Abstract: Mexico's experience under NAFTA is analyzed and used to suggest likely outcomes of a western hemispheric integration agreement. Overall gains in Mexico from NAFTA appear small. The maquila sector, but not non-maquila manufacturing, experienced rapid growth through late 2000. That growth was driven largely by the U.S. boom and the 1995 peso devaluation, though unique preferential access to the U.S. market granted by NAFTA did play a role. Likely explanations for the maquila sector's decline beginning in late 2000 include the peso's real appreciation, the U.S. recession of 2001, and the sharing of preferential access to the U.S. apparel market with Caribbean Basin countries under the Caribbean Basin Trade Partnership Act. Factors depressing unskilled wages include increasing competition from China, displacement of farmers through increasing imports of U.S. corn due to NAFTA's removal of Mexican corn subsidies, and the impact of the 1995 crisis, devaluation and inflation, during which unskilled workers were least able to defend their real wages, due probably to increases in labor supply, the low minimum wage, and lack of labor rights and unionization in the maquila sector. Other possible inequalizing effects between and within countries, such as intellectual property rights enforcement, are discussed and institutions are proposed to support greater equity.

What can we learn from the U.S.-Mexico experience about how to avoid worsening inequality in the process of hemispheric integration? Critics of globalization mostly point to a problem of increasing inequality between labor and capital, arguing that regional integration agreements harm wages and benefit profits. In contrast, many recent economic studies define inequality more narrowly as a wage or income gap between production and non-production workers, or else between more-educated and less-educated workers (called the "skill premium"). Such studies typically fall short of addressing the central concerns of critics in the anti-globalization movement. To be sure, critics are concerned that the poor are likely to fare worse than other income groups under the impact of economic integration, but their focus is on the power of multinational firms, both vis-à-vis labor and vis-à-vis the public interest. By providing multinational firms with freedom to move among regions with little or no government regulation, they argue, economic integration agreements allow these firms to extract concessions from local or national jurisdictions, as well as from groups of workers competing against one another in different regions. They contend that this happens because of both trade and non-trade components of these agreements.

Among non-trade components, critics focus on national treatment of foreign investors, extension of intellectual property rights protection to developing countries, broad anti-expropriation provisions such as NAFTA's Chapter 11, and objectionable dispute settlement procedures. For example, they hold that in practice the dispute settlement procedures operate to privilege trade liberalization over social goals such as environmental protection, even when such goals are explicitly written into the agreements. They hold that the procedures, and the rules they seek to enforce, usurp or restrict the powers of national and regional jurisdictions. In the course of this paper I try to address some, but not all, of these broader concerns.

1. Mexico's liberalization

In order to understand NAFTA's effect on Mexico, we need to understand what liberalization had already occurred before NAFTA was implemented in 1994. Mexico has undergone three liberalization phases. The first was selective liberalization in the form of the Border Industrialization Program of 1965, better known as the maquiladora program. At a time when Mexico had moderately strong protectionist trade policies and restricted foreign firms to minority ownership of newly established firms (Whiting 1992), the maquiladora program allowed foreign firms located in Mexico and dedicated to producing solely for export to be exempt from tariffs on imported inputs as well as receive other concessions, and to maintain 100% foreign-owned production operations. This made it possible for U.S.-based firms to take greater advantage of the production-sharing program known as 806-807 and created in 1962, allowing firms to send components out of the U.S., have them assembled in another country, and then re-import them, paying tariff only on the value added abroad. The Generalized System of Preferences (GSP) in 1974 granted even greater advantages to firms in certain developing countries (including Mexico) producing certain goods, since these goods were made completely duty-free.

The second liberalization was a major reduction in tariffs and elimination of import licensing as Mexico joined the GATT in 1986, as well as extensive investment liberalization decreed in 1989. An important change in U.S. trade policy in the late 1980s also affected Mexico's production and trade. Until 1986 Mexico and the Caribbean Basin (CB) countries were still - like other countries - restricted by quotas on shipments of apparel and textiles to the U.S. under the Multi-Fiber Arrangement. The U.S. relaxed apparel quotas for certain CB countries beginning in 1986-87 under its new Special Access Program, and then gave Mexico similar privileges in 1988 under the Special Regime, for garments using fabric formed and cut in the U.S. (USITC 1994, 1990). This led to rapid growth of apparel production in the region.

The third liberalization phase was NAFTA in 1994, whose effect on trade was generally to eliminate tariffs, subsidies and other barriers, for some goods at once, for others in a series of steps over a period up to 15 years. NAFTA also changed the U.S. production-sharing law in a way that particularly benefited U.S. textile and apparel firms operating in Mexico and their subcontractors, an issue that will be discussed in a later section. Because considerable liberalization had already occurred before 1994, we might not expect the additional impact of NAFTA to be large. About 45% of the value of U.S. imports from Mexico was already duty-free by 1990, either under HTS 9802 because they were U.S. components of goods assembled in Mexico and re-exported to the U.S., or because they were goods exempt from duty under the GSP, or because they were duty-free under Most-Favored-Nation (MFN) tariff rates (USITC 1993).

