The North American automobile industry has been in a process of spatial and organizational restructuring since the early 1970's. Restructuring has however exhibited two fairly distinct trends so that no clear consensus exists about the current direction of the industry. Some argue that automobile production will continue its exodus from Canada and the United States to low wage areas, especially Mexico, with the ratification of the North American Free Trade Agreement (NAFTA). Others argue that the increasing use of microelectronics and flexible production will not only halt the shift of manufacturing to low wage areas but reverse the trend. Activities exported to low wage areas will return to the core due to both the difficulties associated with operating in peripheral regions and a significant reduction in the labor costs resulting from the use of these new technologies of production.

Both of these trends have some basis in reality because both refer to observed tendencies in North American auto production. Determining which of these trends will predominate over the next five years is important for several reasons. There are specific concerns about the future of North American automobile production. "Mid-range" concerns about the future of the Canadian and American economies. Finally, there are general concerns about the global restructuring of production. One trend, exodus, implies a further deindustrialization of Canada and the United States and the continued growth of manufacturing in peripheral areas. The other trend implies a reindustrialization of Canada and the United States and a possible slowing in growth of peripheral industrialization.

By way of an introduction, I will briefly anticipate my answer to the question posed in the title: where is North American automobile production headed? In a word, south -- it will continue to flee Canada and the United States for low wage regions to the south of the US, especially Mexico. Given the previous discussion of one trend in the restructuring of automobile production, namely, the shift of production to low wage areas, it could be argued that this is nothing new. North American automobile companies have been sourcing labor intensive and relatively sophisticated components from Brazil and Mexico for the last 20 years. I will argue that the ongoing move to the south is something new. It is different from the earlier relocation of production to these areas along two important dimensions: the sites where production is being relocated and, more importantly, the "organizational logic" underlying this relocation.

This paper will analyze and discuss the ongoing shift of North American automobile production to Mexico by comparing and contrasting it with the two trends already identified. The earlier movement of production to low wage areas is part of a larger strategy among North American automobile producers that I call "transnational production." Underlying the claims about the return of manufacturing to core areas is a strategy derived from Japanese automobile companies which can be called, after Womack et al. (1990), "lean production." I will discuss the benefits and limitations of "transnational" and "lean" production and then relate these to the shift of automobile production to Mexico, which I call "low wage lean production". This comparison will demonstrate how "low wage lean production" represents, on the one hand, a continuation of the trend in automobile production to reduce labor costs through the use of lower waged labor, evident in transnational production, but also a break with the organizational logic of transnational production because it incorporates several aspects of the organizational logic of "lean production."

TRANSNATIONAL PRODUCTION

In response to increasing competition by the Japanese for the low price market, and rising production costs, in the mid- 1970's Ford and General Motors (in North America and Europe) embarked on a strategy of "transnational production" (Dassbach, 1989), "global Fordism" (Lipietz, 1987) or "global factories" (Hamilton, 1981; Hill, 1990; Dicken, 1992).This strategy combined old, established Fordist manufacturing practices with a new spatial distribution, integration, and coordination of production.

Transnational production embodied three established Fordist manufacturing practices. One was the endeavor to increase economies of scale by locating similar activities at the same site. The second was the continuation of what Lipietz (1987: 71) calls

Fordism's...division of activities into three levels: 1) conception, organization of methods, and engineering..., 2) skilled manufacturing, which requires a fairly skilled workforce; and 3) unskilled assembly and execution, which in theory requires no skill.

The third was Fordism's shop floor organization, namely, the extensive and intensive use of dedicated automation, the lack of stringent in-process quality control, the production of large stocks of inventory in anticipation of possible disruptions or "just-in-case" inventory practices, and the restriction of workers to a limited number of highly specialized tasks through an elaborate system of job classifications.

To these, transnational production added several new elements. One was a new spatial distribution of manufacturing operations. Similar or related activities were not simply concentrated at some site to exploit economies of scale but concentrated at those sites in the world which enabled companies to exploit economies of scale in conjunction with some local advantage such as R & D experience, specialized labor skills, manufacturing expertise and, most importantly, pools of low wage labor. In other words, where Fordism tripartite division of activities was previously articulated nationally and organizationally as a division between activities in the same nation, region or city, transnational production articulated this division globally. Perhaps the first to identify this trend was Stephen Hymer (1979: 157-158) when he wrote in 1971:

A regime of multinational corporations would tend to produce a hierarchical division of labor between geographical regions corresponding to the vertical division of labor in the firm. It would tend to centralize high-level decision making in a few key cities in the advanced countries, surrounded by a number of regional subcapitals, and confine the rest of the world to lower levels of activity and income.

