive impacts on efficiency and growth. Liberalization results in a reallocation of labour and capital to areas of comparative advantage, raising productivity and potential growth. This in turn provides the basis for improvement in wages, working conditions and social standards. This is held to be true even in the case of closer economic integration between relatively high wage, high social standard countries such as Canada, and lower wage, developing countries. For example, the International Monetary Fund (1997) argues on the basis of a survey of the mainstream literature that increased North-South trade and investment flows have not been a significant driving force behind deindustrialization, the erosion of the relative position of unskilled workers and increased wage and social inequality in the advanced industrial countries.

Other, more balanced accounts, such as that in the ILO World Employment Outlook of 1996/1997 find some evidence that increased North-South trade has had negative impacts on relatively unskilled workers, but lay more emphasis on macro-economic and institutional factors in terms of explaining increased inequality and declining labour market and social standards in the advanced industrial countries. The ILO has laid particular emphasis on the role of contractionary macro-economic policy in raising unemployment and inequality in the advanced industrial countries, and has also consistently drawn attention to the linkages between declining rates of unionization and labour market deregulation, and increasing inequality among workers and in the wider society. However, the linkages between trade and investment liberalization and institutional changes in the labour market is left unexplored in this broad overview of trends in employment in the advanced countries. These linkages are explored in this paper.

The dominant neo-classical theory of gains from liberalized trade (and most econometric models are based on this theory) rests upon assumptions which abstract from reality in a number of key respects. Perhaps most importantly, the theory assumes full employment - workers displaced from some jobs move to others. But this assumption is clearly unrealistic in the context of the high and rising rates of unemployment which have existed in most advanced countries, including Canada, for at least the past two decades. High unemployment may be the result of macro-economic policy, but it is nonetheless the context in which liberalization of trade and investment has taken place. Further, there is a key linkage between liberalization of trade and investment and macro-economic policy in that it is more difficult to pursue expansionary policies in a context of globalized financial and product markets. The possibility of capital flight limits the scope for expansionary monetary policy, while leakages to imports limit the scope for expansionary fiscal measures.

Theory also assumes that capital flows from declining to expanding sectors, creating new jobs to replace those lost to imports as an economy is restructured through greater exposure to external market forces. But this assumption may be unrealistic given the increased integration of national capital markets and the real possibility of capital flight. Today's "free trade" agreements - including the FTA and NAFTA - are explicitly intended to reduce barriers, not just to the flow of goods and services, but also to direct investment capital flows between countries. It is by no means certain that direct investment inflows will balance outflows.

Finally, theory assumes that trade will be balanced by compensating movements in the exchange rate which reduce trade deficits or decrease surpluses. But, in the real world, shifts in the exchange rate do not necessarily balance trade, at least in the short term, and balance of trade deficits financed through foreign borrowing can and do persist for long periods of time. The assumption of mutual gains from trade through balancing movements in exports and imports cannot be assumed.

Anwar Sheikh (1996) has argued with others that the real world is increasingly one in which absolute advantage rather than comparative advantage holds sway. As in an integrated national economy, in countries integrated through trade and investment agreements, production, investment and jobs will tend to shift to those locales which are most cost competitive in the sense that they provide the best combination of wage and other costs in relation to productivity. Countries with the lowest unit labour costs will tend to have growing and dynamic export sectors, while countries with high unit labour costs will tend to suffer slower growth or recession and severe adjustment problems in the form of rising unemployment and declining labour market and social conditions as a result of lack of competitiveness. Of course, the constraints of cost competitiveness can, to some degree, be lessened through specialization in high quality and innovative products and services as opposed to commodities and commodity like goods, but this requires a shaping of competitive advantage through strategic investments in plant, innovation and skills which cannot simply be assumed to arise from "free markets".

The key point is that in the real world as opposed to the world of the neo-classical models, it cannot simply be assumed that there will be mutual gains from increased trade. There is the possibility of significant loss of jobs and production for any one country. And this possibility makes it even more likely that the "losers" from integration driven adjustments will not be compensated from overall increases in economic efficiency.

Shaikh argues that economic integration is likely to be much less disruptive between countries with broadly comparable levels of productivity and wages. Many labour market economists have also recognized that the impact of economic integration will be different if labour and social standards differ between countries. If competition in integrated product and capital markets tends to shift jobs, investment and production to regions where labour and social costs are low in relation to productivity, then there will be downward pressures on such standards in relatively high standard jurisdictions. As leading Canadian industrial relations academic M. Gunderson argues: "the economic forces of free trade, global competition and capital mobility put pressure on countries to harmonize their labour regulations, and these pressures tend to be in the direction of harmonizing downwards towards the lower common denominator. The pressure is especially great for measures that are the most interventionist and involve the protection of rents. They are also substantial, unfortunately - for regulations that serve an equity role if they conflict with efficiency." (Gunderson, 1993).

In short, economic integration and institutional factors at work in the labour market are not separate and distinct, but are rather closely connected.

2. The background to economic integration between Canada and the United States

A. Labour and social standards in Canada

The impact of growing economic integration between Canada and the United States offers an interesting example of the impacts of economic integration on labour markets, social standards and national distinctiveness.

Canada has traditionally had a modestly lower level of income than the United States, reflecting in part lower levels of productivity. Measured in terms of purchasing power parity, disposable income per capita in Canada was 77 per cent of the United States level in 1985, but this had fallen to 68 per cent in 1995 (Conference Board of Canada, 1995). The widening of the income gap reflects both macro-economic and structural developments related to economic integration, as explored below.

Notwithstanding lower incomes, labour market and social standards have, since at least the 1970s, been higher in Canada. Certainly, there have been and remain key differences in terms of rates of unionization. Overall union density in Canada has been more or less steady at about one in three workers in Canada in the 1980s and into the 1990s, but the United States unionization rate has fallen by half from 23 per cent to about 12 per cent over the same period. While this difference partly reflects high rates of unionization in a larger Canadian public sector, private sector unionization in Canada has been and has remained much higher, at 20 per cent to 25 per cent in the 1980s and into the 1990s, compared to a fall from 15 per cent to 10 per cent in the United States. Union density in US manufacturing has been about 10 per cent since the early 1980s, compared to about 35 per cent in Canada (Chaykowski and Slotsve, 1996; Kumar, 1987). Minimum wages and employment standards have also played a more significant role in Canada.

Canadian labour markets thus have been and remain significantly more regulated by unions and by governments than those in the United States. The generosity of income support programmes for workers has also been significantly greater. In the 1980s, roughly twice the proportion of unemployed workers qualified for Unemployment Insurance benefits in Canada compared to the United States, and the Canadian programme generally covered the vast majority of the unemployed. Social welfare programmes in Canada in the 1980s covered all unemployed workers who did not qualify for unemployment benefits, while US welfare programmes have typically been much more limited. The tax/transfer system has equalized incomes to a much greater extent than in the United States. Strikingly, in the 1980s, a family at the mid point of the income distribution was better off in Canada than in the United States, notwithstanding higher average incomes in the United States, because of greater equality in the distribution of income. A final major difference lies in the fact that Canada has had public Medicare since the late 1960s.

These differences in labour market institutions and social programmes bearing on labour reflect many factors, but certainly have been shaped by the relatively much stronger social democratic tradition in Canada. Put simply, Canada has weaker unions and welfare state programmes than is typical of European countries, but has been much closer to the European social market economy model than the United States.

A major NBER funded study of the impact of these differences in the 1980s by US labor economists David Card and Richard Freeman, tellingly titled Small Differences that Matter (1993) found that:

"Labor market institutions have resulted in less income inequality and poverty in Canada than in the United States. The United States chose to give relatively free play to market forces during the 1980s, whereas Canada pursued a more activist strategy of providing broader social safety nets and labor regulations that were more favourable to trade unionism. The American policies generated substantial employment growth, but did little to mitigate the redistribution of income towards higher income workers and families. Canada's programmes produced comparable employment growth, but also eased the forces that tended to promote inequality and poverty. Taken together, these studies show that subtle differences in unemployment compensation, unionization, immigration polices and income maintenance programmes have significantly affected the levels of poverty, unemployment and income inequality in the two countries"

Empirical work confirmed that transfer programmes markedly lowered Canadian poverty rates compared to those in the United States in the 1980s, while the rise in earnings inequality in Canada was much less than that in the United States in part because of the relative strength of Canadian unions.

Labour economists - as opposed to orthodox neo-classical economists - have tended to see Canadian distinctiveness in terms of labour market institutions as subject to erosion by competitive pressures in an integrated economic environment. For example, Gunderson (1993) argues that:

"the threat of capital mobility and plant location decisions will put pressure on different jurisdictions to harmonize their labour law and regulations. This will be true between Canada and the United States, especially under free trade. It is also the case that harmonization will be towards the lowest common denominator since jurisdictions with more costly regulations will be subject to pressures from investment flight and business relocations. In essence, the legislative authorities will be under more political pressure to pay attention to the concerns of employers over regulatory costs, including those emanating from the labour market. They will be under more pressure to compete with other jurisdictions to attract and retain business and the jobs that are involved."

Chaykowski and Slotsve (1996) have argued that private sector union density in Canada is likely to decline under the competitive pressures of free trade, unless the union wage premium declines or productivity gains in the more heavily unionized Canadian environment outpace those in the United States. Increased competitive pressure will not necessarily lead to deunionization, but it will undermine the bargaining power of labour vis-à-vis capital and make employers more resistant to unionization. Of course, it is possible for unions to help raise productivity to offset these pressures, but while unionization tends to be associated with higher than average productivity growth, unions will rightly insist that the gains be shared. High productivity does not necessarily translate into high rates of return on capital (Jackson, 1993)⁽¹⁹⁾.

While labour laws and regulations and income support programmes which strengthen the bargaining power of workers, notably Unemployment Insurance, are subject to downward pressures in an integrated economy, other social programmes have more mixed effects. Canada's Medicare system has given Canadian employers an important competitive advantage vis-à-vis US employers, or at least those US employers who contribute significantly to employee medical costs. Medicare also has given a significant cost advantage to auto production in Canada. The same could be argued of higher levels of Canadian public investment in transportation, education and other services. However, taxation on business to finance such public services is clearly subject to some competitive pressures, not least because corporations operating in both Canada and the United States can allocate taxable profits to one jurisdiction or the other through transfer pricing.

B. The economic role of the state

Canada has also been highly distinct from the United States with respect to the role played by the state in economic development. Close trade and investment ties with the United States begin to displace those with Britain from the 1920s, with Canadian resource exports flowing to the United States in return for US manufactured goods, and with US corporations establishing Canadian branch plants to produce behind the then high Canadian tariff wall. Canada and the United States were already significantly integrated in terms of trade flows in the 1950s and two way trade grew strongly as tariffs were successively lowered through GATT rounds. By the 1970s, US direct investment in Canada accounted for a large - often majority - stake in the resource and manufacturing sectors. In brief, long before the FTA and NAFTA, Canada and the United States were already highly integrated economies.

But the nature of Canada-US economic integration was significantly shaped by government policies until well into the 1980s. The broad purpose of such policies was to broaden the industrial and employment base away from resources and resource based products to more advanced products and services, and to capture more resource rents. In the 1980s, tariffs were quite low for most products, but trade in the large integrated auto sector was managed through the Auto Pact with the United States which guaranteed production in Canada broadly equivalent to market share. Public ownership was significant even in commercial, non-regulated sectors such as energy, and foreign direct investment was subject to review and approval, conditional upon Canadian employment and other benefits. Canadian based producers were given preferential access to Canadian resources, notably energy in the National Energy Programme. Regulation gave US companies limited access to several key sectors, including transportation, communications, cultural industries and the financial sector. In short, regulation, public investment and a wide range of industrial and regional development policies shaped and managed a close economic relationship integration with the United States.

The negotiation of the Canada-US Free Trade Agreement - concluded in 1988 - was very much part of a wider federal government agenda of deregulation and privatization and "structural reform" consciously intended to give wider sway to market forces. The FTA was, of course, liberalizing in the sense that it eliminated, or phased out, remaining tariff barriers. Prior to the agreement, one third of Canadian industrial production was protected by tariffs of 7.5 per cent or higher, with tariff protection being particularly significant in some labour intensive sectors such as clothing, textiles and footwear as well as in food and beverages, rubber and plastics. But it was more deeply liberalizing in the sense that it prohibited or limited traditional tools of economic policy such as conditional access to resources, two price systems for resources, regulated market access as in the Auto Pact, foreign investment review, "discrimination" by governments and crown corporations, and use of public procurement, among others.

The NAFTA went beyond the FTA in some respects, further liberalizing trade in some services and further limiting regulation of investment. It also, of course, added Mexico to the FTA region, but this was of modest direct significance for Canada given that direct trade and investment ties with Mexico were and remain of minor importance (with the partial exception of trade in auto and electronic products). However, close integration with the United States clearly means that NAFTA impacts on the United States arising from the integration of Mexico can have significant repercussions in Canada.

It should be underlined that the FTA/NAFTA is much more than a "free trade" agreement and goes well beyond tariff elimination to limit government actions which can be seen as affecting the evolution of markets. The agreement is, in essence, a declaration that Canadian and US corporations have the right (subject still to certain limits) to ship products and sell services across the border and to invest in the other country and to be treated by governments in a non-discriminatory fashion. The key message to Canadian and US corporations is that the locale of production and investment and jobs can be determined on the basis of the logic of profit maximization rather than political criteria, and that henceforth the United States and Canada should be seen as a single market. It is worth noting that, unlike the EC, the FTA/NAFTA is almost completely a liberalizing agreement designed to break down "barriers" and prevent government "discrimination" and does not create new institutions which pool sovereignty. It deregulates the market by limiting the scope for national government intervention without creating new international institutions. The environmental and labour commissions created by NAFTA stand as minor, weak exceptions.

3. The Canadian policy debate

There were four major strands to the Canadian economic policy and political debate over the FTA and then NAFTA. The first had to do with whether the FTA secured one of its ostensible key purposes - namely secure access to the United States market and protection for Canadian exports against recurrent US protectionism. The FTA did not exempt Canada from US countervail and anti-dumping laws, though it did establish a dispute settlement process to determine if US law has been fairly applied. (This may have resulted in some modest benefits over and above GATT/WTO tribunals, but the United States has, post FTA, continued to periodically limit Canadian exports of lumber, steel, agricultural and other sensitive products.)

The second strand had to do with limits on the role of government in the economy. Nationalist, labour and other critics of the FTA argued that tools of policy, sacrificed under the agreement were needed to build productive capacity and to broaden Canada's narrow advanced industrial base, while supporters argued that market access and free markets would narrow the productivity gap with the United States and lead to more and better jobs in sophisticated industries. It was recognized on both sides that Canadian industry was less advanced and competitive than US industries, particularly in technologically sophisticated sectors.

Third, critics argued that restructuring driven by the agreement would lead to significant job losses and adjustment problems, while supporters said that adjustment would be modest and that there would be net gains from reallocation of production of investment. Critics expected that companies which had traditionally operated production facilities on both sides of the border (mainly US companies which had operated "branch plants" in Canada to avoid tariffs and comply with Canadian content requirements) would shift production to larger US plants, while supporters expected that better access to the United States would encourage modernization and expansion of Canadian plants. There was evidence for both positions in the evolution of the Canadian industrial structure in the 1980s.

Finally, critics said that economic integration and the competitive pressures to which it gave rise would lead to downward harmonization of labour and social standards to the lower levels prevailing in the United States. This process would result not just from the more competitive environment created by tariff reduction, but also through the wider deregulation of the economy driven by the FTA.

The key assessment of the FTA released by the Department of Finance in 1988 argued that the FTA would provide secure and enhanced access to the United States market and would lead to a "more efficient and lower cost economy" via economies of scale, higher productivity and higher rates of innovation. The lower consumer prices driven by tariff cuts and the productivity gains from comparative advantage and economies of scale in production specialization were expected to result in a long term real income gain of at least 2.5 per cent. More intangible results were expected in the form of "increased flexibility and dynamism".

"The economic benefits from the free Trade Agreement will start to be realized shortly after the implementation of the Agreement on January 1, 1989. Prices for a wide range of consumer goods will begin to decline, expanding the purchasing power of Canadian households. Investment in plant and equipment will expand as Canadian firms move to take advantage of their enhanced access to the huge US market place. Increased consumer and investment spending will lead to stronger economic growth and more job creation. Department of Finance estimates of the impacts of the Agreement on employment over the government's medium-term fiscal planning horizon indicate that 120,000 net new jobs will be created by 1993, only five years into the agreement."

