Skip to navigation

Skip to main content

Research & Seminars

contact us

The Structural Embeddedness of Demand for Immigrant Labor in California and Japan -- Wayne Cornelius

Preliminary draft; do not quote
without author's permission


By Wayne A. Cornelius

Department of Political Science and

Center for U.S.-Mexican Studies

University of California-San Diego

La Jolla, Calif. 92093-0510

Prepared for the Winter 1998 Workshop of the
Comparative Immigration and Integration Program
University of California-San Diego
February 20, 1998


By Wayne A. Cornelius


A central paradox posed by the most recent wave of international migration to the United States is that poorly educated, low-skilled immigrants are being incorporated readily into a high-tech, knowledge-intensive, postindustrial economy that ostensibly has no need for this type of labor. In Japan, where migrant workers from the Asia-Pacific region are predominantly unskilled and those from Latin America typically have higher educational and occupational skill levels (many have held white-collar jobs prior to emigration), the vast majority of foreign workers of all nationalities are being used employed in low-skilled manufacturing, construction, and service jobs.

The orthodox, neo-classical labor economics paradigm commonly used to rationalize more restrictive or more skill-based immigration policies in the United States stresses that job opportunities for low-skilled workers - natives as well as immigrants - are already quite limited and rapidly disappearing. Moreover, the economic benefits resulting from an ample supply of predominantly low-skilled foreign workers are frequently depicted as minimal and constantly diminishing (Borjas, 1994, 1996; Briggs, 1995; McCarthy and Vernez, 1997; Chiswick, 1998).

In Japan, government officials and scholars seeking to prevent their country from becoming a major importer of unskilled labor argue that the restructuring of the economy - manifested by the relocation of production overseas, and a constant rise in the share of finished products in Japan's total imports - will eliminate the need for large-scale immigration. Thus, deindustrialization, or the "hollowing out" of the Japanese manufacturing sector, will cause a sharp decline in the need for unskilled and semi-skilled, manual labor. Moreover, like their U.S. counterparts, Japanese employers are believed to have a variety of alternatives to the use of foreign labor, from moving more production abroad to greater investment in labor-saving technologies, rationalization of production, and more aggressive recruiting of native-born workers - part-timers and so-called "underutilized" labor sources like middle-aged women, the elderly, and the disabled. It is assumed that native workers can be substituted almost infinitely for low-skilled immigrants now in the labor force, if both employers and native-born residents can be suitably incentivized by government. Pursuing some combination of these options should enable employers to meet their labor requirements without recourse to foreign workers (Iguchi, 1998).

It follows from such arguments that purging the superfluous foreign workers from U.S. and Japanese labor markets would involve, at most, disruptions that are very short-term and entirely manageable. Unwanted immigrants - particularly illegal entrants and visa over-stayers -- can be dislodged from the work place, and penetration of industries and job categories in which immigrants are not yet dominant can be reversed or prevented. Failure to do so is simply a reflection of employer greed and lack of political will.

This line of argument rests on a series of largely untested empirical assumptions about the character and determinants of the demand for immigrant labor in the United States and Japan. Why do employers use this labor source? (We cannot simply assume, "Because it's cheaper.") How did they come to rely upon it - through what processes or mechanisms? Why do many employers actually prefer to hire immigrants, even if native-born workers are theoretically available to be recruited? What alternatives to the use of foreign labor do employers perceive, and if they have pursued any of these options, how well have they worked?

If the demand for immigrant labor has become a structural feature of the receiving economy, what are the empirical indicators of such structural embeddedness? Does demand expand or contract significantly with the business cycle, or does it it remain constant or continue to grow at a steady pace, regardless of macroeconomic fluctuations? Do major changes in the regulatory environment -- in public policies, laws, and law enforcement strategies -- affect the propensity of employers to rely on immigrant workers?

Finally, how do changes in law and public policy affect the supply of immigrant labor that is available to employers? For example, has the supply of undocumented immigrant workers in California been reduced by tighter border controls since 1994? Or, is the employment of immigrant labor so highly institutionalized that it is no longer sensitive to changes in the legal and public policy environment that are intended to deter new immigration or discourage permanent settlement by immigrants already here?