The second was the administrative integration and coordination of spatially dispersed activities through telecommunications networks. Specifically, the emergence of transnational production was both accompanied by, and made possible through, a massive expansion of high speed data processing and communications links within enterprises. These telecommunications systems enabled North American auto firms to quickly and easily share R & D performed at different sites in the world, transmit manufacturing specifications and product modifications instantaneously from development to manufacturing centers, specify rates of output at manufacturing sites around the world and, most importantly, monitor and control the globally dispersed activities of the enterprise from a central site. Activities were materially integrated through the transhipment of large quantities of components from production site to final market where they were assembled into vehicles.

Concretely, transnational production's assumed different guises depending on its geographical scope. If production was dispersed and integrated throughout a "region" such as Europe or North America, this was referred to as "regionalization." If production was dispersed and integrated across different geographical regions such as Europe, North and South America or Asia, this was referred to as "globalization." Central to both "regionalization" and "globalization" was a common product designed to be sold in all or most of the markets where its components were manufactured.1 Perhaps the best known, but by no means the only,2 examples of such products were "world cars" such as the Ford Escort, Ford Fiesta and GM J-car. These small, four cylinder automobiles, were sold, albeit at different market levels, in most of the countries where the components were manufactured.3 By the late 1970s, it appeared that transnational production represented the future of automobile production and North American companies expanded and strengthened their transnational networks.4

By the mid-1980's however, internal and external limitations in the organizational logic of transnational production became evident. External limitations took the form of inadequate infrastructural support for advanced manufacturing activities in peripheral countries and the failure to develop the expected levels of local demand for vehicles. Auto companies in peripheral regions found it extremely difficult to locate the supplies, replacement parts, and maintenance/repair technicians for their advanced, imported machinery. 5 Because peripheral markets failed to develop the anticipated levels of demand for world cars,6 local assembly facilities never ran at capacity, thus raising costs and further undermining sales. Globally, weak sales in peripheral markets, though relatively small compared to total production, resounded through the entire system. Total production never achieved the anticipated volumes and this served to raise overall production costs.

The internal limits of transnational production merely reflected the internal limits of Fordism as practiced by U.S. auto companies. In exporting production to low wage regions, auto companies also exported Fordist practices. As a result, the problems confronting North American plants were replicated and, to a certain extent, exacerbated in foreign plants. Like North American plants of the early 1980s, foreign plants were inflexible. Technically, inflexibility was a consequence of the use of dedicated automation. Socially, inflexibility was the result of the reproduction of a system of rigid job classifications which restricted workers to a limited range of activities (Katz, 1985). For example, Ford's Mexico City plant had as many separate job classifications, 150, as Ford plants in the United States and Canada (Shaiken, 1987).7 Foreign plants also experienced the same quality problems as American and Canadian plants which were compounded by larger than average production runs. Fordism's lack of stringent in-process quality control, combined with large production runs because components were manufactured for export, meant that foreign plants could produce large stocks of defective components. In many cases, these defects were only discovered at remote assembly sites months after they had occurred.

LEAN PRODUCTION

While the limitations in transnational production forced auto companies to reappraise this strategy, the success of Japanese automobile companies in North America and Europe undermined the organizational logic of transnational production. By the mid-1980's, several Japanese companies had established "transplants" in these markets and were successfully building vehicles comparable in quality and price to those built in Japan (Rowand, 1983; Hoerr, 1989). Ironically, in the U.S., Japanese companies such as Toyota or Mazda were building cars with U.S. workers, some unionized, in plants that American companies had abandoned a few years earlier. The success of the transplants was mainly due to the fact that the Japanese had successfully recreated their distinctive organization of production, or "lean production," in these overseas markets (Institute of Social Science, 1990; Womack, et al. 1990; Kenney and Florida, 1993).

There are several important differences between the organizational logic of lean production and Fordism. One is shop floor organization in terms of the actual social organization of production and labor-management relations. According to Kenney and Florida (1988: 131-133), the shop floor organization of lean production:

replaces the fundamental characteristics of Fordism -functional specialization, task fragmentation, and assembly line production - with overlapping work roles, job rotation, team-based work units, and relatively flexible production lines... Work organization is based on self-managing teams. Work roles are assigned to groups of workers, who then reallocate them internally.... Multi-skilling is absolutely essential for this strategy to be successful.... Rotation within teams allows workers to familiarize themselves with various aspects of the work process...This rotation scheme extends to the entire plant. Workers subsequently master the different tasks and grasp the interconnectedness among them.