The government funded Economic Council of Canada study (1988) estimated the stimulus to consumption and investment which would come from price reductions flowing from lower Canadian tariffs, and the impacts of trade driven rationalization and specialization on the manufacturing sector. It forecast a 2.5 per cent gain in real GDP over 10 years. This result was heavily dependent upon an assumed productivity increase as a result of rationalization and economies of scale. The Council forecast modest 2-3 per cent job losses in heavily protected sectors as a result of tariff reduction, and calculated that there would be net job losses if only tariff changes were considered. However, the adjustment problem was viewed as very small. Both of these major studies argued that there would be modest but positive benefits mainly because of an expected increase in productivity. Tariff driven changes to employment were seen as small and manageable. Neither study saw any threat at all to labour and social standards or to Canadian distinctiveness vis-á-vis the United States. In his foreword to the Department of Finance report, the Minister wrote "a stronger and more productive economy will allow us to better support our social programmes and further advance our cultural identity." The Economic Council argued that "there is no reason to think that free trade can be used to alter or undermine Canada's Unemployment Insurance, medicare or other programmes fundamental to the social safety net. Rather, by strengthening economic growth and employment in Canada, the free-trade agreement will increase government revenues and improve the ability of governments to address social issues." Neither the Department of Finance nor the Economic Council study looked at Canada-US wage and productivity differences or at the implications of different labour market institutions in the two countries in a more integrated environment. The Department of Finance study NAFTA: An Economic Assessment (November 1992)⁽²⁰⁾ similarly dismissed the "downward harmonization" argument, with specific reference to Mexico. The basic argument which was not disaggregated by sector or even to the traded goods sector alone, was that the huge Canada-Mexico wage gap was more than paid for by higher productivity. It forecast tiny income gains from trade and investment liberalization.

Critics of the FTA forecast significant job losses as manufacturing industries in Canada were restructured and rationalized and two way trade and investment flows increased as a result of tariff reduction and more secure rights for foreign investors (Cameron, 1988). While there was concern over job displacement from tariff reduction, the leading concern was loss of the economic tools needed to advance an interventionist Canadian industrial policy. Economists opposed to the FTA were highly sceptical of the view that market forces alone would lead to industrial diversification and the growth of more sophisticated industries in Canada, and feared that the United States would continue to dominate such sectors as Canada became even more reliant on resources. In short, much of the debate was less about "free trade" per se, than over the role of the state in economic development and job creation. Critics also forecast a process of downward harmonization of labour and social standards. In their brief to the House of Commons Committee on International Trade, the Canadian Labour Congress, the peak trade union organization, argued that "if this deal is consummated we can count on business on both side of the border to threaten not to invest in Canada if they think costs and conditions are out of line with those in the United States. That way we get a further endless round of harmonization through the market that could erode everything from medicare and environmental protection to the corporate income tax and workers' rights to organize and bargain collectively." Similarly, the CLC submission to the government on NAFTA argued that the extension of the FTA to include Mexico would "pose the very real threat of further job losses, wage cuts and erosion of labour, social and environmental standards as the international competitiveness threshold to which Canadians are directly exposed is dramatically lowered."

The debate over NAFTA in Canada, as in the United States, centred around fears of downward harmonization, with many Canadians arguing that the inclusion of Mexico in NAFTA would increase pressures that had already become clear under the FTA. The basic argument was that NAFTA would increase both capital mobility and competition, leading employers to attempt to lower wages in relation to productivity. Contrary to the assumption of neo-classical economists that wages reflect relative productivity, it was argued that in the United States and, even more dramatically, in Mexico productivity and wages had become delinked. Examples of high productivity, low wage production in Mexico - as in the auto engine industry studied in detail by Shaiken - made it clear that labour market institutions as well as macro-economic conditions are key intervening variables. With the option of relocation to a lower wage, lower labour standard jurisdiction with little or no loss of quality or productivity, employers will be placed in a position to successfully push workers for wage concessions and to better resist unionization (Robinson, 1994; Stanford, 1995).

4. Impact of FTA and NAFTA

A. Canada in the 1990s

By almost any measure, the post FTA period of 1989 through 1996 has been one of the most dismal in Canadian history. Leading economist Pierre Fortin has characterized this period as "the great Canadian slump" - the longest period of below potential growth since the Great Depression.

B. Growth

As shown in table 1, the recession of the early 1990s was deeper and more prolonged in Canada than in the United States, and the recovery more tentative. Prior to 1997 - a year in which the economy has finally begun to grow at or above 3 per cent - 1994 was the only year of growth above 2.4 per cent. The contrast to the pre FTA recovery period of the mid to late 1980s is marked. Economic growth averaged 3.2 per cent, 1980-88, compared to 1.4 per cent 1989-1996. These periods are both ones of recession and recovery. Canada grew significantly faster than the United States, 1980 to 1988 (notwithstanding labour market institutions and social programmes which were stronger than in the United States, and a more regulated economy.) However, growth in the post FTA period has been significantly slower than in the United States, averaging just 1.4 per cent compared to 2.0 per cent. Because of slower growth, Canadian incomes have fallen significantly behind those in the United States. Given population growth, there has been almost no growth in real living standards in Canada since 1988.

C. Major labour market trends

The national unemployment rate soared from a low of 7.5 per cent in 1989 to more than 11 per cent in the early 1990s, and remains well above 9 per cent. The average duration of unemployment is approximately six months, so the unemployment rate indicates that at least one in five Canadian workers have been unemployed each year in the 1990s. Regionally, unemployment has been highest in Quebec and the Atlantic Provinces, and lowest in Western Canada. However, the recession was deepest in heavily industrialized Ontario and Quebec. The more resource industry based Western provinces were less directly impacted by the FTA.

An historically unprecedented gap has emerged between unemployment rates in Canada and the United States, largely because of much slower job growth in Canada. The absolute number of full-time jobs recorded in 1990 was not regained until 1997, reflecting the severity of the recession and the concentration of job growth for women in part-time employment. The participation rate has fallen steadily from 67.5 per cent in 1989 to below 65 per cent today, with most of the decline concentrated among young people.

Table 1. Canada and the United States - major indicators (per cent)

Were it not for the fall in the participation rate since 1989, the unemployment rate in 1996 would have been above 13 per cent. The unemployment rate among young people aged 15 to 24 has been close to 20 per cent through the 1990s, and this is on the basis of an extremely steep fall in the participation rate since 1989 (10 per cent for young men and 7 per cent for young women)⁽²¹⁾. The incidence of part-time employment among young people has also increased sharply, from 36 per cent in 1988 to more than 50 per cent today.

The incidence of part-time work has increased from 20 per cent to 25 per cent among adult women since 1989, and this is almost entirely explained by the increased incidence of involuntary part-time employment. The trend towards part-time jobs for women has also been somewhat disguised by the growth of multiple job holding - about one in four jobs are now part-time, though only one in five workers work part-time because a growing number work full-time hours by combining jobs. Part-time workers - overwhelmingly women - earn just two-thirds the wage of equivalent full-time workers, and less than 20 per cent receive benefits from their employer. Increasingly, part-time work has become more and more casual, with hours in sectors such as retail trade, restaurants and hotels being highly variable from one week to the next.

The casualization of employment in the 1990s is also revealed in the rapid growth of temporary work, which increased from 5.0 per cent to 11.6 per cent of total employment between 1991 and 1996 according to the Statistics Canada Survey of Work Arrangements. Self-employment meanwhile grew by 15 per cent between 1991 and 1995, and has contributed about one-half of all "job" growth in the 1990s. While there is a layer of skilled professionals in high demand among the ranks of the self-employed, the growth of casual and self-employment is driven above all by the unwillingness of employers to hire workers into regular jobs, and by their ability to meet their needs on this basis in a very high unemployment economy. The phenomenal growth of self-employment has not been extensively analysed, but it is clear that earnings generally average well below those of paid workers with comparable skills. Measures of income from self-employment in the National Accounts show little growth, indicating that the average incomes of the self-employed are falling.

The very marked casualization of employment in the 1990s, particularly among women, young people and other disadvantaged groups in the labour market, has been much greater than in the 1980s, and undoubtedly reflects very weak demand for workers and high unemployment combined with major cuts to income support programmes. It is driven by employer competitiveness strategies of contracting-out and out-sourcing work, and making the hours of work highly variable in order to minimize the fixed costs of permanent, full-time workers who usually have access to benefits such as pensions, and are frequently unionized.

Hours worked in full-time jobs increased in the 1980s, and the rate of increase has picked up in the 1990s. The proportion of adult men working more than 41 hours per week rose from 18.0 per cent in 1980 to 21.4 per cent in 1989 and to 24.3 per cent in 1995. (The comparable data for adult women are 5.6 per cent, 7.3 per cent and 8.6 per cent). The proportion of adult men in the goods sector working more than 41 hours per week rose from 15.8 per cent in 1989 to 19.4 per cent in 1989 to 24.6 per cent in 1995 (Statistics Canada, 1996). Long and increasing hours tend to be worked by professionals - both women and men - and by male production workers in manufacturing and the resource sector. Average weekly overtime hours for hourly paid workers have averaged 3.2 hours 1994-96, significantly higher than the previous cyclical high of 2.2 hours in 1988.

Measured solely in terms of hours worked, there has been a significant intensification of work for many workers in blue collar occupations, as well as in the public sector (where overtime is frequently unpaid.) High levels of overtime result from the high costs of training new hires, and from the fact that a significant proportion of wage costs for workers with pension and other benefits coverage are fixed. Work has also been greatly intensified in most workplaces by delayering and "downsizing", and by the widespread adoption of "lean production" techniques both in industrial and non-industrial settings. The intensification of work has gathered pace in the 1990s (Dagg, Schenk and Anderson, 1995).

Table 2 shows wage increases for all workers and for unionized workers in the business sector and the increase in the Consumer Price Index. Over the 8 year period 1989-96, there was a very modest 6 per cent increase in the broadest wage measure (which covers salaried as well as hourly paid workers) while wage settlements for unionized workers in the business sector basically just matched inflation. Wage settlements for public sector workers averaged well under those in the private sector because of wage freezes and legislated controls.

Table 2. Earnings per cent change

Table 3. Earnings trends 1989-1993: Change in real annual earnings (per cent)

Real annual earnings fell for virtually all adult male workers between 1989 and 1993. As shown in table 3, real earnings of the bottom decile of adult men fell 31.2 per cent between 1989 and 1993, and those of the second decile fell by 20.2 per cent. Real annual earnings of the top two deciles of men fell by just 2 per cent. There has thus been continuing polarization in earnings of men, though this trend has been driven more by a decrease in time worked as a result of high unemployment than by polarization of wages. Nonetheless, the proportion of adult male workers earning less than $8.80 per hour (in 1993 dollars) rose from 7.9 per cent in 1989 to 8.9 per cent in 1993, while the proportion of such workers earning more than $27.60 per hour rose from 9.3 per cent to 11.6 per cent. Inequality of hourly earnings has increased, but the rate of increase has not accelerated since the early 1980s.

The real annual earnings of the bottom 50 per cent of adult women fell between 1989 and 1992, again because of changes in hours worked. Annual earnings fell by 10.8 per cent for women in the bottom decile and by 5.9 per cent for women in the second decile. The proportion of women in very low wage jobs did not increase, and hourly earnings among women have not polarized in the same way as those of men. However, women are heavily concentrated in the lowest pay categories - 19.9 per cent of women earned less than $8.80 per hour in 1993 compared to 8.9 per cent of men, and just 5.4 per cent of women earned more than $27.60 per hour compared to 11.6 per cent of men (Picot, 1996).

The negative impact on incomes of working people and families of increased unemployment and the erosion of earnings, particularly for the low paid, has not been offset by income transfers to the same extent as in the 1980s. Three major rounds of cuts to the national Unemployment Insurance programme since 1988 cut benefit entitlement periods drastically - often in half - and raised the qualification period to 26 weeks of work for full-time workers. In 1989, 85 per cent of unemployed workers qualified for UI benefits, but less than 50 per cent qualify in 1997, and the proportion has fallen to one-third in Ontario. The benefit rate as a proportion of prior earnings has also been reduced. The Department of Finance estimates that the effect of these cuts has been to eliminate the impact of the major improvements to the system made in the 1970s. The programme is certainly very substantially smaller than it was in 1988.

Income support from social assistance grew rapidly in the recession and the early recovery, but benefits have been significantly cut in most provinces since 1992, and it has become more difficult to claim benefits. Thus, despite continuing very high rates of unemployment, transfer income has been shrinking relative to employment income. The inevitable result has been a sharp increase in poverty, particularly among families with children. The incidence of low income for families with children has risen from 15.3 per cent in 1989, to 21.0 per cent in 1995.⁽²²⁾

5. Macro-economic policy and economic integration

Most economists - certainly in the Department of Finance, the Bank of Canada, the OECD and business economists - would argue that Canada's dismal recent history is largely unrelated to the FTA. Indeed most would argue that increased exports to the United States flowing from the FTA have led growth in the recovery. The argument would be that economic performance has been weak and unemployment very high because of the impacts of monetary and fiscal policy. Extremely high interest rates and exchange rate overvaluation lie behind the recession of the early 1990s, while subsequent recovery has been held back by stringent fiscal policy as governments have sharply cut spending to deal with a growing debt load. It would also be argued that major structural adjustments to "globalization" were inevitable, notwithstanding the FTA and NAFTA (an argument which glosses over the fact that continental economic integration is precisely the form that "globalization" has taken in Canada.) (Lipsey, 1995).

There can be no doubt that macro-economic policy has been the major cause of the Canadian slump and labour market trends. In particular, blame has to be placed primarily on monetary policy. From the low 70 cents US level of the mid 1980s, the Canada-US exchange rate rose by more than 20 per cent to at or above 86 cents US in 1989 through 1991, before falling back to the low 70 cent level today. In 1990, short-term interest rates were 5 per cent higher than in the United States, compared to a normal average of about 2 per cent, and this fuelled the sharp exchange rate appreciation which came almost coincidentally with the introduction of the FTA in 1989. The traditionally large Canadian merchandise trade surplus with the United States (needed to finance the usually still larger deficit on trade in services and dividend flows) fell in the 1989 to 1991 period, when the manufacturing and much of the resource sector contracted sharply. Extraordinarily high real interest rates put the domestic economy into deep recession.

As prominent Canadian macro-economist Pierre Fortin and others have argued, it was the contractionary monetary policy of the late 1980s which led to the very severely contractionary fiscal policy of the 1990s. Government debt loads soared in the recession because of the combination of high interest rates on accumulated debt, and high unemployment and sluggish revenues. The significant Canadian debt problem was caused by monetary policy, rather than by "excessive" spending. Programme spending had been cut in the 1980s as a share of the economy, and has been cut sharply in the 1990s (Osberg and Fortin, 1996). Fiscal retrenchment in the 1990s has been, according to the OECD, the most severe in any major country in the post-War era. Real federal government spending on programmes has fallen by more than 20 per cent since 1993, cutting real growth rates by well over 1 per cent per year in the recovery. There have been massive job cuts to public services, directly and, via cuts in federal transfers, to provincial and local services and to health and education.

While the fiscal tightening has been rooted in monetary policy, it is not unrelated to economic integration in that the drive for international competitiveness has taken priority over stimulation of internal demand. For example, cuts to Unemployment Insurance have been justified in terms of perceived negative impacts on the labour market, and the traditional counter cyclical role of the programme has been consciously scaled back. Further, tax increases, particularly tax increases on business, have been rejected as an alternative to social programme and public service cuts because of the perceived negative impacts on competitiveness. As in other countries, much of government's role has been reconstrued to be the enhancement of business competitiveness via domestic policies of austerity, a trend which has been speeded by the fact that more and more domestic demand is now met through imports so fiscal stimulus in less effective.

Critics of monetary policy rightly lay much of the blame on the determination of the Bank of Canada to achieve an explicitly stated "zero inflation" objective under Governor Crow. Notwithstanding a rather low peak rate of inflation of 5 per cent in 1989, the monetary brakes were tightened to a much greater extent than in the United States. Many prominent economists, including Paul Krugman, have argued that the Bank of Canada has pursued a much more stringent anti- inflation objective than even the United States Federal Reserve, and that the Canadian case stands as an example of the large costs of adopting too low an inflation target.

There, is, however, much more to the shift in monetary policy than an error in macro-economic judgement. The Bank of Canada was responding not so much to higher consumer price inflation than in the United States as to the perceived loss of Canadian cost competitiveness vis-à-vis the United States in the 1980s. As shown in table 2, inflation in 1989 was 5.0 per cent, only 0.2 per cent higher than in the United States, and wage increases were broadly in line with price increases even at the end of the expansion and were not out of line with productivity growth.

Put bluntly, the Bank deliberately increased unemployment in order to discipline Canadian workers in a more integrated economic environment. From the perspective of business and many economists in the 1980s, a key Canadian problem was eroding cost competitiveness with the United States in manufacturing and the traded goods sector generally. The expansion of the 1980s saw a sharp increase in unit labour costs and an erosion of Canada's cost competitive position which was offset only by continued exchange rate depreciation. The crux of the problem was that productivity was growing more slowly than in the United States, while wages were growing somewhat more rapidly in relation to productivity than in the United States (Economic Council of Canada, 1992). The problem was not "excessive" wage growth in Canada per se - real wage increases were barely increasing and were more than justified by productivity - but rather the potent competitive threat posed by the particularly sharp delinking of wages from productivity growth in the largely deunionized United States.