Research Design and Methodology
This paper is based primarily on data from a comparative study of how immigrant labor is being utilized in the economies of San Diego County and Hamamatsu, an industrial city of about 500,000 located south of Tokyo. Fieldwork in both research sites was carried out simultaneously during the first half of 1996. In San Diego, a randomly selected sample of 112 employers were interviewed, together with 501 immigrant workers who were employees of the same firms in which employers were interviewed (five or six randomly chosen immigrant employees per firm were interviewed), and 116 irregularly employed "street-corner workers" - migrant workers who seek day-labor in construction and landscaping by standing along public thoroughfares in various parts of San Diego County. In Hamamatsu, random samples of 104 employers and 336 foreign workers were interviewed.

In each research site, the universe for our surveys of employers consists of firms that were known, through telephone screening interviews and other methods, to be employing at least some foreign-born workers. Therefore, we can compare heavy users of foreign labor with firms that are not so dependent on this labor source, but we cannot compare foreign-labor users with firms that make no use of foreign workers.

In both San Diego and Hamamatsu, all sectors of the local economy in which foreign-born workers constitute a significant portion of the labor force are represented in the study (see Figure 1), with the exception of private household service (a significant source of employment for immigrant women in San Diego but not in Hamamatsu). In San Diego, interviews with employers and immigrant workers were distributed evenly across ten sectors: agriculture and horticulture, high-tech manufacturing, low-tech manufacturing (except clothing), apparel, food processing, construction, hotels and motels, restaurants, landscape and building maintenance, and miscellaneous services (dry cleaners, car washes, laundries, and convalescent homes). These industries account for 34.4 percent of nongovernmental employment in San Diego County. Four out of five of the sample firms were chosen at random from complete lists of businesses in the ten sectors covered by the study, compiled by a market research firm. The remainder consisted of firms that had been studied in two previous surveys of immigrant-using businesses in California, conducted by the author in 1983-84 and 1987-88 (see Cornelius, 1989). In Hamamatsu, interviewing was conducted in manufacturing, construction, food processing, hotel, restaurant, and service businesses, with manufacturing firms constituting 51 percent of the sample.

All interviews were conducted in person, using standardized questionnaires developed for each of the three groups of interviewees (employers, immigrant employees, and, in San Diego, street-corner workers). In each firm, the person having primary responsibility for hiring decisions was interviewed (in most cases, this proved to be the firm owner). Interviews with employers averaged about one hour in duration; interviews with immigrant workers lasted about 45 minutes. In San Diego, over half of the workers interviewed were undocumented (77 percent among the street-corner workers). Seventy-nine percent were Mexican nationals, although a dozen other nationalities are also represented in the sample.

Among our sample of foreign workers in Hamamatsu, only 8.7 percent were working in Japan illegally. Illegals are undoubtedly underrepresented in our Hamamatsu sample, as they would be in any survey study of Japan's foreign worker population. Illegal workers in Japan are extremely difficult to locate, since they tend to be employed in firms so small that they are not included in the official lists of businesses maintained by the Labor Ministry, local governments, and chambers of commerce. Moreover, they keep an even lower profile than undocumented immigrants in the United States. Brazilian nationals make up 45 percent of our foreign-worker sample in Hamamatsu, but 17 other nationalities are also represented.

In most respects, San Diego County, with a population of about 2.9 million, is representative of the large metropolitan areas that now absorb the bulk of both legal and illegal immigrants to the United States. However, its economy differs from that of other major immigrant-receiving cities like Los Angeles, Chicago, Houston, and New York in one important way: the San Diego region has a major, still-expanding agricultural and horticultural industry, which in 1996 produced $1.1 billion in crops and flowers. Agriculture and horticulture in San Diego are highly labor-intensive, and this sector is among the heaviest users of Mexican migrant labor (mostly originating in the impoverished southeastern state of Oaxaca). Like the other major destinations for long-distance Mexican migrants to the United States, San Diego has lost much of its traditional manufacturing base since the 1980s, but growth in the service, retail, and construction sectors has more than offset the loss of manufacturing employment, and high-tech manufacturing is robust.