2. Growth after NAFTA

Did the Mexican economy grow more rapidly after NAFTA? It is always difficult to compare growth rates meaningfully over short periods of time, since cyclical fluctuations can be large, and the comparison depends heavily on the beginning and ending years (or even quarters) chosen. It is best for beginning and ending points to be in the same phase of the business cycle: both trough years or both peak years. Economists tend to cite a growth rate from a trough to a peak year when they want to claim that growth was exceptionally rapid. For example, strong advocates of economic integration tend to cite the change in Mexico's real GDP after NAFTA in growth rates from 1996 to 2000, which essentially compares the level of real GDP in the trough year of 1995 to its level in the peak year of 2000. Such a comparison is misleading.

In reality, there was little change in the growth rate of Mexico's real GDP after 1994, compared to 1986-93, the second liberalization phase. Figure 1 shows the exponential growth path that best fits the data for the period from the last quarter of 1985 to the last quarter of 1993, that is, the second pre-NAFTA liberalization period. (If we begin instead with the first quarter of 1985, the trend line is lower and makes 1999-2000 look like it significantly improved performance over the 1985-93 trend growth path; if we begin with the first quarter of 1986, the trend line is higher and makes NAFTA look like it caused the economy to significantly fall short of the trend growth path. The point here is not that NAFTA definitively failed to raise Mexico's trend growth rate, but that any change was small and the outcome is too close to call.) Notice that in the post-NAFTA period, while real GDP fluctuated below and above the trend line, 2001was a year of slowdown, returning very close to the trend growth path. Arguably, the 1999-2000 growth rate was driven at least in part by the U.S. high-tech boom, now revealed to have been partly driven by inflated earnings and questionable practices by investment analysts.

As I argue more fully in a later section, part of any acceleration in growth that may have occurred was likely temporary because it was based on Mexico's temporarily unique preferential access to the U.S. apparel market. As of the implementation of the Caribbean Basin Trade Partnership Act in October 2000, Mexico now again shares that access with the Caribbean Basin countries - just as before NAFTA. This suggests that we cannot expect exceptional growth of Mexican real output in coming years.

3. Employment, output and wages in manufacturing

The dynamic manufacturing sector since NAFTA has been the maquila sector, while non-maquila manufacturing has stagnated. In fact, while maquila manufacturing value added grew from 1.3% of GDP in 1988 to 3.1% in 2000, the manufacturing sector as a whole fell from 21.9% of GDP in 1988 to 18.7% in 2000 (Primer Informe de Gobierno 2001:153, 163). However, as Figure 2 shows, with the peso crisis in 1995-96 and the sharp decline in the dollar wage, the manufacturing share of GDP did jump up by more than two percentage points as maquila employment surged; by 1997, though, it had leveled off and it later continued its downward drift. The lower line in Figure 2 shows that non-maquila manufacturing value added also increased after the peso crisis, but still declined as a percentage of GDP even over 1993-2000.

The fact that the uptick occurred in 1995, not 1994, suggests that the peso crisis was at least as much a cause of maquila growth as was NAFTA. For the maquila sector it is useful to look at employment rather than value added, since reported value added in the maquila sector may be significantly distorted by transfer pricing (Tang 1997). Figure 3 shows growth since 1980 in total maquila employment. Two things stand out: (1) the acceleration in maquila employment growth appears to have become strongest in 1995, not 1994; and (2) a sudden and precipitous decline in maquila employment occurred beginning in October 2000.

Before offering an explanation of these facts, we look at trends in manufacturing wages in the maquila sector, since the wage trend may be important either as cause or as effect of a change in employment. In both the 1980s debt crisis and the 1995 peso crisis, the combination of contraction and inflation caused the real maquila wage to fall sharply. Figures 4 and 5 show the evolution of real wages for production workers (obreros), technicians (técnicos), and non-technical white-collar workers (empleados) in the maquila sector. In Figure 4, 1982 is taken as the base year to display the divergence of wages following the crisis, devaluation and inflation. In Figure 5, 1993 is taken as the base year to show divergence following upon NAFTA and the 1995 peso crisis. White-collar workers, among whom are professional and managerial personnel, clearly fared much better than either of the other two groups in both periods.

One striking fact is that the wage divergence in 1983 and the following years shown in Figure 4 is even larger than the wage divergence in the 1995 crisis. This certainly suggests that the divergence was a product, at least in part, of the crises and inflationary episodes themselves, and less of trade liberalization - which had not yet occurred to any significant degree in the early 1980s. I will argue in a later section that the decline in the real value of the minimum wage contributed to the decline in the incomes of lower-wage households. However, the severity of the 1995 crisis itself, and by implication the severity of its effect on wages and wage inequality, may be partly attributed to the political and economic forces to which the NAFTA project gave rise. Larudee (1998) argues that capital flows were larger because of the political process surrounding NAFTA, and Mexico's measures to postpone the crisis less appropriate, because of political exigencies related to NAFTA. As a consequence, the crisis was more severe than it might otherwise have been.