Beginning from at least the mid 1980s, business interests in Canada expressed growing concern about loss of competitiveness. While much of the blame could be placed on corporate under-investment in innovation and research and skills, it was argued that Canadian labour was "too strong" in the new, more competitive, more integrated economic environment which flowed from the overall process of deregulation, tariff reduction and looming free trade. Put in more technical terms, it was argued by the Bank of Canada, the Department of Finance (1995) and the OECD (1996) that Canada had a higher "natural rate" of unemployment or NAIRU than the United States in large part because of stronger unions and more "generous" social programmes⁽²³⁾. "Natural rate" theory was the justification for the sharp tightening of monetary policy in the late 1980s and the successive rounds of cuts to UI made in the late 1980s and 1990s.

To summarize, monetary tightening and the consequent squeeze in fiscal policy were not unrelated to the context of much closer economic integration with the United States. Inflation per se was not a problem in the late 1980s. What was a problem was the perceived higher propensity of the Canadian economy to inflation than the United States. This difference was seen as rooted primarily in labour market institutions. High unemployment was used deliberately to promote long term competitiveness with the United States, even at the cost of a short run deterioration in relative costs.

It can be added that there is some evidence that the FTA would not have been approved by the United States Congress if an exchange rate appreciation had not been anticipated. Indeed a member of the federal Cabinet that negotiated the FTA, Sinclair Stevens, has said that there was a "secret deal" to placate hostile members of Congress.

6. The FTA, trade and direct employment impacts

Analysts basically agree that the FTA and NAFTA have had a significant impact on Canadian trade patterns. Both exports and imports have risen sharply as a share of GDP, and trade has become even more geographically concentrated. Exports have risen from 26 per cent of (nominal) GDP in 1988 to 38 per cent in 1996, matched by an almost equally large increase in the import share of GDP from 26 per cent to 35 per cent. The Canadian merchandise trade surplus with the United States moved in a narrow range between 1988 and 1992, in the early FTA period, and then rose very rapidly, from $15 Billion to $40 Billion. (The growth of the merchandise trade surplus accounted for about 25 per cent of economic growth, 1992-1996). Exports to the United States rose from 75 per cent to 81 per cent of total exports over the same period, while imports from the United States rose from 69 per cent to 75 per cent of the total.

Measured in nominal dollars, both exports and imports of goods and services have about doubled in the post FTA period. This expansion of both exports and imports was much more rapid than in the 1980s or in the 1970s. In short, the Canadian and US economies have indeed become much more integrated in trade terms at a rapid pace. The extent of integration is now remarkable. A greater share of Canadian manufacturing production is now exported to the United States than is consumed in Canada, and Canadian manufacturers supply less than half of the Canadian market for manufactured goods. It should be noted that a very high portion of Canada-US trade - 40 per cent - is intra-company trade, particularly in the highly integrated automotive sector. Trade is made up of inputs to cross border production chains as much as resources and finished goods.

There have been some changes in the broad structure of Canada-US trade under the FTA and NAFTA. Resource based goods as a share of Canadian merchandise exports have diminished slightly in importance from 34 per cent to 31 per cent of the total, while the share of machinery and equipment in exports has risen from 16 per cent to 23 per cent, 1988 to 1996. This has been driven by exports of telecommunications equipment, a traditional area of Canadian strength, and by growing exports of office machinery and software. Canada's small "high technology" sector has been the major beneficiary of the FTA. However, Canada remains a very large net importer of machinery and equipment, exporting 80 per cent of the value of imports in 1996. The export to import ratio would likely fall if depressed Canadian industrial investment recovered strongly.

A detailed analysis of changes in the pattern of trade by Schwanen (1996) shows that the growth of trade has been particularly strong in sectors liberalized by the agreement, and that the growth of trade with the United States has been greater than would have been expected given slower growth in the United States than in other markets, and the greater depreciation of the Canadian dollar against other currencies. While overall exports to the United States grew by 99 per cent, 1988 to 1995, exports in liberalized sectors grew by 139 per cent. As a result of the FTA, the United States market share in Canada increased significantly in sectors liberalized by the agreement, notably clothing, furniture, processed foods, steel and chemicals. Canadian exports to the United States have grown particularly strongly in liberalized sectors also. A US Congressional Research Service Report (Wilson, 1996), similarly concludes that the FTA has had an important independent impact on trade flows. If anything, the impact of the FTA on trade has probably exceeded expectations, indicating that corporations reconsidered production strategies in a new light after the agreement was concluded, rather than making marginal adjustments.

Shifts in trade in response to the FTA in combination with exchange rate movements and domestic economic conditions resulted in a major restructuring of the manufacturing sector. There was a massive wave of plant closures and mass layoffs in industrial Canada, particularly Ontario and Quebec, in 1989 through 1992. Over this short period, about one in five manufacturing jobs was lost. Since 1992 there has been a modest recovery in payroll employment in manufacturing. Job losses were often in US (or, more rarely, Canadian) companies which operated production facilities in the United States and decided to rationalize higher cost operations in Canada. Other jobs were directly lost to competition from imports. As shown in table 4, there have been significant job losses in those industries which enjoyed significant tariff protection before the FTA - notably clothing and food. Strikingly, jobs have also been lost in significant numbers in sectors where exports have expanded, notably machinery and electrical and electronic products (though employment in some sub-sectors producing telecommunications and office equipment has grown.)

University of Toronto economist Daniel Trefler has calculated that employment in sectors with tariff protection of more than 10 per cent (about one-third of manufacturing employment) fell by 17 per cent 1988 to 1996 and that 138,000 of 290,000 manufacturing jobs lost between 1988 and 1996 can be attributed directly to the FTA rather than to other factors. He finds that these protected sectors, such as clothing and textiles, have restructured mainly by shrinkage - there have been limited productivity but not jobs gains even in the smaller sector which has survived. Conversely, there was rapid growth of output and productivity in sectors which had more modest tariff protection before the FTA (Trefler, 1997).

Table 4. Employment in manufacturing (000s)

Between 1988 and 1995, there was a very sharp 18.7 per cent fall in the number of manufacturing establishments in Canada. 7,544 establishments closed with the most severe declines in clothing (1,094 establishments, 39 per cent of the 1988 total) furniture (612 establishments or 31.4 per cent of the 1988 total), printing and publishing (1,190 establishments or 21 per cent of the 1988 total). The fact that the number of establishments shrank more than the fall in employment indicates that restructuring resulted in increased concentration of production. This appears to have taken place mainly in medium sized plants, since the number of large (more than 500 worker) manufacturing establishments fell from 436 to 365 (Statistics Canada, 1996).

Under the FTA, cross-border trade in services with the United States has also grown, though less rapidly than merchandise trade. Such trade - travel, transportation and commercial services - accounts for only 14 per cent of the value of merchandise trade. Canada's traditional deficit in the trade of services with the United States widened from $4.6 Billion in 1988 to a high of $10.7 Billion in 1992, and has since gradually shrunk to $9 Billion. Canada currently exports $2 of services to the United States for every $3 which is imported. There have been significant FTA and NAFTA related impacts on Canadian transportation industries, notably cross border trucking. Competitive pressures have greatly increased in all transportation industries.

Under NAFTA, direct Canada-Mexico trade has about doubled in volume but continues to be modest, amounting to only about 3 per cent of the value of Canada-US trade. This may be underestimated in that US goods assembled in Mexico may be counted as US goods. There has been a large and growing imbalance in direct Canada-Mexico trade, driven by the shift of auto industry parts sourcing to Mexico and by the slump in Mexican imports following the peso crisis. In 1996, Canadian imports from Mexico were five times greater than exports ($6 Billion vs $1.2 Billion) and this large imbalance has grown. In the first five months of 1997, Canadian imports from Mexico were six times greater than Canadian exports to Mexico. While still modest in dollar terms, the trade deficit undoubtedly has translated into direct job losses. Several Canadian auto parts producers (eg. Magna, Ford) and electronic equipment producers (eg. Northern Telecom) have established production facilities in Mexico. In these sectors, the threat - real or potential - of relocation has been used in collective bargaining.

7. The FTA and economic performance

As noted above, proponents argued that the FTA would lead to a revitalization of the manufacturing sector based on higher productivity and new investment. Performance in this respect has been disappointing. As shown in table 5, labour productivity growth in manufacturing (output per hour) averaged 2.0 per cent 1989 to 1995, down from 2.3 per cent 1981 to 1988. Productivity growth was just above 1 per cent in 1995 and 1996 (based on preliminary data for 1996). The productivity gap with the United States has not been closed and has in fact recently widened. Labour productivity growth in US manufacturing has been rising at above 3 per cent since 1993, and averaged 2.6 per cent, 1989-95. The same trend is equally apparent with respect to total factor productivity in manufacturing (Conference Board of Canada, chap. 5, 1995).

While poor Canadian productivity performance undoubtedly reflects depressed domestic conditions, it has been low considering the closure of many low productivity firms (which should have raised growth through a concentration effect) and considering that free trade gave Canadian manufacturers access to a faster growing market. The 1996 OECD Country Review of Canada expresses surprise and concern that the expected impacts of structural reform, including free trade, have yet to appear, and the same tone of puzzlement is present in the recent Conference Board of Canada report on Canadian Economic Performance. Comments are made on the negative effects of the rapidity of needed adjustments - begging the question of whether more gradual liberalization would have resulted in better performance.

Poor productivity performance likely reflects relatively poor levels of manufacturing investment. Measured in nominal dollars, investment in construction of new manufacturing facilities was more than $4 Billion in 1989 and 1990 but, after a sharp fall, has been consistently below $3 Billion per year even in the 1992 to 1996 recovery. Investment in manufacturing plant and equipment has been stronger, but only regained the nominal dollar level of 1989 in 1996. Measured as a share of GDP, non-residential investment has fallen from 12 per cent in the late 1980s to the 10 per cent level in the 1990s, with much of the investment effort focused in the trade and financial sectors. Canadian manufacturers' historically poor record of investment in research and development and in skills has not appreciably increased.

While there is no doubt that exports - and manufacturing exports in particular - have led the weak recovery, there is very limited evidence that free trade has produced a structurally stronger manufacturing sector. To be sure, some sectors have invested significantly and have grown on the basis of exports - auto, aerospace, telecommunications equipment, software - but the overall competitive position of the manufacturing sector vis-à-vis the United States has been maintained only through depreciation of the exchange rate. The Conference Board of Canada has calculated that 80 per cent of the improvement in Canada's competitive position in the United States in the recovery has been based on exchange rate depreciation, with the remainder coming from slower nominal wage growth than in the United States, (see table 5).

The FTA and NAFTA have not resulted in a significant inflow of net new foreign direct investment into Canada. Overall, foreign direct investment inflows from the United States have about matched outflows since 1989. US FDI in Canada increased by $47 Billion or 61 per cent, 1988 to 1996, while Canadian FDI in the United States increased by $42 Billion or 82 per cent. Most of the inward flow to Canada represents reinvestment of earnings in the modernization of plant and equipment, while outflows represent new investments by Canadian companies in the United States or elsewhere. While there has not been FDI disinvestment in net dollar terms, the ratio of Canadian FDI in the United States to US FDI in Canada has risen from .67 to .75. The United States has been a more attractive locale for new Canadian corporate investment than Canada has been for US corporations. There have been very few new "greenfield" manufacturing investments in Canada by US corporations under the FTA and NAFTA but many major Canadian manufacturing corporations have established new facilities in the United States and, increasingly, in Mexico. Canadian FDI in Mexico doubled between 1993 and 1994 to $1 Billion, and rose to $1.3 Billion in 1996.

It is worth noting that the United States controlled share of operating revenues of Canadian corporations grew from 17.3 per cent in 1988 to 20.1 per cent in 1995 even though the United States share of assets grew only from 11.8 per cent to 11.4 per cent. This reflects the fact that export growth has been strongest in areas where US transnationals are dominant players - notably auto, aerospace, electrical and electronic products, chemicals and wood and paper.

Table 5. Growth of hourly wages and hourly labour productivity in manufacturing: Canada versus the United States (per cent)

8. Assessing the broader FTA impact on labour markets

Aside from the direct impact on jobs, critics forecast that the FTA and NAFTA would increase the bargaining power of capital vis à vis labour, resulting in slower growth of wages, possible deunionization, and a downward harmonization of standards. There is mounting evidence that this has indeed been the case, and that the FTA and NAFTA are, therefore, contributing to the dismal overall labour market trends described above.

As detailed in table 6, wages have grown significantly more slowly in relation to productivity in the manufacturing sector than in the business sector as a whole. Between 1989 and 1995, real wages in manufacturing rose by an average of just 0.2 per cent in both manufacturing and the business sector as a whole, even though productivity growth in manufacturing averaged 2 per cent per year compared to 0.8 per cent in the business sector as a whole. This suggests that competition from the United States and now Mexico has a significant impact upon the growth of wages in relation to productivity, since the manufacturing sector is much more directly exposed to competitive pressure from the United States and Mexico than is the business sector as a whole. There is some evidence of a growing gap between productivity and real wage growth in manufacturing which could be attributed to the FTA and NAFTA. Real wage growth as a proportion of real productivity growth in manufacturing has clearly fallen, though this has also been true for the business sector. It is difficult to determine trends in that workers are highly resistant to wage cuts, and concessions made in collective bargaining have taken the form of rollbacks on other issue such as benefits and work rules. This has shown up in the increased work-time noted above.

The data in table 6 indicate the intense downward competitive pressures on wages to which Canadian workers have been subjected by US competition. Despite reasonably strong productivity growth in US manufacturing, real wages of US manufacturing workers have barely increased in the 1990s.

Strikingly, workers in those manufacturing sectors identified as "winners" under free trade have not benefited in the form of higher wages. Trefler found "no link" between wage growth and export or productivity growth in different manufacturing sectors, and Schwanen found no link to more rapid wage growth in sectors with fast growing exports. Between 1988 and 1997 (January) average hourly earnings of hourly paid workers increased quite uniformly across the manufacturing sector, and rose no higher in the fast growing electrical machinery sector (30 per cent in nominal terms) than in manufacturing as a whole (33 per cent). Again, this suggests that competitive pressures have eroded the bargaining power of labour.

Table 6. Annual growth of real hourly wages and hourly productivity - Canada and the United States (per cent)

Erosion of worker bargaining power has resulted in a redistribution of income from labour to capital. In the post FTA period, corporate profits plunged in the recession, and then recovered strongly. Profits as a share of GDP have still not recovered to pre-recession levels, but this reflects the still depressed domestic economy. Statistics Canada has reported (in the Daily, 3 October 1996) that profitability for large corporations has returned to the peak levels of the 1980s and rates of return are very high in auto, electrical machinery and equipment, pulp and paper, and other winning sectors. In short, the income gains from the growing export sector have mainly been appropriated by shareholders.

Between 1988 and 1995, production worker wages as a share of value added in direct manufacturing activity fell from 31.2 per cent to 27.2 per cent and total salaries and wages in manufacturing as a share of value-added also fell, from 43.0 per cent to 37.4 per cent. The decline in labour's share of value-added has not been as significant in successful and profitable export sectors such as transportation equipment as in the manufacturing sector as a whole, as indicated in table 7.

The overall unionization rate has remained stable at 32 per cent (as measured by the CALURA survey) and unionization in the private sector has also remained stable. However, the unionization rate in manufacturing fell from 35.0 per cent in 1988 to 33.4 per cent in 1992. This may reflect the fact that heavily unionized industries have suffered disproportionately high job losses, perhaps in part due to the growth of contracting out to non-union firms (Gaston and Trefler).

Leading Canadian industrial relations analysts Meltz and Verma have summarized the impacts of the FTA on Canadian unions as follows:

"Some degree of union avoidance has always characterized Canadian management practices. What gives them a new flavor is the growing ability of employers to stay non-union in greenfield sites. But perhaps the strongest weapon that employers have used with success against unions and workers in the 1980s is the threat of closure. In response to increasing competition as a result of the Free Trade Agreement (FTA) with the United States and lower tariffs in general, a number of employers (especially, but not only, the United States manufacturing companies) began to wind up their branch plant operations in Canada. Even as these plants closed, other employers have missed no opportunity to point to these cases to win concessions or to defeat organizing campaigns.

Despite management resistance the Canadian labour movement as a whole is far from weakened in the way that the United States labour movement found itself circa 1980." (Meltz and Verma, 1995).

Canadian employers have extensively used the argument of international competitiveness vis à vis the United States and Mexico to press governments for changes in labour laws and regulations and social programmes. For example, the Canadian Manufacturers Association has proposed that all policies should be subject to a "competitiveness test"

Government now plays a more pervasive role in the economy than ever before. Tax rates and their coverage; tax expenditures and support programmes; public spending for social programmes and public infrastructure; regulation and administrative measures of many kinds all have a significant impact on the economic system of the country.