Until the early 1980s San Diego served primarily as a corridor for Mexican migrants headed to Los Angeles and other cities to the north and east; during the last 15 years, however, San Diego County has become an important destination for Mexicans seeking U.S. employment. Much tighter border enforcement by the U.S. Immigration and Naturalization Service in the San Diego sector since October 1994 has further reduced San Diego's role as a mere way-station for Mexican migrants in transit (see the discussion of public policy changes, below). San Diego's immigrant population is dominated to a greater extent than other major immigrant-receiving cities by a single nationality: Mexican. As a consequence, Mexican immigrants in San Diego usually do not find themselves in competition for the same kinds of jobs with immigrants from the Caribbean (e.g., Mexicans vs. Dominicans in New York City), nor from Central America and the Asia-Pacific region (as in Los Angeles).

Hamamatsu is representative of the industrial satellite cities in Japan where large numbers of "newcomer" foreign workers have clustered since the late 1980s, especially the ethnic Japanese (Nikkeijin) from Latin America. The city is home to hundreds of small and medium-sized subsidiaries and subcontractors to several of Japan's largest manufacturing conglomerates, including Yamaha, Honda, Suzuki, Nissan, Mitsubishi, and Fuji. Because more than half of its total employment is in the manufacturing sector, a somewhat higher percentage of Hamamatsu's foreign-born residents are factory workers than would be found in large metropolitan areas of Japan having significant foreign-worker populations (e.g., Tokyo, Osaka). Hamamatsu also has a higher proportion of Latin American Nikkeijin in its foreign worker population, which reflects social network-mediated migration as well as employer recruitment in the Latin American source countries (mainly Brazil and Peru).

Characteristics of the Demand for Immigrant Labor

We found high levels of reliance on immigrant labor throughout the San Diego economy. While the percentage of immigrant workers varies considerably by sector, from 50 percent in manufacturing to 92 percent in agriculture and food processing (see Figure 2), on average two-thirds (66.7 percent) of the production workers in our sample firms were foreign-born, according to the employers interviewed. One out of five firms in the sample reported that more than 90 percent of its local work force consists of immigrants, and one out of ten admitted that they have a 100-percent foreign-born labor force, i.e., they "never" hire U.S.-born people, at least as production workers. This stratum of firms is not just immigrant-dependent; it is totally reliant on this labor source.

Japanese employers typically have a much lower level of dependence on foreign workers, averaging only 7.5 percent of the total production work force in our sample of Hamamatsu firms. As in San Diego, the percentage of foreign workers varies considerably by sector of the economy, ranging from 2-3 percent in hotels, restaurants, and food processing to 7 percent in construction, 10 percent in manufacturing, and 19 percent in service businesses. The proportion of foreign workers in these firms is creeping upward, as has been the case nationally during the 1990s. And as reported below, only a small minority of the firms studied in Hamamatsu expect to be able to reduce their utilization of foreign labor in the foreseeable future.

The already strong representation of foreign workers in Hamamatsu's service sector may be a harbinger of Japan's future as a country of immigration. Most services cannot be relocated abroad, and service businesses tend to rely heavily on unskilled and casual work forces. Thus, foreign workers are likely to be employed increasingly in the service sector. Foreigners have not yet been able to break into domestic services (house-cleaning, in-home care of children and the elderly), which is in sharp contrast to the United States and most West European countries; but there are plenty of other niches in Japan's service industries that foreign workers can fill.

What types of firms use immigrant labor? In both San Diego and Hamamatsu, the vast majority of them are small businesses: the median total number of employees in our sample of San Diego firms is 46; in Hamamatsu, it is 55 workers. But most of these immigrant-using firms do not fit the stereotype of the small, financially precarious, high-turnover enterprise that hyperexploits immigrant workers.

The vast majority of our sample firms do not seem financially endangered. In San Diego, 74 percent of employers described their current operations as somewhat or very profitable. Another indicator of financial viability is the fact that nearly half (48 percent) of these San Diego firms are involved to some extent in exporting, and more than one out of ten has at least one production site in a foreign country (usually Mexico). More than 80 percent of the Hamamatsu employers characterized their businesses as somewhat or very profitable. Looking ahead, 92 percent of the San Diego firms were somewhat or very optimistic about their profitability and sales during the next three years. Hamamatsu employers were considerably less optimistic, reflecting the generalized discouragement in Japan about the country's economic prospects, after six years of recession. Even so, a majority of them (58 percent) were optimistic about future profitability and sales.