As Figure 3 shows, maquila employment grew dramatically after NAFTA, from 547,000 in December 1993 to a peak of 1,339,000 in October 2000 (INEGI, BIE 2002). Much of this gain, however, will be shown below to be a result not of NAFTA but of the traditional sensitivity of maquila employment both to fluctuations in U.S. GDP and changes in the dollar value of the maquila wage (or to changes in the real exchange rate). Figure 6 shows the trend in the average Mexican non-maquila manufacturing wage in dollars as a fraction of the average U.S. manufacturing wage; the maquila dollar wage followed a similar trend. Naturally, it shows a sudden and very large drop in the ratio of Mexican to U.S. wages in 1995 with the devaluation.

The influence both of this factor and of U.S. GDP can be seen in Figure 7, which shows a 5-month moving average of the per cent change in maquila employment using the same data as in Figure 2. Employment declined during U.S. recessions in 1982, 1990-91 and 2001, and grew during the intervening booms. Major devaluations were followed by acceleration of maquila employment growth in both 1983-85 and 1995-96.

Figures 3 and 7 also show the sudden reversal and precipitous decline in maquila employment beginning in October 2000. In a 15-month period ending in December 2001, employment dropped to 1,082,000 - a loss of 257,000 jobs from the peak, or 33% of the post-NAFTA job gain. Of these employment losses, 97,000 were in electronics (43% of that sector's post-NAFTA job gain) and 59,000 were in apparel (26% of its post-NAFTA job gain). Apparel and electronics together accounted for about three-fifths of both the maquila sector's post-NAFTA gain and its subsequent loss in employment. The U.S. recession starting in early 2001, and the related slump in the high-tech industry, were among the causes of this decline. An additional factor was the effect of the real appreciation of the peso from late 1998 onward, whose effect we saw in Figure 6.

Statistical regression shows a relationship between the annual percent change in maquila employment ¹ and two independent variables, the annual growth rate of real U.S. GDP and the real peso/dollar exchange rate, with both variables significant at better than the 5% level. In addition, dummy variables were used to test for a possible shift at the beginning of 1988 (due to the Special Regime relaxing Mexico's apparel quotas into the U.S. market) or with NAFTA in 1994, but neither was significant even at the 20% level, and both had the wrong sign.

A regression was also run using quarterly data on conceptually similar variables from 1993:I through 2001:IV. The quarter-to-quarter growth rate of maquila employment was significantly associated (at the 5% level) with the (annualized) growth rate of U.S. GDP lagged one quarter, and at the 6% level with the ratio of the Mexican non-maquila manufacturing wage to the U.S. manufacturing wage, with a break in October 2000, but no break in January 1994 (not significant even at the 20% level).²

The cause of the October 2000 break appears to have been the implementation in that month (after passage in May) of the Caribbean Basin Trade Partnership Act (CBTPA), which extended to the Caribbean Basin countries preferential access to the U.S. apparel market approximately on a par with what was granted to Mexico by NAFTA (Rodríguez-Archila 2000). Thus NAFTA's effect on the Mexican economy appears to have been both small and fragile - fragile because it depended not just on Mexico's preferential access to the U.S. market, but on the uniqueness of NAFTA's preferential access - which was removed with the CBTPA.

4. Mexico's preferential access to the U.S. market

NAFTA granted to Mexico in 1994, among all developing countries, a degree of preferential access to the U.S. market which none had ever had. There followed rapid growth in the maquila sector, especially in apparel and electronics. However, the uniqueness of that access was temporary, and it has already begun to be reversed. The limited success that Mexico had from 1994 to 2000 was to some degree a result of that uniqueness - not of the access itself so much as the fact that Mexico was the only country that had it - and so we would not expect to see a repeat performance from an FTAA. In fact, Mexico's benefits from this source may decline as it shares preferential access with other western hemisphere countries.

As we saw in Section 1, in the late 1980s both the CB countries and Mexico were granted relaxation of apparel quotas, in contrast to other developing countries which still faced rigid quotas. NAFTA changed three important facts about the apparel sector. First, it created a new category, HTS 9802.00.90, under which apparel and other textile goods assembled in Mexico from fabric wholly formed and cut in the U.S. would enter the U.S. duty-free as well as quota-free. Among all U.S. production-sharing imports in the apparel and finished textile products sector during 1994-96, a little over 70% entered under this new category, to which no other country had access (USITC 1997: 6-25).

Second, under this new category, apparel and finished textile products could undergo finishing processes such as wrinkle-free processing and stone-washing without disqualifying the product for 9802 treatment - even though a similar finishing process applied to similar goods assembled in any other country still would disqualify the goods from the 9802 tariff exemption on their U.S. components. Likewise, certain categories of garments could receive tariff preferences even if not made from fabric formed and cut in the U.S.; for example, silk qualified even if the fabric was produced outside North America and cut and sewn in any NAFTA country (U.S. Customs Guide...2002). Telephone interviews with several U.S. apparel firms by staff of the U.S. International Trade Commission confirmed that these advantages were paramount in their decision to locate garment production in Mexico (USITC 1997: 6-25).

Third, NAFTA also reduced tariffs into the U.S. on garments that did not qualify for 9802 treatment, including garments produced in the non-maquila sector. These changes did impart some acceleration to maquila apparel employment growth in Mexico in 1994, as shown by Figure 8, but the peso was strongly overvalued in that year, and the acceleration was only moderate until after the peso crisis.