Table 7. Production worker wages as per cent value-added in manufacturing activity in selected industries

"Governments in other countries do the same thing, and every nation has its own unique mix of public and private programmes and public and private cost. While national sovereignty makes any such mixture possible the realities of international competition and economic interdependence reduce the range of choices which are feasible in practice.

Now more than ever, governments must recognize that their choices about taxation, spending and regulation cannot be made in isolation. Companies make decisions every day about where to invest, produce and employ. Selecting new locations around the world is becoming easier as time goes by. The government cost and regulatory burden is part of the environment which firms evaluate in making these choices. In this context, governments do compete, at all levels, for business, for jobs and for the tax revenues that support and strengthen our social fabric." (Canadian Manufacturers' Association, 1992).

While labour laws and employment standards have tilted in both directions since the FTA came into effect, the recent trend - notably in Ontario, Alberta and Manitoba - has been to severely limit the effective right of workers to organize, and to roll back even basic employment standards. While employer acceptance of the legitimacy of collective bargaining and employment standards has always been tenuous, the depth of opposition to unionization has grown and this reflects, at least in part, the greatly increased pressures of international competition.

The shift against the bargaining power of organized labour attributable in part to the FTA is a major factor behind the overall labour market trends described above, notably casualization of work and increased polarization of incomes and working-time.

9. CONCLUSIONS

Advocates of the FTA and NAFTA drew on the neo-classical theory of mutual gains from free trade to argue their case. By contrast, it was argued above, that integration of trade and investment in a context of high unemployment and international capital mobility open the possibility of significant losses, particularly to workers, as a result of intensified international competition. In such a context it was argued that bargaining power in the labour market will shift in favour of capital.

The above analysis detailed dismal developments in the Canadian labour market and wider society in the post-FTA era - high unemployment, increased inequality and insecurity, and erosion of the social safety net. It was argued that macro-economic policy underlay many of these developments, but links between economic integration and contractionary macro-policy were also noted. It was also argued that there are direct and demonstratable links between the FTA and NAFTA and job losses, stagnation of real wages relative to productivity, the erosion of union bargaining power, and downward harmonization of the Canadian social safety net relative to that in the United States. Finally, it was noted that the expected efficiency gains of the FTA and NAFTA have not materialized - undercutting the possibility of providing assistance to "losers" from the adjustment process. For workers, closer continental integration has meant job losses, and little or no real wage growth, even in "winning" sectors. Meanwhile, some sectors of business have done well under conditions of continental economic integration after a major process of re-adjustment.

The major policy issue that arises from this analysis is how to ensure that the possible gains from closer trade and investment ties are equitably shared. This requires that labour's bargaining power not be diminished through the integration process. For working Canadians, this is a particularly key issue since labour market and welfare state institutions have been generally superior to those in the United States and Mexico. Unfortunately, the erosion of labour's position in Canada has not been offset in any significant way through the development of new regulatory instruments and institutions at the NAFTA level, since the FTA/NAFTA integration process has been almost exclusively one of deregulation.

In recent years, the Canadian labour movement has pressed strongly for "social clauses" to promote and protect labour rights in all trade and investment agreements, including the WTO and the proposed Free Trade Agreement of the Americas (FTAA). At the same time, the movement has remained strongly critical of purely liberalizing agreements. The Canadian dilemma is that economic integration in North America has undercut national labour and social standards, but has provided little or no basis for reconstituting those standards at a regional level (not least given political realities in the United States.). Regional integration in the Americas will continue to be much more unbalanced than in Europe.

III. NAFTA's impact on US labour markets, 1994-1997

1. Introduction

The backdrop against which we hope to assess NAFTA's effects on US output, trade, employment and wages is that of a decline more than two decades long in labour's share of productivity gains. Real hourly compensation in the United States has stagnated since 1978, and income inequality in the United States notably worsened during the 1980s and much of the 1990s. Workers at lower skill levels have been particularly hard hit, as the wage penalty for the lack of a college education has grown⁽²⁴⁾ (Wood 1995). At the same time, in 1995 after-tax profits for US manufacturing corporations reached 16.1 per cent as a percentage of stockholder's equity, their highest level since 1988 (ERP 1997: Table B-92.).

While there is dispute over the size and nature of the role that trade (or the relocation of production that gives rise to trade) has played in reducing labour's bargaining power in the United States, it is clear that it has played some role. The world labour market has become more competitive, and it is harder than before for US manufacturing workers to maintain their wages and working conditions, despite the fact that the value of their output has continued to grow. The expansion of production in China alone has led to massive US imports from that country of clothing, toys, tools, household furnishings and other goods, produced at wages with which US workers cannot possibly compete.

The debate among economists over the main causes of increasing wage inequality between more- and less-educated workers in the United States during the 1980s and 1990s has been (inappropriately) narrowed to two possible causes: 'technology' and 'trade'. By 'technology' is meant skill-biased technical change, which increases the demand for skilled labour relatively to less-skilled labour. By 'trade' is meant import penetration which forces domestic low-skilled labour to compete with low-skilled, low-wage labour in other countries; this includes out-sourcing by US-based multinational firms. Wood (1995), Freeman (1995) and Richardson (1995) provide useful surveys of the issues; among these, Wood is particularly persuasive that the evidence is best interpreted as implying a large role for trade in causing trends in wage inequality in the United States and other developed countries.

But labour market institutions, and in particular the long-term decline in the real value of the United States minimum wage, surely also play a role. More generally, the advance of the neoliberal programme - even apart from NAFTA - is central. Globalization, in other words, is not simply technologically driven, or outside the political decision-making process. On the contrary: a series of changes in laws and institutions, under the broad rubric of market-oriented reforms, has contributed to the weakness of labour's bargaining power, more so in the United States than in other industrialized countries. For example, it has been suggested that the overall long-term decline in the value of the United States real minimum wage is one factor undermining the average real wage: expressed in 1995 dollars, the minimum wage peaked at $5.97 in 1979 and steadily declined to $4.12 in 1989; several increases since then have raised its value, but only to just over $5.00 an hour. While there is dispute over the main cause of the decline in the real wage in the United States, in countries with stronger labour institutions the trend in the real wage has been more favourable to labour (see, e.g., OECD 1993:166). Some other factors accounting for labour's weakness are the erosion of eligibility for unemployment benefits during the 1980s; the rise of part-time and temporary work that undermines organizing efforts; the determination of the Federal Reserve throughout much of this period not to allow the unemployment rate to fall 'too low'; and the steady decline in the percentage of the workforce represented by unions, from 24 per cent in 1979 to just 15 per cent in 1994 (Mishel et al. 1997).

Meanwhile, national and local governments have tried to increase the advantages of capital, by removing regulations on capital movements, and granting tax concessions to firms, on the premise that this will make them more competitive and lead to greater prosperity. This process of advantaging capital includes various rules written into NAFTA, but it also includes the Uruguay Round of the GATT and the creation of the World Trade Organization (WTO), and it includes the Multilateral Agreement on Investment being drawn up by the OECD, and in the United States Congress, the Republican agenda of deregulation and tax breaks for business.

NAFTA, then, is just a piece of a much larger picture. But because of the fierce debate over its passage, and the extravagant claims made both for and against it, as well as the mobilization against it by labour, environmentalists and community groups, it has in a sense come to represent globalization itself. This is surely because it was and is one of the few ways that the public can get a handle on globalization and exercise some control over its advance. Globalization as such will never be put to a democratic vote; but NAFTA was a concrete proposal that was to be voted up or down by the United States Congress, and related proposals such as fast-track legislation have the same status. Hence in the public and legislative eye the question of whether NAFTA has had devastating effects is a crucial one.

Based on the best estimates available, NAFTA's effects on labour in the United States overall appear to have been negative, moderate in size, and harsh in some sectors and regions. For example, job losses have been substantial in the apparel sector, in El Paso, Texas, and in some small communities heavily dependent on factories which shut down and moved to Mexico. These effects should not be minimized, and the workers affected ought to have received full compensation for their losses, but did not. Overall, however, the fact that NAFTA's first three years occurred during a business cycle upswing in the United States cushioned its effects. Some of its effects have therefore been a reduction in the rate of creation of new jobs, rather than an increase in actual layoffs. In a growing economy, the duration of unemployment for those affected has also been shorter than it would have been in a declining economy. The question suggests itself whether the United States business cycle upswing was itself accelerated or facilitated by NAFTA; existing studies differ on whether there is any evidence that this is so, but the most detailed study, by the United States International Trade Commission (ITC 1997), finds no significant statistical evidence of a net change in GDP due to NAFTA during 1994-1996.

This paper mainly considers impact on employment, with brief consideration given to impact on wages and on labour rights later in the paper. (Environmental considerations and broader social issues are dealt with in other parts of the larger report.) The employment impact may be thought of in two conceptually different ways, which are often confused in public discussion. One is the net employment impact: the difference between the number of jobs which actually existed (say, at the end of 1996), given that NAFTA was implemented, and the number of jobs which would have existed if NAFTA had never been implemented. It is this kind of number that was predicted by some studies of employment impact done before 1994⁽²⁵⁾. It includes three things: jobs created due to NAFTA, jobs actually eliminated due to NAFTA, and jobs which failed to be created due to NAFTA (but which in the absence of NAFTA would have been created).

The second concept is the gross employment impact, perhaps better described as the number of displaced workers. This concept includes only the middle of the three categories that make up the net employment impact, that is, jobs actually eliminated.

Of course, the number of jobs affected is not only a consequence of NAFTA but of labour's response. In some cases, by making concessions to employers over wages and working conditions, workers have managed to keep jobs in the United States, at least for some period, so that the impact of NAFTA is felt in labour incomes and working conditions rather than in job losses per se. Likewise, displaced workers may be displaced for shorter periods if they are willing to accept lower-wage jobs, and in that case the impact is more on incomes than on employment. Thus, strictly speaking, employment impact makes most sense if calculated in a situation in which real hourly compensation is constant. Some pre-NAFTA studies made predictions of employment impact in exactly this way. However, no post-NAFTA assessments of employment impact have been conducted in this way or have attempted to calculate the income impact of displacement of workers. What follows tries to distinguish clearly between net and gross employment impact; the whole issue is discussed in greater detail later in the paper.

Among the studies to be reviewed here, the best estimates available indicate that NAFTA to date has brought a nearly zero net gain in output in the United States, probably together with a small percentage overall loss to US labour income and a gain of a comparable dollar amount to property incomes. While the December 1994 peso crisis brought in its wake a severe worsening of the United States trade balance with Mexico, implying a significant negative impact on US employment creation, in the context of a much larger expanding US economy this impact has fortunately not been devastating.

This is not to say that the usual mechanisms creating 'gains from trade' have been absent. Rather, gains from trade are typically thought of in the context of an assumption of full employment before and after trade opens up - or, in practice, of the same level of employment (apart from growth of the labour force) before and after trade liberalization. But I will argue that NAFTA brought with it a set of political pressures which led to an exceptionally severe currency crisis in Mexico, and thus the macroeconomic difficulties it helped to create partially offset the gains from trade over the few years NAFTA has been in existence.

Assessments of NAFTA released in late 1996 or during 1997 range from defining it largely as a success (USTR 1997) to defining it as a clear failure (EPI et al. 1997). Between these two polar views, several extensive studies find that though NAFTA has had significant effects on trade, it has had remarkably small net effects on output and employment, except in a few sectors (Hinojosa et al. 1996; ITC 1997).

The study by Weintraub (1997) takes a somewhat different tack. Weintraub washes his hands of the debate over employment effects almost entirely, arguing that different criteria should be used to assess NAFTA, such as the increase in total trade and investment, the increase in the rate of growth of productivity and wages, effects on the environment, and the extent to which the institutions of economic integration are developed. Though he expresses the usual concern for compensation and retraining of displaced workers, he argues that bilateral trade balances, typically used as the basis for calculation of employment effects, fluctuate for all sorts of reasons and are not a sound basis for evaluating NAFTA's effects. While his observations are important, insufficiently precise data are available on many of the criteria he suggests (productivity, wages) to make an adequate assessment at this time; or, to put it another way, there is no statistically significant evidence of overall productivity or wage increases or decreases due to NAFTA, according to ITC (1997). And the increases, say, in total trade - for which data are available - hardly seem an appropriate category for a definitive assessment of NAFTA's success.

We proceed by briefly summarizing theory and previous studies on the effects of trade liberalization, and then turn to a detailed discussion of the recent NAFTA studies.

2. Pre-1994 predictions of NAFTA's effects

It is a common public misconception that neoclassical theory holds that, always and everywhere, free trade is good for everyone. Nothing could be further from the truth. While the public cannot be blamed for drawing this conclusion from hearing the repeated refrain, 'win, win win', the NAFTA debate did not in fact provide an accurate representation of what trade theory says.

Trade theory contains many different models, with different sets of assumptions. Among them, the most basic textbook model (the Heckscher-Ohlin) says clearly that when a high-wage country opens up trade with a low-wage country, workers in the high-wage country will suffer a fall in their real wages. In models which categorize workers into at least two skill levels, low-skilled workers in the high-wage, high-skill country may experience a decline in wages relative to workers with higher skill levels.

It may be surprising, then, that in the early 1990s studies predicting NAFTA's effects sometimes found that hardly any category of the population would lose. One reason for this result is that these studies typically assumed that capital flows as well as trade effects would occur: there would be a substantial increase in the inflow of investment into North America from outside the continent, and this often was the main source of the beneficial effects. It was not just the free trade among NAFTA countries, but the protectionism against non-NAFTA countries, that would create this tariff-jumping investment. Firms would switch production from some Asian countries to Mexico because Mexico offered similarly low wages, but without the necessity of paying a tariff on goods entering the United States. To some extent this has occurred in practice; for example, ITC (1997:6-58) reports that about 80 German auto parts firms have established Mexican production in recent years to enable Volkswagen to meet NAFTA's rules of origin requirements for exemption from tariff. However, the size of the contribution to North American income depends in part on what fraction of profits are repatriated to non-NAFTA countries. Another reason for optimistic projections was the assumption in some studies of substantial scale economies. In an econometric analysis of data on Mexican production, Tybout and Westbrook (1995) have challenged this assumption, however, at least for Mexico.

Studies predicting NAFTA's effects typically found that the effects of tariff liberalization alone were likely to be small for several reasons. For Canada, of course, a free trade agreement already had been implemented beginning in 1988. For Mexico, the maquiladora programme, in existence in various forms since 1965, permitted US firms to locate assembly production south of the Mexican border and be exempted from Mexican tariffs on intermediate and capital goods entering Mexico for use in those plants; later the same opportunity was provided throughout Mexico. About one-third of the value of US imports from Mexico entered duty-free before NAFTA. The majority (about one-half the value of about 45 per cent of US imports from Mexico in 1989) entered under the production-sharing provisions of the United States tariff code, HTS 9802, which allows US firms to assemble goods abroad from US components and re-import them into the United States, paying tariff only on the value added abroad (ITC 1991:1-5; ITC 1997). A smaller share (about 9 per cent in 1989) entered duty-free under the Generalized System of Preferences (ITC 1991:1-5). In addition, of course, NAFTA was expected to have a larger impact on Mexico's economy than on the US because of their relative sizes.

3. Recent assessments of NAFTA

In late 1996 and in 1997, a number of studies were released which attempted to assess NAFTA's effects during its first three years or so. Among relatively substantial studies, the most positive assessment has been by the office of the United States Trade Representative (USTR 1997), whose study was endorsed by President Clinton and submitted as his required report to the US Congress on the first three years of NAFTA's operation. The USTR report draws heavily on other, more detailed studies, such as the voluminous report by the United States International Trade Commission (ITC 1997), to some extent on the preliminary report by Hinojosa et al. (1996), and on several other studies much more limited in scope, such as DRI (1997), Kouparitsas (1997) and Gould (1996).

At the opposite pole, a study by the Economic Policy Institute (EPI et al. 1997, hereafter 'EPI 1997') asserts that NAFTA is a failure, because it has failed to promote high and rising living standards for the great majority of the population. While it recognizes that NAFTA has not been the primary cause of declining living standards in North America, it asserts that NAFTA has made matters worse. My assessment is that their analysis is largely correct. Moreover, in actuality little or nothing in EPI's assessment conflicts with the data or even many of the conclusions (if carefully understood) which are presented in other studies.

Between these two opposing views, several substantial studies argue that the net effect of NAFTA on output and employment has been very small, some say effectively zero. Apart from a nearly zero net effect, these studies do recognize that some gross job displacement has occurred in the United States. ITC (1997) summarizes its findings by saying that in general NAFTA had positive but modest effects on the United States economy. This conclusion is puzzling, since for the United States it finds no effects on GDP or its growth rate, no aggregate employment or earnings effects, and no effects on aggregate investment, and the study notes that it was unable directly to analyse productivity effects. It is unclear, then, what are the positive effects which it finds 'in general' - perhaps the reference is to increased trade or to the 'improvement in the business climate' or other effects it describes as not easily quantified or observed. Hinojosa et al. (1996) find that there has been a very small net positive effect on US employment from the changes in trade which occurred during 1994-96.