The firms in our sample are not rock-bottom, sub-minimum-wage employers. In San Diego, the mean hourly wage paid to entry-level production workers in 1996 was $5.62. Only agricultural enterprises and restaurants were paying their entry-level workers close to the legal minimum wage (see Figure 3). Average wages were not being kept down by employer-induced turnover aimed at preventing workers from accumulating seniority and demanding higher pay; average annual turnover among production workers in the sample firms was only 10 percent, and most employers claimed that they wanted as stable a workforce as possible. In our Hamamatsu sample firms, the mean hourly wage paid to entry-level production workers averaged the equivalent of (U.S.) $12.81 per hour, with less intersectoral variation than we found in San Diego. The most poorly-paid workers in Hamamatsu were in the restaurant industry, averaging $9.72 per hour (see Figure 3).

There is no evidence from our study that immigrant-using firms discriminate significantly on the basis of legal status. The difference in average hourly wages between legal and illegal immigrant workers in our San Diego employee sample was small ($6.50 per hour for legals, $6.05 for the undocumented), and overall monthly earnings including overtime pay were also similar (averaging $1,167 for legals; $1,066 for illegals). Nearly all of the wage gap can be attributed to the legal immigrants' greater job seniority, longer residence in the United States, and ability to do jobs requiring English - not to legal status per se. A similar pattern was observed in Hamamatsu.

The wage gap between native workers and immigrants in general (irrespective of legal status) is more significant, particularly in Japan, where foreign workers typically earn about 20-30 percent less than the average native-born worker. However, it is possible for Japanese employers to hire foreigners at half the total cost of Japanese workers, not because of wage discrimination but because they do not pay foreigners the full range of employee benefits and annual cash bonuses expected by native workers. In this respect, the treatment of foreign-born workers is similar to that of Japanese part-time workers (mostly middle-aged housewives). In fact, the cost differential between hiring foreign workers and Japanese part-timers is minimal, and as noted below, there are significant advantages to using foreign workers in terms of convenience and flexibility, which explains why many employers prefer foreign workers to native-born part-timers, or make equal use of both types of labor.

The job skill requirements of the immigrant-using firms in our sample are of particular interest. Many of these firms use highly advanced technologies, but they continue to require substantial numbers of low-skilled workers. On average, our San Diego employers classified 83.2 percent of their production work force as unskilled or low-skilled. In our Japanese firms, a much lower but still significant portion of the work force - 37.6 percent - was considered unskilled. This is a key finding, in view of the conventional wisdom that the new information-based technologies require numerous highly-skilled, college-trained workers but very few of the less-skilled. The reality is exemplified by one of our San Diego sample firms that grows gourmet-quality mushrooms, using the latest hydroponic techniques developed in Japan. The company nevertheless relies on unskilled Mexican immigrants as production workers, to harvest and package the product. This high-tech agricultural employer is just as dependent on immigrant labor as the average tomato grower in San Diego County, who typically uses the most traditional methods of production.

The employers interviewed in both San Diego and Hamamatsu reported that only minimal training was needed for newly-hired, entry-level production workers in their firms: eight days of training in San Diego, and even fewer (five days) in Hamamatsu. These training requirements were corroborated by the immigrant workers whom we interviewed.

In San Diego, nearly half (46.4 percent) of the employers told us that there were no tasks in their business that they could not assign to immigrant workers. In Hamamatsu, the proportion was even higher: 58.8 percent of employers reported that they had no tasks that foreign workers could not perform. Among the San Diego employers who said that there were some tasks that they could not give to immigrants, 90 percent characterized these as jobs that required English speaking or comprehension ability, or tasks requiring direct contact with English-speaking customers (so-called "front-of-the-house" jobs).