The CBTPA more or less restored parity with Mexico for the CB countries with respect to access to the U.S. apparel market. As we have seen, at almost exactly this moment the long-term rising trend of maquila employment turned downward, although there was no break in the U.S. economic upswing then, and no sharp real appreciation of the peso. The decline in production in Mexico also extended to the textile, apparel and leather goods sector of non-maquila manufacturing, as shown in Figure 10.

5. Low-wage competition from Asia

Overshadowing western hemisphere integration is the immense presence of China as a site for low-wage production, especially of consumer goods which Mexico now produces, like clothing, footwear and electronic goods. This puts fierce competitive pressure not just on the employment and wages of U.S. workers but also of workers in middle-income countries like Mexico (Hanson and Harrison 1999; Wood 1997; Freeman 1995; López-Córdova 2001). In fact, Mexico had dreaded and resisted China's entry into the WTO because it would mean that Mexico would have to reduce to a maximum of 35% the 400% tariffs it imposed in 1994 on imports of certain Chinese products, mostly apparel and textiles ("WTO Door..." 2001). Only after negotiating a recent commercial agreement with China, allowing Mexico to extend the exceptionally high tariffs for six years, did Mexico agree to let China become a WTO member.

Since 1995 Mexico has also maintained its external tariff on textiles, apparel and leather goods from non-North American WTO members at 35%, indicating that it is not just protecting against China but against other producers as well. Until China's strong export performance eventually appreciates its currency and narrows the labor cost gap with middle-income countries (or Mexico further devaluates the peso), Mexico has little hope of being highly competitive on the world apparel market under a free trade regime.

6. A vulnerable agricultural sector

Mexico has an agricultural sector which is especially vulnerable to a free trade agreement. About half of Mexico's arable land was planted to corn before NAFTA , but productivity in corn cultivation was and is quite low compared to the United States. Average output of corn per hectare in Mexico is 2.3 metric tons, while in the U.S. it is 8. While large U.S. subsidies and provision of services to agriculture explain part of the productivity difference, part of it comes from Mexico's lower rainfall, its more uneven terrain, and its latitude closer to the equator, where there are fewer hours of sunshine during the height of the growing season. With more irrigation and other government assistance, Mexican corn yields could probably be increased, but U.S. corn would still likely be produced at lower cost.

The result of Mexico's removing subsidies to its corn growers was expected to be displacement of corn farmers and their migration to urban areas or the U.S. seeking work. Levy and van Wijnbergen (1994) and Hinojosa and Robinson (1991) both used computable general equilibrium models to estimate that about 800,000 labor force members would migrate with their families as a result. This would tend to depress wages of low-skilled workers in Mexico, both rural and urban, and if much migration to the U.S. occurred, it could also exert downward pressure on low-skilled wages in the U.S. Real agricultural wages in Mexico decreased at an average annual rate of 4.3% between 1989 and 2000 (Zahniser and Treviño 2001). The Mexican government has implemented programs to try to cushion the impact on the rural sector, but there are reports that a considerable number of corn farmers are giving up farming.

How does agriculture fit into a standard trade model? Economists sometimes use the factor proportions or Heckscher-Ohlin model to analyze trade liberalization between the U.S. and Mexico. If a version of the model with just capital and labor is used, and if its assumptions are satisfied, and capital-labor ratios in the two countries are not too different ³, then wages in Mexico ought to rise significantly and in the U.S. ought to fall (even if very slightly) as a result of liberalization. However, in the capital-labor version, one assumption is violated: corn production is labor-intensive in Mexico relative to most other sectors but capital-intensive in the U.S. relative to other sectors, a fact which if both countries share the same technology constitutes a factor intensity reversal (Larudee 1995). Hence wage equalization will not necessarily occur.

Alternatively, it could be argued that the assumption that both countries have access to the same technology is violated, since Mexican producers have a disadvantage due to latitude, and in many regions also due to rainfall (or water supply) and terrain. But the key issue is the same however we look at it: since Mexican corn growers cannot produce corn as cheaply as the U.S., despite their much lower wages, the demand for the unskilled labor used in producing corn in Mexico will decline when U.S. corn is allowed to be imported in increasing quantities. The resulting displacement of farmers will increase the supply of labor, especially unskilled labor, to other sectors, tending to depress average wages, especially of unskilled workers.

7. Volatile capital flows, crisis, recession, and wage inequality

In the last two decades the world has seen repeated episodes of major balance of payments crises leading to devaluation, recession and inflation. Such episodes almost invariably result in a decline in the real wage (Morley 1995: 32, Table 2-2). While some of these episodes happened to countries which had not liberalized trade or capital flows, such as Mexico in 1982, liberalization of capital flows does increase the likelihood of such crises, and therefore of their consequences.

Why do these events exacerbate wage inequality? One reason is that poorer households are less able to defend themselves against inflation by, for example, putting their wealth into dollar-denominated assets. Another is that in a recession workers' bargaining power declines, and it becomes more difficult than in a period of economic growth for them to achieve wage increases at the same pace as inflation. There are other likely causes as well, which we discuss in some detail.

One source of increased wage inequality is an increase in the supply of labor during a crisis, especially the unskilled labor of women and children who were not previously in the labor force. It is well known that under some circumstances, a household in which one earner becomes unemployed or underemployed, or suffers a decline in wage or self-employment income, decides to send another household member into the labor force. Figure 11 shows that in the 1995 crisis, the percentage of all urban workers earning less than the minimum wage rose markedly.