In addition, we consider a few other studies whose analyses are far more limited (Gould 1996; Kouparitsas 1997; DRI 1997), but which need to be discussed since they are cited in USTR (1997) as evidence of NAFTA's positive effects. Essentially, I judge that none of the conclusions of these limited studies is reliable, either because of errors in statistical methods, or because the data series used were too short and yielded results with a low level of statistical significance, or because the model was conceptually flawed. The finding of DRI (1997) that NAFTA, apart from the peso devaluation, had a small positive effect on US exports to Mexico (and possibly a smaller positive effect on US imports from Mexico) must be understood in the context of the study's assumption that NAFTA and the peso crisis were completely independent events. The main problem with these studies is that, as I believe, the effects of the peso crisis are partially attributable to NAFTA and the political process set in motion by the NAFTA debate, and these authors instead treat NAFTA as completely separable from the peso crisis.

The study by Weintraub (1997) is of a different nature, sidestepping the dispute over short-term effects on output and employment, and arguing that these are essentially irrelevant. Weintraub urges that NAFTA be judged by outcomes such as the amount of trade created (from which should be subtracted trade diversion from outside North America), the amount of new investment created, the degree of further specialization achieved, the increase in productivity and competitiveness and real wages. In addition, he says the effect on the environment should be considered - but not the number of jobs displaced which, although a concern, according to Weintraub, should not be central to the debate.

4. Changes in trade, output, and employment: Aggregate and sectoral trends

US total trade with Canada and Mexico has been growing rapidly, while the trade balance with both countries sharply worsened during 1994-1996. From 1993 to 1996 US exports to Mexico grew 35.8 per cent, from $40.3 billion to $54.7 billion, while US imports from Mexico grew 91.8 per cent, from $38.7 billion to $74.2 billion, as shown in table 1. The more rapid growth of imports was due largely to the peso devaluation in December 1994, causing large US trade deficits with Mexico of $16.8 billion in 1995 and $19.5 billion in 1996. During 1993-96 US exports to Canada grew 29.7 per cent, from $91.9 billion to $119.1 billion, while US imports from Canada grew 41.5 per cent, from $110.5 billion to $156.3 billion (ITC 1997. The United States trade deficit with Canada thus grew from $18.6 billion to $37.2 billion over the period.

Table 1. Total US trade with NAFTA countries, 1994-96

These increases in US trade deficits with both its NAFTA partners are the basis for the largest estimates that have been made of US job losses. EPI (1997) estimates total US job losses (jobs eliminated, plus jobs that failed to be created) at 420,000 as a result of the worsening trade balance. Since not all of these represent actual jobs eliminated, the number of displaced workers is smaller. In 1994 a NAFTA Trade Adjustment Assistance programme (a programme of compensation and training similar to the existing general Trade Adjustment Assistance programme) was created to assist workers displaced by trade and investment among NAFTA countries. The number of workers certified under the NAFTA-TAA as impacted by trade with Mexico or Canada, or relocation of production to either country, was 134,492 as of mid-July 1997. Because the number of certifications is in some respects an underestimate and in other respects an overestimate of those affected specifically by NAFTA, we believe it is a good rough approximation to the number of workers actually displaced. The issue of the effect of changes in the trade balance on employment is controversial, and will be discussed at length below. Our conclusion will be that, taking account of the ways in which various methods may overstate or understate the employment impact, the EPI estimate of the larger category is in the ball park, if likely somewhat high. However, it is important to understand the differences in the categories used in different studies to analyse employment impact, and this, too, will be clarified below.

Two sectors - auto and auto parts, and apparel - account for most of the United States trade deficit with Mexico and a significant fraction of the trade deficit with Canada as well. Among sectors prominent in NAFTA trade, none is larger than the motor vehicle and vehicle parts sector. This sector accounted for approximately one-fourth of total US trade with Mexico and Canada in 1996, as shown in table 2. In 1996 it accounted for about one fifth of all US exports to NAFTA partners ($35.8 billion out of $173.8 billion in total exports to Mexico and Canada), and over one quarter of all imports ($62.3 billion out of $230.5 billion in total imports from Mexico and Canada). The predominant pattern was for parts to be exported from the United States, assembled or otherwise processed, and then returned to the United States as imports. In fact, some of the trade recorded as US exports of textiles, leather and electronic components, to name a few, also consists of goods destined for auto plants, as seat belts, fabric or leather for seat covers, and electronic components in autos. The United States trade deficit in vehicles and vehicle parts with its two NAFTA partners was $26.5 billion in 1996, accounting for nearly half the total US trade deficit with Mexico and Canada.

The sectoral trade deficit for vehicles and parts together worsened by $12.2 billion during 1993-96, not only because of the peso devaluation, but also because a decline in Mexican demand for autos during the peso crisis caused automakers in Mexico to export a much larger share of their product than before. Among other things this allowed them to retain workers in whom they had invested considerable time and money in training (as well, of course, as to produce the vehicles more cheaply than in the United States). Due to expansion of US domestic output in the sector in a macroeconomic boom, however, these changes did not result in a net loss of employment in the sector; rather, US employment in the two sectors combined grew from 910,000 in 1993 to 1,035,000 in 1996 (ITC 1997: xx, 6-52-3, 6-60). At the same time, however, from 1994 through mid-1997, 9,663 US workers were certified under the NAFTA-TAA programme as impacted by trade in auto and auto parts with Mexico and Canada, or relocation of production to either country (DOL 1997).

The apparel sector also accounts for a large share of trade with Mexico and Canada, and has in addition experienced a large number of job displacements due to changes in this trade, largely derived from relocation of production. These were concentrated during 1995-96, after the peso devaluation reduced Mexican labour costs sharply. As shown in table 3, during 1993-96 US imports from both Mexico and Canada nearly doubled, while US exports to both countries grew more slowly. In 1996 US apparel exports to Mexico were $2.0 billion while imports from Mexico were $4.5 billion. Sectoral trade with Canada was more nearly balanced, with $0.8 billion in exports and $1.2 billion in imports. The United States trade balance in apparel worsened with Canada by $0.4 billion and with Mexico by $1.4 billion form 1993 to 1996. During this same period the total number of US apparel production workers fell by 130,000 to 698,000, a decline of 16 per cent (ITC 1997: 6-22).

Table 2. NAFTA vehicle and vehicle parts trade, 1993-96

Table 3. NAFTA apparel trade, 1993-96

ITC (1997) notes that the growth in exports to Mexico consisted mostly of garment parts for assembly. Movement of apparel production to Mexico using US-formed and US-cut fabric accelerated, in part because NAFTA contained a specific expansion of US production-sharing opportunities under HTS 9802. Under this section of the US tariff code, US firms were already able to send US components abroad, have them assembled, and re-import them into the United States, paying tariff only on the value added abroad. The new provision made the re-importation both duty-free and quota-free from Mexico. It also allowed processing, such as stone-washing and wrinkle-free processing, in Mexico, although if done in other countries, these would invalidate the tariff exemptions on the resulting goods when re-imported into the US Industry officials told the ITC that these provisions were the most important effect of NAFTA in their decision to produce in Mexico (ITC 1997: 6-25). Where the increase in US exports consisted of garment parts for processing and assembly in Mexico, and simply was the continuation of an existing supplier relationship, of course the increase in demand for these US exports was matched by a decline in domestic demand for the same goods, and implied no new job creation in the United States.

From 1994 through mid-1997, 29,247 apparel workers were certified under NAFTA-TAA as displaced by trade with Mexico and Canada, or by relocation of production to either country. Some of the growth in US apparel imports, however, was due to diversion of production from Asian or Caribbean countries to Mexico. During 1993-96, Hong Kong's apparel exports to the United States declined by 1 per cent, The Republic of South Korea's declined by 39 per cent, and Taiwan's by 11 per cent. Apparel imports from China rose by only 2 per cent over the same period (ITC 1997:6-24).

While ITC reports an increase in the average wage in US apparel, it cautions that the 9 per cent increase is explained by the change in composition of the workforce, that is, the exodus of lower-paying apparel assembly jobs, leaving behind the higher-skilled and higher-paying jobs in cutting.

Another major sector which accounts for a large share of NAFTA trade, and for a portion of the worsened US trade deficit with its NAFTA partners, is electronic components and equipment, for which data are given in table 4. The United States has exported an increasing quantity of electronic components to maquiladora and other enterprises in Mexico as inputs into radios and televisions, computer and peripheral equipment, communications equipment, and so forth. Net US exports of electronic components to Mexico increased by $1.7 billion from 1993 to 1996, while net imports from Mexico of computers and peripheral equipment increased by $1.4 billion, net imports from Mexico of radio and television equipment increased by $3.1 billion, and net imports from Mexico of communications equipment increased by $0.3 billion (ITC 1997). Electronic components are also used in auto and auto parts production. While the data indicate a worsening of the United States trade balance with Mexico in these combined sectors, Hinojosa et al. (1996) note that the negative impact may be somewhat greater than it seems. This is because some of the electronic components recorded as exported from the United States are actually goods that were produced in Asian countries, sent to warehouses in the United States and then re-exported to Mexico without any further processing in the United States. However, although this point is noted in their text, it was not incorporated into their numerical estimates of NAFTA's employment impact.

Table 4. NAFTA electronic components and products trade, 1993-1996

Nine sectors were identified by ITC (1997) as ones in which trade among NAFTA countries increased significantly relative to the size of the sector during 1994-96, including several not mentioned above. These sectors were vehicles, vehicle parts, apparel, textiles, women's footwear, appliances, leather, grains and cotton. The nature of these effects (whether the increase was primarily in US exports or imports), as well as the reasons for them, varied among these sectors.

5. The main points of agreement and contention

Reading pro- and anti-NAFTA authors' assessments of the last three years is much like reading the six blind men's reports on the elephant - with one exception: even where they are groping the same part of the animal, one could hardly tell it from reading their accounts. Yet on closer examination, the differences among these studies are somewhat smaller than they seem. True, the reports written from the two most diametrically opposed viewpoints (USTR 1997; EPI 1997) use language that seems crafted to maximize the apparent differences. On casual reading, for instance, one would think that USTR (1997) claims creation of 90,000 to 160,000 jobs in the United States over the three years of NAFTA. But contrary to appearances, the USTR report does not say this; in fact, remarkably, it does not explicitly claim any job creation by NAFTA. Instead it uses a category called 'jobs supported by exports', which is different from 'jobs created', and will be discussed below.

In contrast, EPI (1997) estimates 420,000 net jobs lost (due to worsening of the trade balance with both Mexico and Canada). However, this category includes job opportunities not created, a category which USTR (1997) does not consider at all. Thus the total number of actual layoffs that resulted from NAFTA is judged by EPI to be less than the 420,000 figure. EPI points out that the number of workers who applied for and were certified to receive NAFTA Trade Adjustment Assistance (NAFTA-TAA) is in some measure an underestimate, since some workers who were affected very likely did not apply for assistance, and some who applied were actually affected, but were not certified because of the restriction of the programme to goods-producing workers, or for other reasons. The report does not, however, draw a conclusion about how many persons actually were laid off due to NAFTA, and notes that the prolonged strong business cycle upswing in the United States has obscured for the present the negative effects of NAFTA on actual employment.

There are two main bones of contention between opposing views on NAFTA's effects. The most central is disagreement over whether NAFTA and the peso crisis were completely independent events, whose effects can be statistically disentangled, or whether NAFTA was a cause of the peso devaluation and crisis (as EPI (1997) and Blecker (1997) argue). The majority of studies have assumed that NAFTA had nothing to do with the peso crisis, but have typically not explained or defended this view. These studies use statistical methods to factor out the negative effects of the peso crisis and then attribute the remaining positive effects on US employment to NAFTA (Gould 1996; DRI 1997). ITC (1997) takes a slightly different approach, defining sectoral effects as attributable to NAFTA only if in their statistical analysis the effect is significant in all three years 1994-96, and in addition is either positive in all three years, or else negative in all three years. Although a peso devaluation of some amount was unavoidable, NAFTA and the political process entwined with NAFTA turned what could have been an orderly, smaller devaluation into a major crisis; hence NAFTA is an important cause of the effects attributed to the devaluation.

The second bone of contention is how changes in US exports and imports affected employment. This controversy centres on the appropriateness of using the Department of Commerce export-jobs multiplier, particularly on whether it should be applied to imports. These issues will be discussed in detail below.

6. NAFTA and the peso crisis

What causal relationship exists between NAFTA and the peso crisis? USTR (1997), DRI (1997) and several other studies assume implicitly that there is no causal relationship, and hence in essence give NAFTA credit for positive effects on US output and employment in 1994, while blaming the peso crisis for the worsening of the trade balance in 1995 and any resulting US job losses. In contrast, NAFTA opponents tend to assume, suggest or assert that all effects that occurred after NAFTA was implemented on January 1, 1994 - including the entire peso crisis - are consequences of NAFTA.

The evidence shows that NAFTA was an important cause of the peso devaluation and crisis, but not the only cause. Hence part of the effects of the peso crisis should be attributed to NAFTA, a fraction difficult to determine.

Mexico is no stranger to episodes of overvaluation of the peso, and eventually balance of payments crisis, followed by devaluation. Such episodes (along with other causes) led to devaluations in 1976 and 1982, and another series of devaluations occurred in 1985-1987, though each episode had somewhat different causes. The most recent overvaluation also had historically particular features. The problems of overvaluation in the early 1990s, and the potential seriousness of its consequences, were exacerbated by several factors related to NAFTA. One was the enormous inflow of capital, particularly portfolio investment, into Mexico in 1991-93 in anticipation of gains after NAFTA was passed. This capital inflow was partly caused by Mexico's huge sell off of state enterprises to the private sector - including foreigners - but it was also a consequence of the euphoric atmosphere created by pro-NAFTA official and unofficial pronouncements. Of course, liberalization of capital inflows, such as Mexico implemented in 1989, may cause currency appreciation with or without a NAFTA. This has happened in a number of countries in Latin America (see Agosin and Ffrench-Davis 1995). However, the capital inflow into Mexico was on an unprecedented scale, reaching in 1993 the astonishing level of $29 billion (Banco de Mexico 1996), and the overvaluation of the peso proceeded apace. This overvaluation, among other things, improved the United States trade balance with Mexico, helping lend credence to the claim by NAFTA supporters that US exports would grow under NAFTA.

Moreover, Wall Street investment firms such as Goldman, Sachs and others became deeply involved with Mexico, to the extent that in March and April 1994 they took an active role in prolonging the overvaluation of the peso and building up the financial house of cards which collapsed catastrophically in December 1994. According to a June 1994 Wall Street Journal report (Torres & Vogel 1994), a group of investors reacted with alarm when the assassination of the PRI Presidential candidate Colosio in March led to a sudden sell off of peso-denominated assets, forcing Mexico to use up $10 billion in foreign currency reserves to defend the value of the peso (Weintraub 1997: 55). The investment group urged the Mexican government to take several steps, including assuming much more exchange-rate risk on new debt by issuing a large quantity of securities called tesobonos indexed to the US dollar (Torres & Vogel 1994).

The government proceeded to do this, and as a result, in December when the peso collapsed, the burden of servicing the tesobono debt was far greater than it would have been to service an equal value of debt not indexed to the value of the dollar; thus the crisis was deeper than it might otherwise have been.

The hype about the marvels of economic integration probably also played a role in the motivating US bankers to extend substantial loans to private enterprises in Mexico, and many firms took on a heavy burden of dollar denominated debt (Danby 1997), which after the devaluation became far more burdensome to service, contributing to the prolongation of the crisis.

In short, the hype during the NAFTA debate, together with the role played by US investment firms, both support the argument that NAFTA, and the political and economic processes set in motion by the NAFTA proposal and debate, contributed substantially to the seriousness of the peso crisis. The fall in the value of the peso was evidently much deeper than it might have been absent NAFTA; the fragility of the debt structure caused interest rates to rise higher, and hence consumer demand and Mexican imports from the United States to fall to a lower level than they might otherwise have done; and the result was a larger worsening of the United States trade balance with Mexico than might have occurred in the absence of NAFTA. For these reasons the assumption by USTR (1997) and other authors that NAFTA had no causal role in the peso crisis is indefensible.

While we cannot attribute an exact fraction of the peso crisis to NAFTA, if even half of its effects flow from NAFTA, assessments by USTR (1997) and others cited in it that NAFTA had a net positive effect on US employment would likely have to be reversed.

7. Trade and employment

In addition to the controversy over the relationship between NAFTA and the peso crisis, there is a second bone of contention: how have changes in exports and imports affected US employment? Several main methods have been used to derive the employment impact, none of which is free from controversy. Before we discuss these, however, it is useful to provide a framework for analysis.