The basic organizational unit on the shop floor is not the isolated, individual worker found under Fordism but rather the work team. Teams are headed by an hourly paid team leader and consist of four to eight workers. The emphasis within work teams is on group identification, worker involvement, and job flexibility. To insure flexibility, lean plants have replaced the 150 or so rigid job classifications of Fordist plants with one classification for production workers and one, or perhaps two, for maintenance workers. Workers are cross-trained and regularly rotated to different jobs. Unlike traditional plants, most workers in lean plants are also responsible for performing routine maintenance on their equipment.

Workers in these plants have a far higher degree of control over both their working conditions and final products than in conventional plants. They are constantly encouraged to make suggestions to improve their job, associated jobs or finished products and these suggestions are regularly implemented. A system of signal lights or, in Japanese, andon board, enables workers to summon help if they are experiencing difficulties with their work. There are few quality inspectors and quality control is the responsibility of the workers. One of lean production's most radical breaks with Fordism is empowering workers to stop the line if they are experiencing difficulties with their work which cannot be resolved without assistance or if there are significant problems in quality. Workers who stop the line are not upbraided or punished by management. Instead, line stoppages are viewed by management as an indication of worker involvement and as a means to uncover and resolve difficulties.

Through the use of techniques originally developed by Demming and subsequently refined in Japan during the 1950s, lean producers achieve far more stringent in-process quality control than Fordist firms. Hence, there are fewer manufacturing defects. Lean production is also far more flexible than Fordism. Socially, flexibility is the result of the lack of rigid job classifications and the ability of workers to perform a variety of tasks. Technically, flexibility results from the fact that lean producers avoid the use of dedicated automation. In fact, unlike Fordist firms who often implement new types of automation without considering whether workers are replaced or just converted into machine attendants, lean producers eschew the use of automation unless it truly displaces at least one worker (Monden, 1983).

Lean producers have also recast labor-management relations in the automobile industry by replacing traditional adversarial relations with relations of cooperation and consensus. This has entailed eliminating many of the privileges and perquisites accorded to management in traditional plants, requiring managers and workers to wear the same company uniform, and adopting an "open door" policy where workers are free to go to any level of management with a problem or complaint. Lean producers stress that management and workers are partners in the firm. As such, both parties will share in the profits resulting from improved production as well as the employment and salary reductions occasioned by downturns.

A second, and perhaps more important 8 difference between lean production and Fordism is the organization of the enterprise. Historically, Fordism has internalized or incorporated previously external activities within the enterprise. Internalization enabled Fordist firms to eliminate market transactions between suppliers, capture or eliminate intermediate profits and extend their span of control over an ever wider range of activities (Chandler, 1977). This strategy proved to be extremely successful, and highly profitable, as long as sales were expanding (1950-1970), but internalization also had severe and unanticipated limitations which only became apparent with a protracted downturn in sales.

First, as companies increased in size, more and more administrative, or "non-value adding," workers were needed to coordinate and control the various activities.9 Barring massive lay-offs, these administrative workers became a fixed overhead which had to be carried. Second, daily control and coordination of activities became increasingly difficult as the number of administrative layers multiplied.10 Third, connective bargaining and the standardization of wages raised wage costs to the same level in all the activities controlled by the enterprise (Katz, 1985). Finally, companies even abandoned Sloan's basic principle that the cost of internally supplied components had to be competitive with externally supplied components in order to keep their component divisions working (Dassbach, 1989).

Where the organizational logic of Fordism has been to internalize, the organizational logic of lean production has been to externalize. Instead of creating a giant, integrated, enterprise, each lean producer has created their own cooperating network, in Japanese, keiretsu, of banks, sales and export companies, and several layers of quasi-independent and independent subcontractors and suppliers (Cusumano, 1985; Smitka, 1991; Sheard, 1983; Hill, 1989). With the exception of banks, the companies in these networks fall into one of four general categories: firms which were originally departments in the main company that were "spun off" into independently operating companies, companies in which the main firm had an equity interest of at least 20%, companies in which the main company had an equity interest of less than 20%, and finally, totally independent suppliers (Cusumano and Takeishi, 1992). The coordination and integration of activities in these networks is achieved through a complex set of mechanisms which include, in addition to partial ownership of suppliers by the main firm, the movement of personnel between main firm and suppliers, the exchange of R &D and, most importantly, trust (Smitka, 1991; Wright, 1990; Asanuma, 1989; Reich 1988). 11

These networks have several important consequences for the organization of the enterprise. One is that lean firms produce a significantly smaller portion of their components in-house than Fordist firms.