Figure 12 also shows an increase in the share of urban workers working in microenterprises of 1 to 5 employees during 1995-96, suggesting either that new entrants entered such employment, or else that those laid off from larger firms found or made work in microenterprises. A decline in one earner's income is more likely to cause entry of additional household members into the labor force if the household has little savings or access to credit; known as the "added worker" or "secondary worker" effect, it has been noted in various developing countries by Fields and Newton-Kraus (1996). Standing (1981: 76-78, 82-87) reports studies in seven countries which found that married women's labor supply was negatively affected by some measure of the rest of the family's income, or of the husband's wage or income. And while some studies in the U.S. have failed to find the added worker effect, Cullen and Gruber (2000) show that this is probably because unemployment insurance crowds out the effect. So in countries where no unemployment insurance system exists, we should expect to find that the effect is relatively strong.

Figure 13 shows women's labor force participation rate in Mexico, and there is a jump during the 1995 crisis, beyond a general increasing trend over time. However, it is clear that this factor is only one among several factors at work, apparently including both the "push" factors just described and some pull factors such as the expansion of the demand for apparel labor.

We would expect such entry to disproportionately depress wages in the low-skilled segment of the labor market, since new entrants are likely on average to have few skills and little labor force experience. Once wages in this segment of the labor market become depressed, these new workers will likely want to work more hours than at the old wage in order to try to maintain their previous income level. Moreover, entry into the labor force in a recession would in theory not likely be followed by exit from the labor force at the same pace in a recovery, except if the recovery is strong and/or prolonged. That is, there is a coordination failure: those working in low-wage employment would each gain an increase in wages if all reduced their labor supply - but none acting in their individual rational self-interest will withdraw any labor. Hence a crisis that drives households to significantly increase their labor supply will have a downward ratcheting effect on real wages of low-income households, which can be cumulative if a second crisis occurs before real wages have recovered from the first.

Some economists hold that a high enough minimum wage might help to prevent the worsening of wage inequality in recessions, while others hold that the minimum wage will not help, and might even make matters worse. Lustig and McLeod (1996), in an econometric study of 23 countries and over 40 time periods during 1963-90, found an inverse relationship between the real minimum wage and the poverty level, as measured either by the poverty headcount or the poverty gap, and with various definitions of the poverty line. That is, raising the minimum wage would reduce poverty, all else equal and lowering the minimum wage would increase poverty. Nevertheless, the study's results are consistent with the argument I present here. The question of the effect of the minimum wage on the distribution of wages in the economy under various macroeconomic conditions is not yet resolved, and a number of competing models have been offered. Some of these are discussed in Harrison and Leamer (1999).

Bell (1997) assessed whether the minimum wage had any effect in Mexico during 1984-90, a period during which its real value fell. She asserted that in Mexico in this period the minimum wage had essentially no effect on wages or employment in the formal sector, based on data on establishments with 10 or more employees. However, she reported that household survey data showed a different picture: that many employees, especially in the informal sector and especially women, had earnings at or below the minimum wage, and that it looked as though the minimum wage did have an effect at least for these workers. Moreover, the data she reports are consistent with the possibility that formal sector firms with fewer than 10 workers were more likely to pay wages at or near the minimum wage. All these observations are consistent with the hypothesis that economic crisis causes wage inequality, in part through increased household labor supply, especially of women, and that the effect is stronger where the minimum wage and other aspects of labor laws and the social safety net are weaker.

Some economists have objected that raising the minimum wage will only benefit formal sector workers - those who work for employers that follow labor laws and reporting requirements. The huge number of informal sector workers in developing countries like Mexico, they argue, will not benefit and may actually be harmed. But this conclusion implies a model of the interaction between the formal and informal sector labor markets which is controversial. Lustig and McLeod (1996) briefly survey models that could underlie their finding that the minimum wage is inversely related to poverty. Let me offer a model along the lines of one type of model they describe, based on sources of demand for the output of the informal sector.

It is generally argued that the informal sector produces goods of lower quality at a lower price than the formal manufacturing and commercial sectors, including such things as crude furniture, kitchenware and other housewares. Let us make the plausible assumption that formal sector workers earning close to the minimum wage tend to buy exactly these sorts of goods, and so spend a higher fraction of their incomes on informal sector output than do higher-wage formal sector workers. We further assume that the labor demand curve for such low-wage workers is inelastic, that is, raising the minimum wage will increase total wage income of this group, even though some are laid off. Bell (1997) found for Colombia that a 27% increase in the real minimum wage reduced unskilled employment by only 2-12%, so that in this case real wage income for the group did rise; similar results have been found in other cases of an increase in the minimum wage. Based on our assumption, this would augment demand for output of the informal sector. Depending on the wage elasticity of the demand for formal sector labor, and the income elasticity of the low-wage group's demand for informal sector output, informal sector incomes could either rise or fall; but under quite plausible shapes of these curves, consistent with available empirical evidence, raising the minimum wage could easily increase informal sector incomes as well as formal sector wage incomes. Among the results would be a reduction in poverty, as Lustig and McLeod (1996) found.