Whenever there is a change in the demand for labour, its impact may be felt either on employment levels, or on the real wage level, or else partially on employment and partially on real wages. Pre-1994 studies forecasting NAFTA's impact sometimes estimated the impact of a change in labour demand in two different ways; one was to assume the impact occurred entirely on wages, with the employment level fixed (the 'full employment' assumption); the other was to assume the impact occurred entirely on employment, with wage levels fixed. The assumption was that if both employment and wages were affected, the net effects would lie within these bounds.

In the short run the net effects are likely to be more on employment than wages, at least in a country like the United States which has unemployment insurance. As many displaced workers eventually accept new jobs at lower wages, the initial employment impact is converted into a wage impact.

Apart from this, there is gross job displacement, which is typically greater than the net employment effect. Gross job displacement could easily be larger than the upper bound on net employment effect, and imposes burdens on displaced workers, as well as often on the communities in which they live.

In practice the effects of NAFTA are overlaid onto business cycle fluctuations in output, trade, employment and wages. Since 1994-96 saw a business cycle upswing in the United States, we would expect that NAFTA's net employment and wage impact would show up more as jobs that failed to be created than as jobs actually eliminated, and that the wage impact could appear as a slower rate of growth of real wages than is normal in an economic boom, rather than an actual decline. We begin by describing and assessing several methods of estimating net and gross employment impact, and then turn in a later section to the question of NAFTA's wage impact.

Several main methods have been used to derive the employment impact of NAFTA, and none is free from controversy. We consider first the positive employment impact of an increase in exports; the next section discusses the negative employment impact of an increase in imports.

8. How do changes in US exports affect employment?

The most widely used method of deriving the positive employment impact of an increase in exports is the US Department of Commerce export-jobs multiplier. For example, this ratio for 1992 was that a $1 billion increase in exports would create 16,532 jobs (Scott 1997). The number of jobs per billion dollars would be slightly smaller for later years, due to inflation. This multiplier, while accurate when the increase in exports truly represents an increase in demand for US products, is subject to two errors.

First, the multiplier is not based on US exports to a specific country, but on US exports overall. Thus if US exports to a specific country, like Mexico, differ in an important way from US exports to the rest of the world, the employment multiplier could give inaccurate results. While we have no reason to think this is true in general, the Hinojosa et al. (1996) study suggests that US exports of electronic components to Mexico (valued at $4.5 billion in 1996, or 8.3 per cent of total US merchandise exports to Mexico (ITC 1997:3-30,6-186)) may have an unusually low aggregate US domestic content. Hinojosa et al. conducted interviews with electronic industry personnel, who explained that many of these electronic components are actually trans-shipped through the United States, rather than produced in the United States. Asian producers set up warehouses in California to which they ship parts; these parts are then shipped to Mexican assembly plants on a Just-In-Time basis, and are recorded as US exports to Mexico. If such phenomena are significant in United States-Mexico trade, the export-jobs multiplier would overstate how many jobs are created in the United States due to increased US exports to Mexico. It is difficult to determine the extent of the problem, since US data do not distinguish between US exports which are trans-shipped goods, and US exports which have some US domestic content (Hinojosa et al. 1996).

A second and more important problem with the export-jobs multiplier is that US exports can rise for a variety of reasons, not all of them signalling an increase in US output or employment. It is sometimes the case that the increase in export demand is matched by a decline in domestic demand, because a domestic buyer has moved abroad. For instance, when an apparel maker moves from the United States to Mexico, and continues to buy the same quantity of fabric from the same US supplier, made by the same workers as before, this fabric now is recorded as an increase in US exports, but has generated no new US output or employment. Instead, the increase in export demand was exactly offset by an equal decline in domestic demand, as the purchaser moved abroad. If situations of this kind account for a significant fraction of the 1994-96 increase in US exports to Mexico, then application of the export-jobs multiplier to this situation will seriously overstate the resulting creation of new jobs in the United States. There is ample reason to believe that such situations do account for a substantial fraction of the increase in US exports. For many sectors, the detailed sectoral analyses provided in ITC (1997) observe that trade (at least between the United States and Mexico) is generated primarily by 'production-sharing', that is, sending inputs from the United States to assembly or processing plants in Mexico and then re-importing the finished products into the United States. The same sectoral analyses also observe that in many sectors firms have relocated production to Mexico to take advantage of low wages, particularly in 1995-96 after the peso devaluation reduced wage costs initially by about half.

Moreover, the business press regularly reports relocation of production from the United States to Mexico. One example is the announcement in January 1997 by the apparel maker Guess, Inc. of a decision to reduce the percentage of its jeans produced in Los Angeles from 75 per cent to 35 per cent, and to shift the balance to Mexico to save an estimated $1.50 to $2.00 per pair of jeans in labour costs (Scheeres 1997). In addition, the maquiladora sector in Mexico expanded at a breakneck rate during 1995-96, with employment rising from 542,074 in 1993 to 754,858 in 1996, a 39.3 per cent increase, after the peso devaluation reduced labour costs initially by about half; and more than 100,000 jobs have since been added (INEGI 1997). The maquila sector imports nearly all its inputs, typically from the United States, and exports its output, typically back to the United States. To the extent these plants represent production relocated from the United States, then, using US-produced inputs before and after the move, the export-jobs multiplier will greatly overstate the number of new jobs created by increased US exports of inputs to these plants.

USTR (1997) sidesteps this issue by referring to the employment effect derived from the export-jobs multiplier as 'jobs supported by exports' rather than 'jobs created by exports'. This is literally correct. However, the report does a grave disservice by failing to explain the distinction between these two concepts, and hence leaving the false impression with unwary readers (the press? the public?) that 'jobs supported' means the same thing as 'jobs created', which it assuredly does not. 'Jobs (newly) supported by exports' includes two categories: (1) jobs that already existed, but are now supported by exports (that is, they produce goods which used to be bought by domestic buyers and are now exported to those same relocated buyers), as in the hypothetical apparel-textile example above; and (2) jobs that were, in fact, newly created by an increase in exports which was not offset by a decline in domestic demand. In the example above, the unchanged number of textile jobs supplying the apparel maker are not 'jobs created by exports', but they are jobs that are newly 'supported by exports', whereas before they were supported by domestic demand. In the apparel-textile example, no new jobs were created.

The rub is that it is not at all clear that an increasing number of US jobs 'supported by exports' is of any inherent benefit to the US labour force. In fact, just the opposite is likely to be true: buyers (such as assembly plants) that relocate to low-wage countries will presumably be more tempted than before to switch to suppliers in the low-wage country, if quality requirements can be met. Apparel makers that learn how to operate in the Mexican business environment may advise their textile suppliers how to make the move to Mexico. US 'jobs supported by exports' may in such cases be the first step toward US jobs displaced.

In fact, USTR (1997) makes no claim that NAFTA has created any jobs at all, since the estimated 311,000 increase in the number of 'jobs supported by exports' is nowhere disaggregated into 'jobs newly created' and 'existing jobs newly supported by exports'. The report's only specific estimates of NAFTA's job creation in the United States is its citation of the finding by Gould (1996) that in the absence of the peso devaluation, NAFTA would have created 90,000 to 160,000 jobs. Again, the way this result is reported is misleading, obscuring the fact that it refers to a counterfactual. The report's meaning is that these jobs were not actually created, but that this was the fault of the peso crisis, not of NAFTA. As we have already pointed out, this conclusion rests on the indefensible assumption that NAFTA and the peso crisis were unrelated. In fact, apart from this, Gould's findings should not be cited at all, since this result was not statistically significant even at the 10 per cent confidence level (Gould 1996). In sum, even USTR's (1997) specific claim based on Gould's analysis is not a claim that any US job creation actually occurred in the United States during 1994-96, except perhaps temporarily during 1994.

USTR (1997) also cites the finding by DRI (1997) that NAFTA, separate from the peso crisis, boosted US net exports with Mexico either by about 3 per cent per year (based on aggregate data), or by about 12 per cent per year (based on sectoral data). It translates this into a positive impact on US output and then suggests that the net employment impact must also have been positive. Like Gould's conclusion, this finding is based on the idea that NAFTA was independent of the peso crisis, an idea we challenge earlier in this paper.

Finally, one other caveat must be mentioned with respect to the effect of exports on jobs. There is ample reason to believe that US multinationals engage in transfer pricing, at least in the maquiladora sector, as a way of reducing the profits they report in Mexico. There are evidently two motivations for firms to do this. First, Mexican law mandates that each firm share 10 per cent of its profits each year with its workers. While this law is much honoured in the breach, firms are more able to evade it if they show few or no profits. Second, until US tariffs on imports from Mexico are completely eliminated, the tariff is levied on value added in Mexico over and above the US components contained in the imports. In practice, recently about three-fifths of the value of imports from 'production-sharing' (typically maquiladora) plants in Mexico has been subject to tariff (ITC 1997:3-32). If firms report value added in Mexico as a small percentage above the cost of Mexican wages, this reduces the tariff bite, and means that the amount of profit reported is very small. For both reasons, firms are therefore likely either to overstate the value of components exported from the United States to their plants in Mexico, or to understate the value of finished goods imported back into the United States. In either case, the US trade balance with Mexico will be reported at an inaccurately positive level. In late 1994, the Mexican government announced that henceforth, arm's-length pricing would be required of firms (that is, transfer pricing would not be permitted) (Siegle 1994). However, the peso crisis followed shortly after, and we have not found evidence that enforcement of this decree has occurred. Of course, we are interested here in the change in the trade balance. However, with an expanding volume of trade, any systematic measurement error will be amplified so that the change in the trade balance will be mismeasured in a positive direction. Hence, to the extent transfer pricing occurs in an attempt to reduce profits shown in Mexico, the trade data likely overstate the US trade balance with Mexico and hence overstate NAFTA's positive impact on US employment (or understate its negative impact).

In sum, in the case of US exports to Mexico, the export-jobs multiplier appears significantly to overstate the positive employment impact of NAFTA.

9. How do changes in US imports affect employment?

Assessing the employment impact of imports is even more controversial. Some authors, particularly those critical of NAFTA, have used the export-jobs multiplier to apply to changes in imports, or changes (positive or negative) in net exports. Use of the multiplier in this way has been criticized extensively by USTR (1997), Hinojosa et al. (1996) and Weintraub (1997), among others. While their criticisms are largely technically correct, we believe the size of the error they imply is small, and that the larger errors offset one another to a significant extent, at least in the case of trade between the United States and Mexico. The main reason is that, as reported in the ITC (1997) sectoral analyses, much of the trade between the United States and Mexico is generated by 'production-sharing' operations involving assembly plants which have relocated from the United States to Mexico, and in such cases the employment impact is relatively easy to assess. In addition they neglect one source of error which would cause the multiplier to understate the employment impact.

One problem raised with applying the multiplier to imports to obtain the employment effect is that in some cases increased US imports from Mexico represent a switch from importing from a different (often low-wage) country, rather than a switch from US production. If increased imports from Mexico do not represent an increase in total US imports, but are offset by reduced imports from another country, there should be no employment impact in the United States. There is ample evidence that this is a significant factor in some sectors. In the apparel sector, growth in imports from Mexico during 1994-96 was accompanied by a decline in imports from Asian suppliers. While Mexico's apparel exports to the United States increased by 91 per cent from $2.4 billion in 1993 to $4.7 billion in 1996, Hong Kong's apparel exports to the United States declined by 1 per cent, The Republic of South Korea's declined by 39 per cent, and Taiwan's by 11 per cent. Apparel imports from China rose by only 2 per cent over the same period (ITC 1997:6-24). To the degree that increased US imports are goods diverted from other countries, no US employment impact would be expected. None of the studies available assessed the extent to which this is true in practice in the aggregate. However, it would be a mistake to conclude that US imports of apparel were entirely diverted from other countries. In fact there was also a large shift of apparel production from the United States to Mexico, like the movement by Guess cited above. As described in the statistical summary of the apparel sector earlier in the paper, ITC (1997) makes clear that substantial movement of garment production from the United States to Mexico has occurred. This is confirmed by the certification of 11,532 apparel workers by mid-1997 under the NAFTA-TAA programme as impacted specifically due to relocation of production to Mexico (DOL 1997). Additional apparel workers were certified due to the impact of imports from Mexico.

Second, it has been argued (USTR 1997: 24-25) that to the extent imports are cheaper than the same domestic goods, they save money for domestic purchasers (whether consumer or producers) who then spend that money on other goods or on expanding production. These are the gains from trade. While this point is true, its impact is offset, or possibly more than offset, by the fact that when US imports come from lower-wage countries, if their lower wage cost is reflected in a lower import price then a given dollar value of imported goods represents a larger employment impact than the same dollar value of US exports⁽²⁶⁾. While neglect of gains from trade may cause a small overstatement of the employment impact of imports, neglect of the greater employment impact per dollar of imports with low-wage countries will likely be a much larger error in the other direction.

There is an additional reason why applying the export-jobs multiplier to US imports from Mexico can be expected to understate the negative employment impact. The export-jobs multiplier is an average, taken over a range of different export sector jobs with a range of wage levels. However, according to standard trade theory, imports to the United States from a lower-wage country should be more labour-intensive than US exports. The jobs displaced in the United States (the high wage country) are therefore likely to be lower-wage than the average job in the US export sector. For example, the single largest category of US workers certified under NAFTA-TAA was apparel workers. Hence in this respect as well, the export-jobs multiplier applied to imports from Mexico will understate US job displacement.

Several other arguments which are raised in USTR (1997), ITC (1997) and Hinojosa et al. (1996) against use of the export-jobs multiplier are more or less technically correct, but their impact is likely to be small. For example, USTR (1997:24) observes that trade, including imports, creates indirect jobs in transportation, communication, finance, insurance and other sectors. It is true that indirect employment linked to imports, such as transportation jobs, is not typically counted in these studies. But there is no reason to think the net effect of this factor is large. If we are thinking of imports caused by relocation of production abroad, indirect jobs such as transportation jobs will be destroyed in one location and created in another, or some existing jobs may continue to exist. However, if the cost of such indirect services were substantial, they would add to the retail cost of imported goods in the United States, and the increased imports would not occur at all because the imports would not be significantly cheaper than domestic goods. While technically correct, then, to our knowledge this effect has not been shown by any study to be substantial, and is not likely to be so.

USTR (1997) has also argued that importing from Mexico, for example, is advantageous to the United States because it is better than importing from, say, Asia. The reason is that imports from Mexico have higher US domestic content, on average, than imports from Asia. There are two problems with this argument. First, it is a mistake to compare all US imports from Asia with all US imports from Mexico. Instead, the comparison should be confined to those firms that actually do relocate production, or are likely to relocate production - the import propensity of marginal firms, not of average firms or of a nation as a whole. Those firms are disproportionately US-based multinationals, and it is likely that the United States domestic content in their US imports - from Asia as well as from Mexico - is higher than the average of all US imports from these countries. Second, even if there is some net US employment gain in importing from Mexico rather than from Asia, this is not necessarily the relevant comparison; there is still a net employment loss in importing from Mexico rather than producing the same goods in the United States. ITC (1997) reports that for a number of sectors a central explanation for changes in trade is the relocation of production to Mexico in order to take advantage of lower Mexican wages and establish or expand production-sharing arrangements.

USTR (1997:25) argues too that the export-jobs multiplier, applied to imports, overstates the negative impact of these imports because it assumes that indirect jobs (of input suppliers) are eliminated along with direct jobs. That is, it assumes, for instance, that textile jobs are eliminated along with jobs in the apparel factories to which the textile producers sell inputs. This is technically correct. However, this factor must be weighed against the undercounting noted above for imports from lower-wage countries.

An additional issue is raised by Hinojosa et al. (1996) as well as ITC (1997): if the United States produces no close substitute for imported goods, or if the United States is unable to meet its domestic demand for a good (such as oil), then these imports do not displace US domestic output and employment. In principle this point is valid. But a very small proportion of US imports from Mexico fall into these categories, and only a tiny proportion of the increase in US imports from Mexico during 1994-96 fit this description; it is limited to a few agricultural products like coffee and avocados, as well as petroleum. The vast majority of US imports from Mexico are manufactured goods which either are currently produced in the United States, or were produced there until recently: TVs, computers, vehicles, clothing and leather goods, furniture, and so on. Thus, while technically true, this point has little relevance, at least with respect to Mexico.

10. Summary: How trade affects employment

Let us summarize our analysis of the employment impact of a change in exports, and then of a change in imports. The net employment creation of an increase in exports is overstated in two significant ways: (1) counting as exports some goods (like some electronic components) which were actually trans-shipped rather than produced in the United States, and (2) counting exports which result from relocation and are therefore offset by a reduction in domestic demand and so have no net US employment impact. The USTR category 'jobs supported by exports' sidesteps the issue, but no study has calculated net job creation accounting for these factors, because the data needed are not available. Thus the export-jobs multiplier, if incorrectly interpreted as employment gain due to exports, overstates the positive employment impact of increased exports.