All this suggests that there are many types of businesses that continue to serve as easy points of entry into the U.S. and Japanese labor markets for newly-arriving immigrants, irrespective of their skill level: not just farms and restaurants, but even high-tech manufacturing operations. In every sector of the economy, San Diego firms that hire immigrant workers use them in nearly every type of production job, and often in supervisory jobs as well. The proportion of low-skilled jobs in immigrant-using firms has actually risen during the past 13 years. In 1983, when we first surveyed such firms in San Diego, 58 percent of them reported that more than three-quarters of their workforce was unskilled; in our 1996 survey of the same firms, 68 percent said that more than three-quarters of their workers were unskilled.

The Highly Immigrant-Dependent Firm: An Empirical Profile
To identify the principal defining characteristics of immigrant-dependent businesses, we divided our samples of immigrant-using firms in San Diego and Hamamatsu into two groups: those that scored above and below the mean in terms of the percentage of foreign-born persons in the work force. Firms that ranked above the mean will be described below as highly immigrant-dependent firms, and those that scored below the mean as less-dependent firms. Comparing these two groups of firms, within each city, we find that highly immigrant-dependent firms in our sample differ from the less-dependent firms in the following ways:

They are more likely to be very small businesses. In San Diego, the median number of total employees is 30, as compared with 50 in less immigrant-dependent firms; in Hamamatsu, highly immigrant-dependent firms had an average of 27 employees, versus 62 in the less-dependent firms. In both San Diego and Hamamatsu, annual gross sales in highly immigrant-dependent firms were only half as large as those of less immigrant-dependent firms.
They are more likely to be recently established businesses, in both San Diego and Hamamatsu.
In San Diego, highly immigrant-dependent businesses are more likely to be owned by immigrants (especially Mexican nationals) and by Mexican-Americans. No question about the nationality or ethnicity of the employer was asked in the Hamamatsu survey.
Highly immigrant-dependent firms are somewhat less profitable. However, only 36 percent of such businesses in San Diego and 16 percent in Hamamatsu were just breaking even or losing money.
Highly immigrant-dependent firms rely more heavily on illegal immigrant workers (in San Diego, 65 percent of the employees of such firms were undocumented, as compared with 50 percent in the less immigrant-dependent firms).
In San Diego, the immigrant employees of highly-dependent firms are more likely to be purely "economic migrants" (motivated to come to the United States by job availability, higher wages, or economic crisis in their home country), as contrasted with migrants who came to the United States because of the presence of relatives and other non-economic reasons.
Immigrant employees of highly-dependent firms in San Diego are more likely to have worked in agriculture immediately before coming to the United States, and they are also more likely to be "first-timers" or persons with a limited history of U.S.-bound migration and shorter total residence time in the United States.
Immigrant employees of highly-dependent firms in San Diego have lower levels of educational attainment: an average of 8.0 years of schooling, versus 9.6 years among workers in less immigrant-dependent firms.
In both San Diego and Hamamatsu, highly immigrant-dependent firms have younger work forces. In Hamamatsu, the median age of workers in such firms is 34 years; the median age is 40 years in less-dependent firms. This age difference is probably an effect rather than a cause of high dependence on foreign labor.
In San Diego, highly immigrant-dependent firms have lower turnover among production workers (13 percent annually) than less-immigrant dependent firms (17 percent).
In both San Diego and Hamamatsu, highly immigrant-dependent firms are more likely to have very minimal training requirements for entry-level production jobs, and they are less likely to provide more advanced job training to the workers they hire.
They are more likely to recruit workers informally, through social networks -- i.e., referrals by current immigrant employees or supervisors. In San Diego, 80 percent of the production work force in highly immigrant-dependent firms had been hired in this way, as compared with 64 percent in the less immigrant-dependent firms. The same pattern was found in Hamamatsu, where highly immigrant-dependent firms tend to make more use of family-and-friend contacts to hire workers and make less use of professional labor brokers and other, more formal recruitment mechanisms.
In San Diego, highly immigrant-dependent firms are much less likely to require workers to have English language proficiency (27 percent, versus 64 percent in less immigrant-dependent firms).