Thus one component of a causal explanation of the increase in wage inequality in Mexico appears to be the liberalization of capital flows, coupled with the low real minimum wage and the lack of labor rights and unionization in the dynamic maquila sector. Competition from China, and migration of displaced farmers due to low agricultural productivity and increasing corn imports are also factors. Causes cited in other studies, such as differential removal of tariffs and skill-biased technical change, may well also be operative. But any explanation of wage inequality must carry a plausible account of the time-path of the increase in inequality, and it is here that the roles of crisis and the social safety net come into play.

8. Intellectual Property Rights

One goal that the U.S. has for an FTAA is that it should contain provisions strengthening intellectual property rights (IPRs) and increasing the penalties for violating them. Yet enforcing intellectual property rights more widely is the opposite of free trade: it extends the monopoly power of holders of patents, copyrights and trademarks. It therefore should only be done if sound research establishes that the welfare of society as a whole will be improved by extending these rights in small or poor developing countries. It should give us pause that support for extending IPRs originated with those who profit thereby, and pre-dated almost all of the relevant research. The theoretical argument for IPR protection is that it encourages innovation by allowing innovators to recoup their investment in research and development. For small or poor developing countries, however, Deardorff (1992) argued that the increase in innovation would be small relative to the cost imposed on consumers by monopoly prices, so that extension of IPRs to all countries would reduce the welfare of the smaller, poorer ones. Deardorff's argument is generally acknowledged to be correct, and it is therefore widely agreed that enforcing IPRs in small or poor developing countries has the immediate effect of transferring income from their populations to IPR holders in developed countries, mainly the U.S. (Maskus 2000). A number of researchers hypothesize, however, that in the medium to long run these countries may reap substantial benefits, which might be large enough to more than offset the short-run costs. But while recent research has produced useful evidence and hypotheses, it still unproven that developing countries are likely to experience a positive net benefit over the long run from enforcing IPRs (Maskus 2000). Moreover, one very important argument against this proposition seems to have been entirely overlooked.

Books and computer software are both crucial inputs into education and training. A country in which books are inexpensive due to lack of copyright enforcement is a country in which it is cheaper for people to educate themselves and their children. A nation in which software is widely available at low or zero cost, because IPRs are not enforced, is a nation that can develop computer skills relatively cheaply. Thus one major cost of enforcing IPRs is likely to be slower growth of human capital in the labor force; and human capital has been shown to be an important determinant of economic growth.

This suggests that stronger IPRs will not only exacerbate inequality between developed and developing countries, but is likely also to worsen inequality within developing countries, since the cost of books and software - the inputs into learning - will be a higher percentage of the income of poorer households, and so for them constitute a greater barrier to acquisition of human capital. A similar argument applies to the case of medicines. In some countries, medicines are available at much lower cost than in the U.S., either because compulsory licensing laws force owners of pharmaceutical patents to allow the medicines to be produced at lower cost by local or state-owned firms, or else because IPRs are not enforced, so that local firms can produce and sell the medicines without paying the patent holder. Strong IPR protection and weak or nonexistent compulsory licensing laws make medicines no longer cheap, so that where subsidized health care is not widely available, the poor may suffer most. For it is the poor who have least access to credit, and so are most likely to suffer catastrophic harm - to die or become too ill to work - because they cannot afford the medicines they need.

9. Anti-expropriation provisions like NAFTA's Chapter 11

NAFTA's Chapter 11 awarded new rights to multinational firms to sue governments and claim that laws, regulations or government policies have in effect "expropriated" the firm's profits. No economic argument seems to have been offered to claim that this provision provides net benefits to the public, and it is hard to think of any economic principle that could succeed in justifying it. For we do know that it is economically justifiable for the government to adopt policies to correct a market failure.

For example, suppose fuel is sold with an additive which is know to be toxic. Without government intervention, sale of the fuel with the additive may take place at a price that does not reflect the cost of the pollution and adverse health effects generated by use of the fuel - that is, a negative externality is present. If the government intervenes with an effective remedy to this market failure, such as by prohibiting the sale of the toxic additive, then allowing foreign firms to sue (or to threaten to sue) to recover the cost imposed on them by the remedy in effect cancels the remedy, or intimidates the government from correcting the market failure.

10. What have we learned about the effect of an FTAA?

One lesson that emerges from Mexico's experience under NAFTA is that, where significant liberalization has already taken place, the benefits of further liberalization should not be expected to be large. Moreover, such benefits may come from tariff-jumping investment to take advantage of preferential market access provided only to members of a regional trade agreement - from firms that want to produce inside a free trade area (FTA) in order to avoid paying a tariff for exporting into it. So these benefits may be temporary only, and disappear when non-members of the FTA are later granted the same access to the FTA market.

A second lesson is that countries with low agricultural productivity are not so likely to experience accelerated growth in wage income, especially in wages of unskilled workers, because a steady flow of rural-urban migration is likely to hold urban wages down. The solution is either to exempt agriculture from generalized trade liberalization and permit national governments to subsidize agriculture where it seems appropriate, or else for the wealthier countries party to an FTAA to provide considerable development funding to the poorer countries, with special programs for agriculture.