To summarize the employment impact of imports, application of the export-jobs multiplier to imports understates employment impact in a major way and overstates it in a major way. The understatement comes from the fact that lower wages abroad imply larger US job losses per dollar of imports. The fact that imports are displacing employment which is apparently lower-wage than the average wage on which the export-jobs multiplier is based contributes to even greater understatement. The overstatement of negative US employment impact flows from the inclusion of increased US imports from Mexico which represent relocation of production but reduced US imports from some third country. Finally, several other minor factors discussed above do imply some small additional overstatement of the negative US employment impact of increased US imports.

Since applying the export-jobs multiplier to exports probably seriously overstates job gains, and applying it to imports probably somewhat overstates job losses, applying the multiplier to the change in net exports (as EPI (1997) has done) probably yields a fairly good ball-park figure for the net employment impact. In a rapidly growing economy such as the United States economy during 1994-96, however, much of this impact is felt as a slower rate of employment or wage growth than is normal during such a boom, rather than as actual net US job loss in most sectors. The sectoral analyses in ITC (1997) confirm that in most sectors this is the case. However, a few sectors, notably apparel, have been somewhat hard hit.

In the next section we evaluate an alternative method of estimating employment impact, which was used by Hinojosa et al. (1996).

11. The Armington method of estimating employment impact

Hinojosa et al. (1996) use a different method to calculate the job impact of imports⁽²⁷⁾. That method is to use available information on the degree to which purchasers typically switch from imported to domestic goods in response to an increase in the price of imports, or switch from domestic to imported goods in response to a decline in the price of imports. This information is summarized in 'Armington elasticities', which in this case were estimated from data on all US trade, not on trade specifically with Mexico or Canada. Hinojosa et al. (1996) applied these numbers to US manufacturing trade with Mexico and Canada to estimate how changes in US imports from these countries during 1994-96 must have affected US domestic production in these sectors. Finally, the study applied the export-jobs multiplier to each sectoral change in domestic output to estimate how US employment was affected. Its conclusion was that NAFTA had a very small net positive impact, with exports creating 98,000 jobs and imports eliminating 91,000, for a net gain of 7,000 jobs over the period 1990-1996. About 4,000 jobs were found to have been created on net from 1990 through 1993, and about 3,000 more from 1994 through 1996.

While this procedure makes sense to some degree, in some respects it can be expected to understate the impact of an increase in US imports from Mexico and Canada. It assumes that the only impact of increased imports within a sector (say clothing) is a reduction in US production within that sector (that is, of clothing); it does not consider the possibility that US consumers who begin buying clothing, for instance from Mexico, may be switching from spending that money on something other than US clothing. In this respect the impact of increased imports is to some extent understated. (However, the authors note that several aspects of their procedure would result in overstating the impact of an increase in US imports from NAFTA countries. It is difficult to judge whether the net effect is overstatement or understatement.)

Moreover, the Armington elasticities assume that the mechanism which caused the quantity of US imports to increase was a decline in the price of those imports⁽²⁸⁾.However, US imports from Mexico sometimes have risen even though no price change has occurred; this is possible because much trade is intra-firm. For example, in 1993, before the peso crisis, US automakers were exporting from Mexico between 47 per cent (Ford) and 63 per cent (GM) of their Mexican output (EFI 2/27-3/05/95, Suppl.). After the peso crisis these automakers in Mexico began exporting a much higher proportion of their output, because the huge jump in Mexican interest rates caused sales of autos in Mexico to plummet; it is not clear that changes in vehicle prices occurred, or were in any way a cause of increased auto imports into the United States from Mexico. For years, auto prices have increased about 2-4 per cent per year; only for 1998 models was it expected that price increases would be more moderate - mostly due to competition from Japanese auto prices because of depreciation of the yen (Reitman and Simison 1997). It appears that firms producing vehicles much more cheaply in Mexico, but continuing to produce the same vehicles in the United States, would have no incentive to reduce prices, because a price reduction would make their US-made vehicles unprofitable. In this case it would be inappropriate to apply the Armington elasticities. Instead, it seems reasonable to suppose that in 1995 automakers reduced the rate of growth of US domestic production, more or less one for one, to make room for the increase in imports from Mexican plants. Thus in this case the Armington elasticities would likely understate the true impact on US employment of the increase in auto imports from Mexico. As it happens, during 1995 auto production was expanding in the United States, so that in spite of increased auto imports from Mexico no net negative impact on US auto employment occurred.

In general, the Armington method seems less suited to estimation of employment effects of relocated production than to estimation of employment effects from trade alone. ITC (1997) notes for many sectors that production-sharing and hence intra-firm trade accounts for much of sectoral trade. When the decision to import instead of to produce domestically is made within a firm, the decisions will not necessarily follow the same pattern as when trade is among firms, and there is no reason in theory why Armington elasticities should provide a good predictor of impact on domestic output. At the very least, it is to be hoped that the Hinojosa et al. final report, when issued, should provide a substantial discussion of this issue.

In sum, the advantage of the Armington method is that it takes some account of the fact that an increase in imports cannot be expected to have a 100 per cent offsetting effect on domestic output in the same sector. However, it is not clear that application of the method in Hinojosa et al. (1996) is adequate to produce sensible estimates of net employment impact. We now turn to a discussion of gross employment impact.

12. Gross job displacement and the NAFTA-TAA evidence

The NAFTA Trade Adjustment Assistance programme (NAFTA-TAA) provides benefits to workers whose jobs in goods production are certified by the US Department of Labor to have been adversely impacted by imports from Mexico or Canada, or else by relocation of production to either country. Such workers also may apply for the general TAA programme which was already in existence before NAFTA and which offers a similar array of benefits; in practice many have applied for both. During 1994-96, 182,000 workers (or their unions or employers) applied for NAFTA-TAA, and all but 13,000 of these cases were decided by the end of 1996, resulting in certification of 100,000 workers. By July 1997 this number had grown to 134,492 workers (DOL 1997). Data on NAFTA-TAA certifications by two-digit SIC sector are provided in table 5, and by state in table 6.

ITC (1997) points out that the number of workers certified under NAFTA-TAA should not be taken as a count of all affected workers. Many workers may have not applied due to lack of information, or applied only for general TAA due to differences in the training requirements or other features (ITC 1997:3-8). Some workers, particularly transportation and warehouse workers - even if they can prove they were impacted by NAFTA trade or plant relocation - have been denied certification because NAFTA-TAA is restricted to workers who produce tangible goods. In addition, workers who belong to unions appear to have been more likely to submit applications, evidently due to better access to information and better support. Of all workers who applied for NAFTA-TAA 36 per cent had their applications submitted by labour unions, far higher than the 18.4 per cent of manufacturing workers in the United States who are represented by unions (CLC 1997:61). This suggests that workers who did not belong to unions were considerably less likely to file applications for NAFTA-TAA, even if their jobs were truly impacted.

Table 5. NAFTA-TAA certifications, by two-digit SIC category

Hence if workers impacted by trade with Mexico or Canada, or US investment in these countries, are representative of all US manufacturing workers, then we may estimate the true number of workers displaced at about twice the number certified. On the other hand, employers of higher-wage unionized workers are more likely to shift production to Mexico, and such firms are likely to be more affected by cheaper imports, implying that the true number of workers displaced is less than twice the number certified - perhaps in the neighbourhood of 180,000 or so during 1994-96, and perhaps 230,000 or so by August 1997.

Table 6. TAFTA-TAA certified workers by state as of mid-July 1998

This number, however, would include not just workers impacted by NAFTA, but workers impacted by trade with NAFTA countries or US investment in those countries, whatever the cause. Even if NAFTA had never been approved, the long-term trend toward shifting production from the United States to Mexico would surely have continued, as more firms learned how to do business in Mexico; thus some of the jobs lost can be attributed to this long-term trend rather than to NAFTA. Likewise, the Mexican economic reforms gave further impetus to this long-run trend. Several studies, including Hinojosa et al. (1996) and DRI (1997), suggest that NAFTA's tariff liberalization had remarkably little impact, and that most effects seen in 1994 were due to longer-term trends. The peso crisis is agreed to have had a much greater impact on US jobs. Thus, for example, the perhaps 180,000 workers hypothesized above to have been impacted from all these causes during 1994-96 might hypothetically include perhaps 90,000 who lost jobs due to NAFTA (including its effects through the peso crisis); perhaps 45,000 who lost jobs due to the non-NAFTA aspects of the peso crisis; and perhaps another 45,000 who lost jobs due to other causes such as the long-term trends. However, we are not now able to determine the true number of displaced workers who fall into each of these categories. In sum, due to the various offsetting factors recounted here, the likely number of workers actually displaced by NAFTA (the gross negative employment impact) is probably not far different from the number certified for NAFTA-TAA.

Table 7. NAFTA-TAS certifications, by reason and year

13. Assessing employment effects: An overview

Let us now try to make sense out of the wildly different numbers presented in different studies assessing NAFTA's employment impact on the United States. We will see that, because they refer to different categories or scenarios, they are not quite as far apart as they seem.

USTR (1997) makes three main claims. First, it claims that during 1993-96 the increase in US exports to Mexico and Canada has 'supported' an increased 311,000 US jobs, 189,000 supported by increased exports to Canada, and 122,000 supported by increased exports to Mexico. As explained in an earlier section, USTR does not say how many of these are newly created jobs, rather than existing jobs whose output is now exported as input to relocated plants in Mexico or Canada. In a large number of sectors, ITC (1997) emphasizes that changes in trade have been mainly driven by an increase in co-production arrangements between the United States and Mexico or the United States and Canada, in the case of Mexico mainly to take advantage of low wages. And in some sectors (such as apparel) the ITC specifically attributes the decline in US employment to the shift of production to Mexico. Thus only a fraction of the jobs newly 'supported by exports' are newly created jobs. Moreover, we cannot assume these result from NAFTA's effects; some are due to such causes as the business cycle upswing in Canada. In addition, the exports on which USTR's estimates were based apparently included rather large US grain exports to Mexico, which would have very little direct job impact because the grain sector is among the most capital-intensive in the US economy.

Hinojosa et al. (1996) estimate the impact of increased US exports to Canada and Mexico post-NAFTA as generating only 31,158 jobs, about one-tenth the USTR estimate. Some of the sources of the large difference in Hinojosa's and USTR's estimates include the fact that Hinojosa et al. restrict their analysis to manufacturing trade, and the fact that they calculate only direct job effects, not indirect effects. Given the size of the discrepancy, however, it seems likely that there is some additional explanation for the difference in estimates.

While it would be appropriate to go on to estimate the impact of the much larger increase in imports from Mexico and Canada, USTR (1997) does not do so, citing the lack of adequate general methods to estimate the employment impact of imports. USTR does, however, make other claims about overall employment effects. One is that the number of US workers displaced because of trade with Canada and Mexico (but not necessarily as a result of NAFTA per se) during 1994-96 must lie between 100,000 (the number of workers certified for NAFTA during those three years) and 32,000 (the estimated number of workers who, once certified, actually applied for NAFTA-TAA benefits during those three years). The other workers certified may have applied for general TAA benefits, or possibly taken new employment and applied for neither type of benefit.

A frequent reason for displaced workers to apply for general TAA benefits is the tight time restriction written into the NAFTA-TAA programme. The NAFTA-TAA programme (but not the general TAA programme) requires that, once certified, a worker must enroll in a training programme within six weeks of certification, or within 16 weeks of being laid off. If there is no appropriate training programme that begins during the requisite time frame, the worker has little alternative but to apply for general TAA benefits. Thus the number 32,000 provides little information about how many were affected by NAFTA.

As is argued above, it is also unjustified to assume that the number certified represents the true number of workers impacted by trade and relocation of production within North America; I have argued that the true number should be close to twice the certified number. However, the number impacted specifically by NAFTA itself is only a fraction of this total.

USTR (1997) makes a third claim, about a hypothetical (or counterfactual) situation. It cites the DRI (1997) result that if the peso crisis had not occurred ⁽²⁹⁾, NAFTA would have resulted in net creation of 90,000 to 160,000 US jobs. Unfortunately, however, the peso crisis did occur, in a particularly catastrophic form, and I have argued that the political and economic pressures generated by the NAFTA debate bear an important share of the blame. Thus I reject USTR's (and DRI's) claim. Even those who accept it, however, cannot interpret it as the net employment effect that actually occurred during 1994-96 from trade with Mexico and Canada, and even they must concede that a significant number of US jobs were displaced by the worsening of the US trade balance with NAFTA partners during that period.

At the opposite pole, EPI (1997) estimated the employment impact of the worsening US trade balance with Mexico and Canada at 420,208 jobs (including both jobs which were actually eliminated and jobs that failed to be created), of which 169,498 arose from the worsening of the United States-Canada trade balance and 250,710 from the worsening of the United States-Mexico trade balance. The study arrived at this conclusion by applying the Department of Commerce export-jobs multiplier to the change in the trade balance. Since this estimate of overall 1994-96 negative employment impact is the only one among the studies reviewed which both included jobs which failed to be created, and also included impact due to all causes of changes in trade during 1994-96, it is quite naturally larger than other numbers cited here. This number may be somewhat high, since it ignores the fact that some of the increased imports were a diversion of US imports from Asian countries. In addition, of course, it is to be expected that the number will be adjusted downward as the peso appreciates and the US trade deficit with Mexico shrinks. Indeed, one such adjustment was made by EPI in late 1997.

Since the EPI analysis is based on changes in US net exports, implicitly it accounts both for job creation and job destruction, arriving at a net change in employment. The NAFTA-TAA data suggest, as I have argued, that at least 180,000 or so jobs have been eliminated due to all causes related to trade and investment among NAFTA countries, and probably half of these are directly attributable to NAFTA.

We could then combine USTR's number of 311,000 jobs newly supported by exports to NAFTA countries with ITC's sectoral analyses indicating that a sizeable fraction of these exports probably come from existing suppliers whose customers moved offshore, and therefore imply no job creation in the United States. Perhaps 50,000 to 100,000 jobs were actually created by new exports, then - again, some but not all of them due to NAFTA, since exports were on an upward trend long before NAFTA.

Suppose further that we make the extremely rough estimate that something like 200,000 additional jobs failed to be created due to trade and investment among NAFTA countries (EPI's 420,000, less 180,000 jobs estimated actually to have been eliminated, less an arbitrary 40,000 to account for diversion of imports from other countries). Of those hypothetical 200,000 jobs which failed to be created, again not all would be due to NAFTA, but perhaps half might be.

This hypothetical calculation would lead to a net negative employment impact due to NAFTA of about 190,000 (90,000 actual jobs eliminated due to NAFTA and 100,000 which failed to be created for that reason), and a gross employment impact of 90,000. (The estimate of jobs created, based on USTR (1997), would have been implicitly offset by jobs destroyed by imports, since the 420,000 EPI figure is based on net exports and so has already offset imports against exports.)

While these numbers are highly hypothetical, they seem to be reasonable educated guesses based on the considerations outlined above. They are the basis for our conclusion that the impact of NAFTA on US employment has almost certainly been negative.

14. NAFTA and wages

The total annual income loss to US workers displaced by NAFTA due to the period during which they are unemployed (net of replacement income from unemployment benefits or TAA) was during 1994-96 probably in the range of one to two billion dollars at most. In contrast, if there turns out to be any measurable negative effect of NAFTA on US wages, even a one-time $.01 per hour loss in the US average wage would mean an income loss to labour of over $2 billion. (And a disproportionate amount of this loss would be borne by the same dislocated workers, as they take jobs on average at lower wages than they previously earned.) The dollar impact on labour of measurably slower growth in real US wages, or an actual decline in the real wage, is potentially a larger effect than displaced workers' lost income while unemployed.

Has NAFTA affected US wage growth, relative to the level which would have occurred in the absence of NAFTA? It is not easy to tell, and none of the studies reviewed here reach a definitive conclusion on this point.

NAFTA occurred against the backdrop of a long-term stagnation in real hourly compensation (wages plus benefits) in the United States. For more than two decades, US labour has been unable to secure in the form of compensation a constant share of productivity increases. For example, from 1979 to 1994, productivity in non-farm business sector increased at an average annual rate of 1.0 per cent, while real hourly compensation in the sector increased at an average of only 0.5 per cent yearly. And production and non-supervisory workers, who account for more than 80 per cent of wage and salary employment, have fared much worse. Moreover, the situation worsened in recent years: from 1989 to 1994, the rate of growth of productivity continued to be 1.0 per cent, while the rate of growth of real hourly compensation was just 0.3 per cent (Mishel, Bernstein & Schmitt 1997:134-6).