While lean firms produce fewer parts in-house, they still control, via their networks, a far greater proportion of component production than Fordist firms.

Another consequence of these networks -- a higher output per employee ratio 13 -- is evident in Table 3.

In 1989, the best Fordist firm, Peugeot Citroen produced 14.6 vehicles per employee while the worst lean firm, Nissan, produced 19.6 vehicles per employee. This does not mean, as is generally assumed, that productivity in either individual firms or the Japanese auto industry as a whole is higher than North American or European firms or the North American or European industry as a whole,14 only that a smaller share of the labor in automobile production is directly employed in the main firms. Hence, lean firms have lower operating overhead which is especially important during periods of depressed sales.

A third consequence of these networks is the ability of lean producers to maintain wage differentials across different activities, especially between component manufacture and final assembly. Where Fordist internalization combined with connective bargaining led to an equalization of wages across all the activities incorporated into the enterprise, networks in the Japanese auto industry maintain a hierarchy of wages across productive activities. This is due to the fact that different activities occur in firms of different sizes and the most important variable in determining wage rates, as shown in Table 4, is firm size (Cusumano 1985,Smitka 1991). 15

A fourth consequence of these networks is that they permit an externalization of development costs and the associated overhead. The uneasy and distrustful relationships between Fordist firms and their suppliers have led auto companies to design components in-house and then convey precisely detailed drawings and specifications to their suppliers. As a result, Fordist firms are forced to carry most of the engineering and design costs as well as the associated overhead. In lean networks, suppliers perform this function in addition to actually manufacturing the parts (Helper, 1990; 1991). In 1988, for example, suppliers developed 80% of the parts purchased by Toyota (Smitka, 1991). As a result, main firms are freed from a considerable portion of the cost burden and overhead of product design.

The final two differences between the organizational logics of Fordism and lean production are interrelated. One is inventory practices, which directly impacts on component quality, and the other, which is a consequence of inventory practices, is the spatial distribution of production. Fordism's inventory practices have been characterized as "just in case" while lean production's inventory practices have been characterized as "just in time." Fordist companies have historically tended to produce and store large stocks of components "just in case" of disruptions in supplies due to strikes, transport problems, shortage of materials, etc. As already noted, production of large stocks of parts combined with lax in-process quality control often led to large runs of defective parts. These large runs of parts also meant that considerations of proximity between component producers and main company were not extremely important. If a supplier provided a 60 day supply with each delivery, the supplier could be located anywhere in the country, if not the world, hence, transnational production. The only requirement was that deliveries arrived at the proper intervals.

Under a "just in time" system, suppliers produce components in small batches and deliver these to the main company as needed, often as frequently as two or three times a day. These parts are then assembled into vehicles almost immediately. Not only does "just in time" reduce the capital and the amount of storage space devoted to intermediate inventories but it also quickly reveals any manufacturing defects so that these can be rectified before the next batch arrives. Because JIT operates on the basis of frequent deliveries, spatial proximity becomes a paramount consideration. Suppliers must be relatively close to the main company, if not clustered around the main company as in the most extreme case of Toyota City in Japan.

To a large extent, the successful transplantation of lean production to North America and Europe by Japanese automobile companies was at the heart of arguments that the shift of production to low wage countries would halt. While proponents of this view may not use the term lean production, they suggest that the emulation and spread of these methods in the core regions will halt and eventually reverse the flight of production to low wage areas for three reasons (Womack, 1987). First, lean production drastically reduces labor content and this, in conjunction with political and economic instability of low wage regions, diminishes the lure of low wage areas. Second, the newly industrializing countries have neither the skills nor the infrastructure to support the new manufacturing technologies and techniques. Third, it is logistically "desirable", especially for just-in-time production, to have the majority of manufacturing operations not only located in a given market but in proximity. Some even assert that manufacturing activities expelled from North America over the prior 20 years will eventually return (Castells and Tyson, 1988; Junne, 1987; Drucker, 1986; Lawrence and Litan, 1987; Sanderson, 1987). Whether expelled activities will return is unclear, but there is wide consensus that the diffusion of lean techniques, especially in automobile production, will result in a spatial restructuring and a new geography of production in North America. North American automobile producers will abandon the previous pattern of geographical dispersion and adopt a spatial distribution more akin to the regional concentration in Japan. Restructuring will occur largely by abandoning older sites in Michigan, California and the Northeast and moving to new greenfield sites in the Midwest (Hill, 1987; Ishi, 1988; Mair et al., 1988; Storper and Scott, 1989; Rubenstein, 1992).