A third lesson is that middle-income countries like Mexico may well be squeezed from below by lower-wage countries like China. Under free trade, the competition from China would either drive Mexico's apparel industry under, or else force Mexico to let the value of the peso fall in order to maintain the competitiveness of its most labor-intensive manufacturing sectors. The only reason this has not already happened is the high tariff Mexico has maintained, and will continue to maintain for some years, on certain Chinese goods. Eventually, as China's industry grows, its exchange rate should appreciate and its wage levels become less threatening to countries like Mexico. In the meantime, Mexico will probably need to rely more heavily on industries like auto, where its proximity to the U.S. is a particularly large advantage.

Fourth, when economic crises strike, as they have in many of the developing countries of the world, a weak social safety net is likely to give rise to an increase in labor supply, especially of workers with little or no labor force experience, so that unskilled wages disproportionately are driven down. There is evidence that raising the minimum wage could forestall this effect. While further research is needed both on the effect of the minimum wage on informal sector incomes and on the specific mechanisms through which it acts, available evidence suggests that raising the minimum wage may well raise informal sector incomes, perhaps by increasing demand for informal sector output. Moreover, taming volatile capital flows by imposing a Tobin tax on currency transactions (Tobin 1994) would help to reduce the frequency of financial crises.

Fifth, at least in the short run, increasing enforcement of intellectual property rights will reduce income in developing countries and shift that income to mostly U.S.-based patent, copyright or trademark owners (Maskus 2000). In the longer run, there is speculation but little evidence that the benefits of greater innovation will offset the short-run costs; and in the longer run additional costs will be suffered, such as slower growth of skills for which copyrighted materials (books, software) are inputs. The best evidence now available suggests this will contribute to increasing inequality between countries. In addition, it seems likely to contribute to increasing inequality in access to education, skill training and health care within countries. The solution is to omit from future trade agreements a requirement that intellectual property rights be enforced in all member countries, and instead to allow individual governments to implement IPRs when they become persuaded of their value, after observing and analyzing the experience of other countries. As a second choice, if IPRs are included, then very strong compulsory licensing laws should be included as well in order, at the very least, to make pharmaceutical products available at reasonable prices to people at all income levels.

The remaining gains to be had from free trade agreements are diminishing, and some of the non-trade provisions of these agreements are quite likely to have inequalizing effects, with their burdens falling on developing countries and lower-income households. In light of this, all the provisions should be very carefully scrutinized, and - as argued here - some should not be included in a Free Trade Agreement of the Americas, if we want to avoid exacerbating existing inequalities between and within member countries.

Endnotes

1. The annual employment is taken as the November figure instead of December, since December data are less representative due to the Christmas holidays.

2. The regression was re-run without the NAFTA dummy, and that is what is reported here.

3. That is, both lie in the same cone of diversification.

References

Bell, Linda A. 1997. The Impact of Minimum Wages in Mexico and Colombia. Journal of Labor Economics 15(3), part 2: S102-S135.

Cullen, Julie Berry and Jonathan Gruber. 2000. Does Unemployment Insurance Crowd Out Spousal Labor Supply? Journal of Labor Economics 18(3): 546-572 (July).

Deardorff, Alan V. 1992. Welfare Effects of Global Patent Protection. Economica 59(233): 35-51. February.

Deardorff, Alan V. 1986. FIRless FIRwoes: How Preferences Can Interfere with the Theorems of International Trade. Journal of International Economics 20: 131-42.

Dussel-Peters, Enrique. 2000. Polarizing Mexico. Boulder, Colorado: Lynne Rienner.

Fields, Gary S. and Amanda Newton Kraus. 1996. Changing Labor Market Conditions and Income Distribution in Brazil, Costa Rica, and Venezuela. Cornell University.

Freeman, Richard B. 1995. Are Your Wages Set in Beijing? Journal of Economic Perspectives 9(3): 15-32, Summer 1995.

[GAO] General Accounting Office. 2001a. Trade Adjustment Assistance: Improvements Necessary, but Programs Cannot Solve Communities' Long-Term Problems. Statement of Loren Yager, Director, International Affairs and Trade. Testimony Before the Committee on Finance, Subcommittee on International Trade, U.S. Senate. GAO-01-988T. July.

[GAO] General Accounting Office. 2001b. Trade Adjustment Assistance: Experiences of Six Trade-Impacted Communities. Report to the Chairman and Ranking Member, Committee on Finance, U.S. Senate. GAO-01-838: Trade Adjustment Assistant in Case Study Communities. August.

Gereffi, Gary. 2000. The transformation of the North American apparel industry: Is NAFTA a curse or a blessing? Desarrollo Productivo no. 84. Investment and Corporate Strategies, Division of Proudction, Productivity nad Mangaements, Unit on Investmenta nd Corporate Strategies, CEPAL. Santiago, Chile: CEPAL. October.

Hamrick, Karen S., Stephen A. MacDonald, and Leslie A. Meyer. 2000. International Trade Agreements Bring Adjustment to the Textile and Apparel Industries. Rural Conditions and Trends 11(1): 31-41. (United States Economic Research Service publication).

Hanson, Gordon H. and Ann Harrison. 1999. Industrial and Labor Relations Review 52(2): 271-288 (January).