Highly immigrant-dependent firms in San Diego are more likely to have a sharp division of labor between native-born and foreign-born employees. U.S.-born workers, if present, rarely do the kinds of jobs performed by immigrants in these firms. This difference is not so evident in Hamamatsu, where there is less segmentation between "jobs for foreigners" and "jobs for Japanese," in both highly immigrant-dependent and less-dependent firms. Among highly immigrant-dependent employers in Hamamatsu, 76 percent thought that there were no tasks in their business that cannot be given to foreigners.
In San Diego, highly immigrant-dependent firms pay lower wages, averaging $5.04 per hour for entry-level production workers, as compared with $6.02 per hour in less immigrant-dependent firms. Again, this wage differential is much less pronounced in Hamamatsu, where highly immigrant-dependent firms pay only $0.12 per hour less for entry-level production work than less-dependent firms - not a statistically significant difference. The absence of a significant wage differential between highly immigrant-dependent and less-dependent firms in Hamamatsu illustrates the lengths to which the highly-dependent firms must go to compete for labor, at least in terms of entry-level wages (as noted above, in Japan the major labor cost savings from hiring foreign works result from non-payment of certain fringe benefits).
Highly immigrant-dependent employers in San Diego believe it would be necessary to raise entry-level wages by an average of $3.10 per hour (a 62 percent increase over the current level) in order to attract U.S.-born job applicants; less immigrant-dependent firms believe they would need to raise hourly wages by $2.08 (a 35 percent increase). A comparable question was not asked in the Hamamatsu employer survey.
In both San Diego and Hamamatsu, highly immigrant-dependent firms are much less likely to provide employer-paid health insurance to production workers. In Hamamatsu, highly immigrant-dependent firms are much more likely to provide housing or housing allowances to their workers than less-dependent firms.
During the 1990s recession, highly immigrant-dependent firms in San Diego were more likely to reduce wages, overtime, and employee benefits to adjust for lost business, but not to reduce their work force through lay-offs.
During the 1990s recession, highly immigrant-dependent firms in San Diego found it no easier to recruit native-born workers than before the recession. In Hamamatsu, only 32 percent of the highly immigrant-dependent firms found it easier to recruit Japanese workers during the recession, as compared with 72 percent of the less-dependent firms.
Even in "normal" economic ties, native-born workers are much less well-represented in the job applicant pools of highly immigrant-dependent firms, in San Diego. In 71 percent of highly immigrant-dependent firms in San Diego, native-born workers never apply for the kinds of jobs held by immigrants, versus 35 percent of the less immigrant-dependent firms.
In both San Diego and Hamamatsu, today's highly immigrant-dependent firms expect to have an even more immigrant-dominated work force in the future. In San Diego, 80 percent of the highly-dependent firms expect foreign-born workers to account for a higher proportion of new hires in the future, as compared with 50 percent of the less-dependent firms. In Hamamatsu, 71 percent of the highly immigrant-dependent firms anticipate having an even more immigrant-dominated work force in the future, versus only 27 percent of the currently less-dependent firms.
The most immigrant-dependent firms in both San Diego and Hamamatsu are the least likely to have pursued alternatives to foreign labor utilization. In Hamamatsu, only 56 percent of highly immigrant-dependent firms have done something to reduce their reliance on foreign workers, as compared with 91 percent of the less-dependent firms. For example, the highly immigrant-dependent firms in Hamamatsu are much less likely to have moved any production to a foreign country (only 2 percent had done so, as compared with 18 percent of the less immigrant-dependent firms).
In both San Diego and Hamamatsu, highly immigrant-dependent employers believe that they would be severely impacted if they could no longer hire foreign workers. In Hamamatsu, 56 percent of the highly immigrant-dependent firms reported that they would have to reduce their operations or shut down the company if their supply of foreign labor were somehow to be cut off; only 11 percent of the less-dependent firms said that they would have to reduce production or close down. The pattern of responses in San Diego was quite similar.
To assess the relative importance of these factors in explaining dependence on immigrant labor in San Diego, we constructed a multivariate model that regresses ten firm characteristics on the percentage of foreign-born persons in the firm's production work force (the dependent variable), while controlling for sector of the economy. As summarized in Table 1, this model explains 68 percent of the variance in immigrant dependence among our San Diego sample firms. The strongest predictors of immigrant dependence are firm age (established more recently), entry-level wages (lower), whether U.S.-born persons sometimes apply for the kinds of jobs now held by immigrants in the firm (no), whether the company owner is an immigrant or U.S.-born Latino (yes), and whether the firm reduced its labor force during the 1990s recession (no).