LOW-WAGE LEAN PRODUCTION

Although some manufacturing complexes have developed in the United States and Canada around Japanese transplants as well as newer plants built by North American producers, the relocation of automobile and component production to Mexico appears to be a far more significant trend for the future of North American automobile production. Mexico has been integrated into both Ford and GM's transnational production systems since the mid-1970s, but only recently become an important site for automobile production by North American companies. Engine exports, mainly to Canada and the U.S., which hovered around 400,000 a year in the early 1980's, increased to 1.3 million between 1985 and 1989 (Shaiken, 1990). Vehicle exports grew from 14,000 to 186,000 between 1981 and 1989 (Shaiken, 1990) and, over the next three years, more than doubled, reaching a figure of 393,000 in 1992 (Werner, 1993).

The most important reason for this growth in Mexican production is the lure of low wages. Although advanced production technologies reduce the amount of labor required for production, reducing wages costs still remains of fundamental importance for auto production. Why? Because labor costs still represent approximately 20% of total costs. For example, in an advanced engine plant direct labor costs only represent 10% of the cost of manufacturing an engine but indirect labor costs for maintenance and repair account for another 12% so total labor costs amount to approximately 22% of manufacturing costs (Shaiken, 1987). In assembly, total labor costs are of a similar magnitude - 20% of costs. No mathematical wizardry is required to see the benefits of labor costs of $1.55 an hour in the north of Mexico, or even of $5.50 an hour in Mexico City versus $30 an hour in the United States (Flint, 1991). Auto producers, as already noted under the discussion of transnational production, were cognizant of Mexico's lower labor costs in the 1970's but the tremendous growth of Mexican production since the mid-1980s is rooted in the organizational logic of lean production.16

Most of the recent Mexican plants, like most of the Japanese transplants in the United States, are greenfield sites. Mexican greenfield sites, like greenfield sites in Japan and the United States, 17 shift production from urban centers to rural areas in order to take advantage of lower labor costs and a more tractable work force. According to Middlebrook (1991), wage costs in the north of Mexico are markedly lower for two reasons: the general level of wages is lower than in Mexico City and seniority-based fringe benefits are lower because the labor force in the northern plants consists of new workers. Workers in northern Mexican are more tractable for several reasons. They have little experience in industry and industrial conflict and, unlike workers in Mexico city who are organized into militant "democratic unions," these workers are organized by the CTM (Confederation of Mexican Workers) which has "close ties to the government" and a "generally cooperative attitude towards management" (Middlebrook, 1991: 279).

Unlike the older workers and militant unions in Mexico city which resisted the implementation of lean techniques, the CTM affiliated unions in the north wholeheartedly embraced lean techniques such as rotation, multi-tasking, work teams and the elimination of both seniority based promotions and restrictive job classifications. As a result, many of the new Mexican plants have the same flexibility as the best lean plants. In addition, the normal work shift in these plants is 9.5 hours versus 8 hours, which means less overtime pay. (Middlebrook, 1991; Shaiken, 1990).

Finally, unlike the previous tendency among auto producers to export old equipment and manufacture outdated models for sale in peripheral markets,18 most of the newer component and assembly plants in Mexico are close to "state of the art" (Shaiken, 1993). For example, GM's Deltronics plant in Matamoros, is housed in a modern air conditioned factory and employs 4,000 workers to assemble 3 million radios a year (Flint, 1991). The body shop at the Ford plant in Hermosillo was "designed, fabricated and tested" by Mazda of Japan and contained modifications based on Mazda's production experience with a similar model which were not incorporated into its own body shop (Shaiken, 1990). The tools used in Ford's Mexican engine plant are on par with tools used in American and Canadian plants (Haigh, 1992; Shaiken, 1987). Where previously Mexican production was for local sales or American companies manufactured "gas guzzling models" for export to the U.S.,19 Mexican operations now export a majority of their component production and approximately 40% of vehicle production (Werner, 1993).