Harrison, Ann and Edward Leamer. 1997. Labor Markets in Developing Countries: An Agenda for Research. Journal of Labor Economics 15(3), part 2: S1-S57.

Hinojosa-Ojeda, Raúl and Sherman Robinson. 1991. Alternative Scenarios of U.S.-Mexico Integration: A Computable General Equilibrium Approach. Giannini Foundation of Agricultural Economics Working Paper No. 609. University of California, Berkeley.

[INEGI] Instituto Nacional de Estadística, Geografía e Informática. 2002. Banco de Información Económica.  

[INEGI] Instituto Nacional de Estadística, Geografía e Informática. 1999. Estadísticas Econ&oiacute;micas: Industria Maquiladora de Exportación. Mayo.

Koechlin, Tim and Mehrene Larudee. 1992. The High Cost of NAFTA. Challenge 35(5): 19-26 (September-October)

Larudee, Mehrene. 1998. Integration and Income Distribution under the North American Free Trade Agreement: The Experience of Mexico. Ch. 11, pp. 273-292 in Dean Baker, Gerald Epstein and Robert Pollin (eds.), Globalization and Progressive Economic Policy, Cambridge: Cambridge University Press.

Larudee, Mehrene. 1995. Trade Liberalization and Income Distribution: Three Essays with Reference to the Case of Mexico and the North American Free Trade Agreement (NAFTA). Ph.D. dissertation, University of Massachusetts, Amherst.

Levy, Santiago and Sweder van Wijnbergen. 1994. Agriculture in the Mexico-U.S. Free Trade Agreement. Pp. 151-194 in Joseph F. Francois and Clinton R. Shiells (eds.), Modeling Trade Policy: Applied General Equilibrium Assessments in North American Free Trade. Cambridge: Cambridge University Press.

López-Córdova, J. Ernesto. 2001. NAFTA and the Mexican Economy: Analytical Issues and Lessons for the FTAA. INTAL - ITD - STA [Institute for the Integration of Latin America and the Caribbean - Integration, Trade and Hemispheric Issues Division - Statistics and Quantitative Analysis Unit].

Lustig, Nora Claudia and Darryl McLeod. 1997. Minimum Wages and Poverty in Developing Countries: Some Empirical Evidence. Ch. 3, pp. 62-103 in Sebastian Edwards and Nora Claudia Lustig (eds.), Labor Markets in Latin America, Washington, DC: Brookings Institution.

Maskus, Keith E. 2000. Intellectual Property Rights in the Global Economy. Washington, DC: Institute for International Economics.

Morley, Samuel A. 1995. Poverty and Inequality in Latin America. Baltimore: Johns Hopkins. Primer Informe de Gobierno. 2001. 

Rodríguez-Archila, Laura. 2000. Apparel Market: New U.S. Legislation Places CBERA Countries on a More Equal Competitive Basis with Mexico. Industry Trade and Technology Review (July), pp. 21-32.

Tang, Roger Y. W. 1997. Intrafirm Trade and Global Transfer Pricing Regulations. Westport, Connecticut: Quorum Books.

Tobin, James. 1994. A Tax on International Currency Transactions. P. 70 in United Nations Development Program, United Nations Human Development Report 1994, New York: Oxford University Press.

U.S. Customs Guide to NAFTA Implementation. May 4, 2002.

[USITC] United States International Trade Commission. 1997. The Impact of the North American Free Trade Agreement on the U.S. Economy and Industries: A Three-Year Review. Investigation No. 332-381, Publication 3045 (June).

[USITC] United States International Trade Commission. 1994. The First Ten Years of CBERA. Annual Report on the Impact of the Caribbean Basin Economic Recovery Act on U.S. Industries and Consumers. Report to Congress and the President on Investigation No. 332-227 Under Section 332(b) of the Tariff Act of 1930. Ninth Report 1993. Investigation No. 332-227. USITC Publication 2813. September. Washington, DC: USITC.

[USITC] United States International Trade Commission. 1993.Potential Impact on the U.S. Economy and Selected Industries of the North American Free-Trade Agreement. Report to the Committee on Ways and Means of the United States House of Representatives and the Committee on Finance of the United States Senate on Investigation No. 332-337 Under Section 332 of the Tariff Act of 1930. USITC Publication 2596. January.

[USITC] United States International Trade Commission. 1990. Review of Trade and Investment Liberalization Measures by Mexico and Prospects for Future United States-Mexico Relations. Investigation No. 332-282. USITC Publication 2275. April. Washington, DC: USITC.

Weiner, Tim. 2002. In Corn's Cradle, U.S. Imports Bury Family Farms. New York Times, Febuary 26. Whiting, Van R., Jr. 1992. The Political Economy of Foreign Investment in Mexico. Baltimore: Johns Hopkins University Press.

Wood, Adrian. 1997. Openness and Wage Inequality in Developing Countries: The Latin American Challenge to East Asian Conventional Wisdom. World Bank Economic Review 11(1): 33-57 (January).

WTO Doors Open to China. Business Mexico October 2001: 12-14.

Zahniser, Steven and Florencio Treviño. 2001. Hired Farm Labor: Comparing the U.S. and Mexico. Agricultural Outlook January/February 2001: 14-18. Economic Research Service, U.S. Department of Agriculture.