In the US manufacturing sector, from 1987 through 1996 productivity grew at an average annual rate of growth of 2.4 per cent while real hourly compensation fell, so that workers actually received a less than zero share of productivity increases. Moreover, in 1995 the lag of real hourly compensation growth in manufacturing behind productivity growth was 3.1 percentage points (with 3.2 per cent productivity growth and 0.1 per cent growth in hourly compensation), larger than in any year since 1987. In 1996 the lag was exactly the same, at 3.1 percentage points, with 3.3 per cent productivity growth in manufacturing and 0.2 per cent growth in hourly compensation (BLS 1997). This historical trend, which shows no sign of major reversal, understandably makes US labour sceptical of arguments for free trade which rest on economists' confidence that the benefits of productivity growth will sooner or later be shared with labour.

Some pre-NAFTA studies suggested that NAFTA would shift the composition of employment toward a mix that on average was higher-skilled and higher-wage than before NAFTA. Has such a shift occurred? It is difficult to give a definitive answer. On the one hand, data on NAFTA-TAA certifications suggest that lower-wage jobs have been disproportionately displaced; at least, the sector with the largest number of NAFTA-TAA certifications is the apparel sector, whose average wage (ITC 1997) in 1996 was just $7.89 per hour. In principle, elimination of lower-wage manufacturing employment could increase the average wage in manufacturing, just by changing the mix of jobs, but without raising any individual worker's wage or creating any new higher-wage employment. Indeed, the apparel sector analysis in ITC (1997) explains that this type of effect occurred during 1994-1996 within the apparel sector itself: many lower-wage jobs went to Mexico, while the higher-skill, higher-wage jobs, such as in cutting, remained in the

United States, raising the average apparel wage as 130,000 production jobs were eliminated.

USTR (1997) asserts that the average wage in the US export sector is 13-16 per cent higher than the average wage in the United States. This fact is juxtaposed with USTR's assertion that 311,000 more jobs were 'supported by exports' in 1996 than in 1993, and seems intended to imply that the average wage ought to rise. However, even if the skill mix in manufacturing shifted toward a higher average skill level (even perhaps with creation of little net new high-skill employment, and shrinkage in total manufacturing employment), we would have to ask what happened to the displaced workers. Did those unable to find jobs in manufacturing crowd into lower-wage, perhaps service sector employment, and on average suffer wage cuts? If so, this would be consistent with previous findings on the experience of displaced workers. And if so, any increase in the average manufacturing wage due to disproportionate elimination of low-wage manufacturing jobs would be accompanied by higher weighting of the non-manufacturing average wage in computing the overall average wage; thus the overall average wage would likely fall.

None of the studies reviewed here found that NAFTA had a significant positive or negative effect on US labour compensation as a whole, in most cases because they did not carry out any substantial study to assess that effect. ITC (1997) did find that of the industries identified through econometric analysis to have been affected by NAFTA, industries with earnings reductions outnumbered those in which there were earnings increases. However, it drew no conclusion about overall impact. ITC (1997) also assessed NAFTA's effects on average wages in some specific sectors, and found these wage effects to be negligible except in the apparel sector. There, the rise in the average wage in apparel was attributed to the elimination of low-wage jobs and the resulting change in composition of employment, as explained above.

Estimates of NAFTA's productivity effects are no more conclusive. While ITC (1997) cited substantial changes in productivity in various sectors, it did not attribute them to NAFTA, consistently describing NAFTA's influence as negligible.

One channel through which NAFTA might well affect workers is that employers' threats to close plants and relocate them abroad might have a chilling effect on worker organizing and the level of unionization, and thereby on the average wage level. A study by Bronfenbrenner (1996) found that NAFTA has created a climate in which employers have threatened or carried out plant closings in order to avoid unionization. Since the average wage premium for unionized workers relative to comparable non-union workers has been estimated at about 20 per cent in recent years (Mishel et al. 1997:200), discouraging unionization can be expected to have a negative impact on real wages. None of the studies reviewed here estimated this impact. The Bronfenbrenner study is described at length in the following section on labour rights.

15. NAFTA and labour rights

A study of the effects of plant closing or threat of plant closing on the right of workers to organize in the United States, conducted by Bronfenbrenner (1996), sheds light on ways in which NAFTA might have indirectly driven down US wages, or slowed their growth. The study was commissioned by the North American Commission for Labor Cooperation, established by the labour side agreement to NAFTA, and was conducted in the summer of 1996. Bronfenbrenner studied more than 500 union organizing campaigns and more than 100 negotiations for a first contract. She found that a majority of employers threatened to close all or part of the plant during the organizing drive, either in captive audience meetings, or in letters to employees, or in conversations between supervisors and one or more employees. The threats were often crude, and included posting maps of North America with an arrow from the current plant site pointing to Mexico, and attaching shipping labels to equipment in the plant with addresses in Mexico. Among those campaigns in which threats were made, in 10 per cent of the cases the threat was specifically to move to Mexico, while in other cases the threats were more vague, but sometimes made reference to NAFTA.

Bronfenbrenner found that plant closing threats were effective, on average: where such threats occurred during an organizing drive, union wins were 33 per cent, while where no such threats occurred, the union win rate was 47 per cent. Moreover, in 15 per cent of the plants where the union won the election, the employer did shut down all or part of the plant within two years of the election, in most cases before a first contract was reached. The 15 per cent shutdown rate within two years of a union winning an election was three times the rate found by researchers during the late 1980s, before NAFTA. The report concluded that NAFTA had created a climate which had emboldened employers to threaten more aggressively to close plants, or else actually to close plants, in order to avoid unionization.

16. Conclusion

Standard cost-benefit analysis calls for evaluating any policy change by determining whether the winners gain more than the losers lose, and therefore could in principle compensate them. It is then hypothetically possible, with redistribution, for the policy change to make no one worse off, and some better off. On this basis, it is not clear that NAFTA has qualified. There have clearly been winners and losers: firms that have shifted production to Mexico to take advantage of lower wages have been winners; dislocated workers and the communities in which they live have been losers. But the gains may not have exceeded the losses: the ITC (1997) was unable to detect a net gain in US GDP due to NAFTA. Other studies have claimed that NAFTA, apart from the peso devaluation, did bring about a small boost in US GDP, but if NAFTA was in any significant way a cause of the peso crisis - as we argue it was - this finding would be reversed. In the long run, of course, abstracting from short run macroeconomic fluctuations, gains from trade must be realized.

Standard trade models predict that labour, particularly unskilled labour, in the United States is likely to suffer a net loss in real income, and this would be counteracted only by sufficiently strong scale economies. Evidence of such substantial scale economies has been challenged, at least for Mexico, as pointed out above. While no significant statistical evidence of real wage losses in the United States has emerged, neither is there evidence of real wage gains. And there are a couple of reasons to suspect that a negative wage impact has occurred in the United States, even though it might be small. One reason is the detailed sectoral analyses in ITC (1997), which describe widespread shifts of production to Mexico, particularly to the maquiladora sector, to take advantage of lower wages and duty reductions. Another is the conclusion reached by Bronfenbrenner's (1996) study, that NAFTA has encouraged US employers to brandish plant relocation, and to use plant relocation, as a means of discouraging union organizing.

Moreover, the losers have not been fully compensated for their losses. A survey of studies of the cost of protectionism concludes that the cost of protecting a single job is often over $100,000 (Coughlin, Chrystal and Wood 1995). This amount may be thought of as the size of the gain to the winners after trade liberalization, which in principle is the maximum amount available to offer compensation to the losers. If trade adjustment assistance were available in even half this amount per displaced worker, quite probably labour would have far fewer complaints about NAFTA.

It was initially argued that NAFTA would attract inward foreign investment from outside North America, and this has happened to some degree. Ironically, however, this shows the effectiveness not of free trade but of protectionism. What it says is that protectionism works to some extent in attracting foreign investment if the protected market is large and attractive enough, and includes a low-wage location for production to serve that market. But it is questionable to what extent labour can achieve any gains from this arrangement, or even avoid net losses. What we know so far about the experience of US labour under NAFTA provides little encouragement on this score.

Mexican labour markets under NAFTA

1. Introduction

In order to objectively analyse the impact of NAFTA on the structure of employment and wages in Mexico, one must start by looking at the main economic indicators over a longer period than the negotiation and implementation of NAFTA (1991 to 1997). For the purpose of our analysis it is relevant to go back to 1985 when the process of trade liberalization began in Mexico.

Any analysis that seeks to examine the evolution of employment and wages must take into account not only the effects of trade liberalization, but also those of structural adjustment programmes of which they are a part. Structural adjustment programmes in Mexico have involved the following elements: privatization of state-owned enterprises; a change in the method of regulating economic activity; the promotion of a trade strategy based on non-petroleum exports; control of inflation; wage restrictions; and an effort to maintain stability in the balance of payments. All these measures have directly affected wages and employment.

During the process of trade liberalization a large number of companies have tried to reorganize their productive structures, retrain their employees, and reorient their marketing strategies. This process of 'business adaptation' has meant in many cases a reduction in the number of jobs, changes in duration and rhythm of the workday, and company downsizing and closure. These aspects will be analysed later in this paper.

During the negotiation stage of NAFTA, official statements focused on the promise of a few key benefits that supposedly would result from the treaty. The way in which these were presented to civil society was crucial, as it would have been very difficult to negotiate NAFTA had there not been a minimum consensus in Mexico.

Among the expected benefits were the following: an increase in the competitiveness of the national economy; growth in employment; and a significant increase in wage levels. The first of these was aimed at the national and transnational business sector, and the last two were meant to ease the fears of a large number of Mexicans who had suffered from a worsening of employment and wage levels in the 1980s.

In the following pages we shall demonstrate how the NAFTA promises have not been fulfilled, especially those related to sustainable economic growth. In the first section, we will analyse briefly the recent macroeconomic performance of the Mexican economy. In the second section, we will review labour market trends in the era of economic liberalization and NAFTA. In the third section we will present the main conclusions drawn from the analysis of the previous sections.⁽³¹⁾

2. Macroeconomic performance

2.1 Gross Domestic Product
Mexico's GDP grew at an average annual rate of 1.7 per cent from 1986 to 1994, lower than the rate of population growth, implying a decline in living standards. The only year in this period with a negative growth rate was 1986 when GDP growth was -3.8 per cent. Subsequently, GDP growth accelerated, peaking at 5.1 per cent in 1990.

During the Salinas administration (1988-94) the path of GDP growth took on a classic 'bell shape' (Salas y Rendón, 1996:81). Economic growth which was the main priority of the Salinas government did not develop a sustainable trend over the long term. (See tables 1 and 2.) Furthermore, the crisis of December 1994 caused a dramatic drop in GDP, reaching -6.2 per cent in 1995. (OECD, 1997, p. 175.) More recently, GDP growth became positive again, with a rate of 4.6 per cent estimated for 1996, and is projected to increase 3.7 per cent and 5.5. per cent in 1997 and 1998 respectively. (Dehesa y Camarena, 1997:89.)⁽³²⁾

Table 1. Mexico's GDP by broad development phases

With respect to GDP per capita, we have already noted that the rate of population growth exceeded that of GDP. Assigning GDP per capita for 1980 an index of 100, by 1988 it was down to 85. During the Salinas administration, the GDP per capita index rose slightly to 89 in 1994 (Hiernaux, 1996:20.)⁽³³⁾ Thus, market liberalization did not in general improve the situation of Mexican families. GDP per capita was more than 10 per cent below the average level reached in the 1978-82 oil boom period, the final stage under the import-substitution model.

In the first three years of NAFTA (1994-96), GDP per capita dropped 3.2 per cent. GDP per capita at the end of 1996 was 7.8 per cent lower than in 1981, the last year of real growth before the first crisis.

GDP per capita fell 10.0 per cent during the administration of Miguel de la Madrid, (1982-88) while growing 11.2 per cent during the administration of Salinas de Gortari (1988-94) (RMALC, 1997). Far from demonstrating the success of the neoliberal economic model, within months of Salinas' departure came the financial crisis and the collapse of GDP per capita by 6.9 per cent in 1995, the second largest decline in the economy this century, surpassed only in 1932. (RMALC, 1997:20, citing INEGI data.)

Table 2. Selected economic indicators (per cent change)

Despite the changes made to the Mexican economy during 1988-94, there were few changes in the sectoral composition of the economy. In the primary sector (agriculture, forestry and fishing) there was a steady decline in its contribution to GDP, from 6.3 per cent in 1988 to 5.8 per cent in 1995 (see tables 3 and 4). From 1988 to 1995, the average rate of growth of this sector was 1.2 per cent, among the lowest after mining and construction.

The manufacturing sector's share has been stable, at almost one-quarter of GDP - 24.3 per cent in 1988 and 24.4 per cent in 1995. The December 1994 crisis does not seem to have affected the size of manufacturing sector in relation to GDP. However, in 1990, 1991 and 1992 there was an upturn in manufacturing growth. In the first years of the Salinas administration these were very high, 6.1 per cent in 1989 and 6.7 per cent in 1990. This was due to an upturn in manufacturing and construction in 1989-90. However, during 1990 and 1991 growth started to weaken, falling to almost zero during 1993.

The crisis of 1995 was particularly hard on the manufacturing sector. Manufacturing GDP declined by -7.6 per cent, more than the overall decline of GDP overall (-6.2 per cent). Maquiladora sector employment has grown rapidly, rising from 369,489 in 1988 to 754,858 in 1996, (table 5) and jumped to an estimated 890,412 in the first half of 1997 (Zedillo, Informe, 1997). Between 1994 and 1996 maquila manufacturing value-added jumped by a record 12.2 per cent, and the job growth by an even greater 13.8 per cent. The dynamism of manufacturing GDP is due to the dynamism of the maquiladoras, leading many observers to refer to the process of 'maquilization' of Mexican manufacturing. The participation rate of the maquila sector in manufacturing GDP went from 1.6 per cent in 1975 to 5 per cent in 1988, and has continued grow  - 15.2 per cent in 1995 and 20.2 per cent in 1996 (calculated from Zedillo, 1997).

Table 3. Mexico's gross domestic product: Sectoral composition (per cent)

In 1995 oil exports were $8.4 billion rising to $11.7 billion in 1996. (preliminary estimates for 1997 show a decline to $5.6 billion). In these three years, crude oil exports were greater than gas, petroleum derivatives and petrochemicals.

Meanwhile maquila exports reached the level of $5.0 billion in 1995. The border states have become more dependent on the maquila sector than the rest of the country (table 6). In this region, the maquila sector's contribution to manufacturing GDP went from 7.2 per cent in 1970 to 20.2 per cent in 1988. In the first quarter of 1996, manufacturing as a whole grew by 7.3 per cent while the maquila sector grew by 15.8 per cent; though leading the recovery, its low contribution to GDP continues to be problematic.

Services GDP increased at an average annual rate of 2.5 per cent between 1988 and 1995. However, this average obscures some internal disparities. The most buoyant were financial services with an average growth of 13.9 per cent per year during this period, banking with 5.1 per cent, and transportation, communications and storage with 3.5 per cent. These figures reflect the unprecedented expansion of the financial and banking sectors.

At the same time, the expansion of the transportation sector is related to the growth in the external sector and movement of goods within the country and abroad. This has been noticeable in ports, airports, and customs offices. Moreover, it has been necessary to expand the highway system and modernize ports and airports. This has been accompanied by significant privatization of communications and transportation infrastructure. For example, sea freight grew by 7.4 per cent per year from 1988 to 1995. The number of airline passengers increased by 54.6 per cent in the same period. Freight by road increased by 22.7 per cent annually (data taken from various tables in the Zedillo, Informe, 1996.) The share of services in total GDP has, like manufacturing, been relatively stable (table 4).

Finally, one of the major changes in composition of GDP has been the decline in the public sector, from 20.4 per cent in 1988 to 14.7 per cent in 1995 reflecting the privatization of the Mexican economy.

The growth of GDP has not been stable in recent years. There was a sharp decline even before the crisis of 1994, which considerably aggravated the situation. Moreover, this growth has been has been concentrated in the external sector to the detriment of the domestic market.

During the NAFTA era (1994-97), the largest annual growth has been in communications and transportation (12.4 per cent), due in part to domestic and foreign private investment in telecommunications. The next largest are electricity (11.8 per cent), manufacturing, especially the export component (9.8 per cent), and mining (including oil) (8.1 per cent).

Distribution of the GDP by sectors has been stable during the period, but there have been differences within sectors. Notably, the growth of the financial and banking sub-sectors (6.6 per cent) has been driven by foreign portfolio investment and liberalization. By contrast, sectors linked to the internal market decreased or showed little dynamism. Construction dropped by 7.6 per cent. Despite international tourism, the internal trade category (restaurants and hotels) fell by 6.1 per cent. The agricultural sector grew only 3.1 per cent. The case of the manufacturing sector is noteworthy: two branches alone - machinery and equipment, and food processing - make up 52 per cent of manufacturing production (RMALC 1997:20).

Table 4. Growth of Mexico's gross domestic product by sector%

Table 5. Employment in manufacturing

Table 6. Manufacturing domestic product and value added of the "Maquiladora" sector for particular states, 1975-80-85-88