Perhaps the most important plant for the future of automobile production in Mexico is Ford's Hermosillo plant. Hermosillo represents the most advanced assembly plant in all of North America, and maybe the world, with the best quality of the 80 plants, including Japanese plants and transplants, examined by Womack et al. (1990: 87). Like NUMMI, the GM-Toyota venture, which enabled GM to learn lean production from Toyota, Hermosillo enabled Ford to learn lean production from Mazda. 20 But unlike NUMMI which merely replicated existing Japanese practices, Hermosillo is striving to move beyond these. In Hermosillo, Ford has not only eliminated all classifications among assembly line workers, leaving two categories in the plant, assembly worker and maintenance technician, but eventually plans to eliminate these. Hermosillo will then have one job classification and workers will rotate between assembly and maintenance jobs. Even more incredible, Ford has managed to implement a modified version of "just in time", called "one-day-at-a time" for the production of the first car at Hermosillo, the Tracer, despite the fact that 65% of the components came from Japan.21

While the search for low cost labor is the fundamental force in the relocation of automobile production to Mexico, other factors have also contributed to Mexico's growing attractiveness as a site for component production and final assembly. For example, auto makers have found that Mexican workers are highly motivated and capable of learning advanced production techniques (Chappell, 1991; Eden and Molot, 1993). At GM's Ramos Arizpe plant, the workers have "taken to quality programs with a near-religious zeal...[and] quickly mastered Japanese-style manufacturing techniques to become GM's No. 1 plant, setting company quality records" (Baker, Woodruff and Weiner, 1992: 100).22 In the last few years, the Mexican domestic market for automobiles has boomed. After stagnating since the early 1980s, annual domestic sales grew over 20% a year between 1988 and 1992 (Chappell, 1991) thus providing auto makers with a rapidly growing local market.

Two political changes have been especially important in making Mexico the new center of North American automobile production: the rescinding of a series of restrictive governmental decrees in 1989 which regulated the Mexican auto industry for the previous 20 years and NAFTA. Under these decrees, i.e., prior to 1989, imports of assembled vehicles into Mexico were forbidden, a domestic content of at least 75% was required for all cars sold in Mexico, and manufacturers were restricted to a few car lines. As a consequence, most of Mexican production was isolated from U.S. and Canadian production. In 1989, these restrictions were replaced with two new requirements: imports could not exceed 15% and later 20%, of domestic sales and companies had to export a certain dollar value for every $1.00 they import: $2.50 in 1989 and 1990, $2.00 in 1992 and 1993, and $1.75 in 1994. This meant that U.S. companies, in particular, could begin importing cars and rationalizing model lines across North America (Baker, 1989; Downer, 1991; Eden and Molot, 1993). Where the 1989 decree liberalized conditions for companies already in Mexico, NAFTA restricted access to Mexico and also insured that Mexico would not become an off-shore assembly platform for foreign companies seeking to penetrate the North American market. NAFTA placed a ten year moratorium on new entrants into the Mexican auto industry and restricted duty free access to North American markets to vehicles with at least a 62.5% North American content (Eden and Molot, 1993).

Womack et al. (1990: 265) have predicted that there will not be a large-scale relocation of North American automobile production to Mexico. Instead, Mexican production will be limited to low cost entry vehicles and larger vehicles and trucks will be imported from the U.S. and Canada to Mexico. Mexican production of several larger and more expensive models such as the Buck Century, Chrysler LeBaron, Dodge Ramcharger and new Dodge truck has already disproved this prediction (Berg, 1992). Already, over 75,000 U.S.23 and 24,000 Canadian (Eden and Molot, 1993: 298) auto jobs have been shifted to Mexico and, as producers continue with their plans to close American and Canadian factories, more jobs will be shifted to Mexico despite protests by the UAW and CAW (Kerklewich, 1993).

The fundamental motive for the shift in North American vehicle production to Mexico remains the same as with transnational production, the lure of low wage labor. But the recent shift to northern Mexico differs from transnational production insofar as it incorporates the organizational logic of lean production. At the same time, the growth of Mexican automobile production shows is that the reality of lean production is unlike the "theory" of lean production. Theoretical discussions of lean production often assert that it will result in the retention of manufacturing jobs in the core areas by enabling companies to profitably build automobiles using high wage labor. area. In reality, North American auto companies have come to a different conclusion: if "high wage lean production" can be profitable, "low wage lean production" can be even more profitable.24