The very large and highly significant negative coefficient for the variable concerning native-born applicants for the types of jobs done by immigrants is of particular interest. This variable by itself explains 26 percent of the variation in percentage of workers in the labor force. As discussed below, native-born workers are virtually absent from the applicant pools of most of our sample firms, at least for entry-level production jobs. This is strong evidence supporting the hypothesis that the demand for immigrant labor has become structurally embedded in the San Diego economy. The wage variable is also highly significant but has a rather small coefficient, indicating that cheapness is not the primary reason for hiring foreign workers, or at least that the wage differential is small. The one public policy variable included in this model - concerning the impact of the U.S. Border Patrol's Operation Gatekeeper on the supply of immigrant labor available to our sample firms) does not reach statistical significance. As discussed below, very few of the San Diego employers interviewed in 1996 had seen a decrease in immigrant job-seekers since Gatekeeper began (which is why this variable is insignificant), but the most immigrant-dependent firms were especially unlikely to have experienced such a contraction of their labor supply (which is why the coefficient is large and negative).

A similar regression model, using the same dependent variable and including as independent variables all of the firm characteristics on which data were collected in both San Diego and Hamamatsu, was estimated for our sample of Hamamatsu firms. This model explains 49 percent of the variance (adjusted R-square = .492) in dependence on immigrant labor in Hamamatsu. One of the most significant predictors is whether the firm has the same difficulty recruiting native-born workers with or without a recession (yes), demonstrating again the extremely limited competitiveness of the most immigrant-dependent firms in Japanese labor markets.

In summary, our findings suggest that the stratum of most-immigrant-dependent businesses in San Diego consists of those that are "immigrant enclave" or "Mexican co-ethnic" firms that recruit almost exclusively through social networks, tapping the lowest stratum of the immigrant population (impoverished, recently arrived, undocumented, economically motivated migrants with limited education and English proficiency). Despite offering less attractive wages and fringe benefits, they have more stable work forces (some possible explanations are offered below). They offer employment opportunities to the least skilled, least educated, least English-proficient members of the immigrant population.

In both San Diego and Hamamatsu, the most immigrant-dependent firms do not attract native-born applicants for lower-level production jobs (even if there are some older, long-time native employees still doing such jobs), and they are very pessimistic about their future ability to compete for native-born workers, given the numerous disincentives (unattractive wages, few fringe benefits, poor working conditions, and so forth) which they can do little to change. During recessionary periods, these firms may find it necessary to cut wages, overtime, and benefits, but they are unlikely to shed immigrant labor, because they depend so heavily upon this labor source and have no realistic alternatives to it. Thus, the most immigrant-dependent firms are the least responsive to the business cycle in terms of immigrant worker hiring and lay-offs.

These results underline the crucial role played by immigrant labor in the establishment, survival, and growth of a significant stratum of businesses in San Diego and Hamamatsu. Such firms are the clearest manifestation of a demand for immigrant labor that is structurally embedded.

Creating Structural Embeddedness
The continuing need of even knowledge-intensive businesses for substantial numbers of relatively low-skilled production and maintenance workers helps to explain the pervasive demand for immigrant workers in large metropolitan areas like San Diego. But other factors are clearly at work, too; most importantly: (1) the nature of the jobs that are typically offered to entry-level workers in immigrant-using firms (not just low-paying jobs, but jobs that involve boring, often dirty and even hazardous tasks, with no career ladder); (2) the proliferation of social networks that keep delivering eager new immigrant job-seekers to the employer's doorstep with little or no effort on his part; and (3) the way in which employers perceive and evaluate immigrant workers in comparison with potential native-born substitutes. Data from our surveys of employers and immigrant workers in San Diego and Hamamatsu demonstrate the importance of all of these factors in creating a structurally embedded demand for foreign labor.

Making new hires through immigrants' kinship and friendship networks has become the most common labor recruitment mechanism for immigrant-using firms in the San Diego area. Seventy percent of the employers whom we interviewed reported that their most important way of recruiting production workers was through referrals from current employees. This coincides exactly with what the immigrant workers in our employee sample told us: 70 percent had found their current job through a relative or friend employed at the same firm where they now work (only 16 percent had applied for their job directly, without benefit of social network contacts). The proportion of immigrant workers hired through network referrals ranged from 55 percent in the hotel and motel industry to 80 percent in restaurants, landscape and building maintenance, and low-tech manufacturing firms. In Hamamatsu, 41 percent of foreign workers had found their most recent job through a relative or friend.

From the employer's perspective, immigrants' social networks provide an ideal mechanism for labor recruitment. No costly advertising is required; no employment agency fees need be paid. Job vacancies can be filled almost immediately; in most cases, immigrants already working in the firm know that a vacancy is about to occur even before the employer does. High-quality workers are virtually guaranteed, even without screening of job applicants, since the immigrant's social network vouches for his reliability, productivity, and good character. As Waldinger has also observed in Los Angeles and New York City, the existence of such a convenient, no-cost, highly reliable labor recruitment mechanism in itself serves to perpetuate and, over time, expand the immigrant presence in many firms, while reducing the representation of native-born residents in their applicant pools (Waldinger, 1996, 1998).

However, it is also clear from our interviews that many employers who use immigrant labor do so not just as a matter of convenience: employers develop a distinct preference for it -- a preference that strongly influences the level of effort that they make to recruit outside of immigrant networks. Employers like the fact that most immigrants are highly reliable and punctual. They appreciate the immigrants' flexibility - their willingness to work overtime, weekends, or night shifts if needed (the illegals in our San Diego immigrant worker sample were especially likely than legal immigrants to be working the night shift). In Hamamatsu, 77 percent of the employers whom we interviewed said that foreigners were more willing to do overtime than the native-born Japanese. Employers also value the immigrant's strong work ethic: 54 percent of our San Diego employers told us that, in general, immigrants were more hard-working than native-born employees, which is up from 28 percent who expressed this belief in our 1983 survey of San Diego employers.

Labor market segmentation is also a key factor promoting utilization of foreign workers. Immigrants and native-born workers tend to be channeled (or channel themselves) into quite distinct segments of the California labor market. Native-born workers seldom compete directly with immigrants for low-skilled production jobs in firms and industries that have come to be dominated by immigrant workers. If there is job competition, it typically occurs between recently-arrived and long-settled immigrant residents. As shown in Figure 4, immigrants have all but replaced U.S.-born workers in nine of the most common types of low-paying jobs in California, statewide. The representation of immigrants in all of these job categories has increased dramatically during the last 16 years.

This phenomenon is very evident in San Diego, where nearly half (49 percent) of the employers whom we interviewed reported that U.S.-born persons never apply for the same types of jobs held in their firm by immigrants, or apply only rarely for these jobs. In just one-third of our sample firms do the immigrant and native-born job applicant pools overlap substantially. The segmentation of San Diego's labor markets into immigrant-dominated and native worker-dominated sectors is partly a product of social network-mediated labor recruitment, which effectively removes the jobs filled through referrals by other immigrants from the "open" labor market. But such segmentation also reflects the prevailing attitudes of native-born workers - especially the young -- toward certain types of jobs and industries. Employers who get few, if any, native-born applicants may not be trying very hard to recruit them; but it is also true that native-born workers typically avoid jobs that have become stereotyped as "immigrant jobs," particularly in firms where immigrants already comprise a high percentage of the total work force. In those firms, the culture of the workplace - including the predominant use of Spanish among production workers and between workers and their immediate supervisors - constitutes an additional disincentive for native-born workers to seek jobs there.

A final factor that contributes to the demand for immigrant labor in San Diego and Hamamatsu is the growth of immigrant entrepreneurship. Increasingly, immigrants in both cities are not just employees but owners of the businesses where they work. Among the firms in our 1996 San Diego sample, more than one-third were immigrant-owned, with Mexico-born entrepreneurs being the most numerous. A recent survey of 102 Mexican-owned businesses in Los Angeles (Guarnizo, 1996) found that 81 percent of their employees were Mexican immigrants and another 17 percent were Latino immigrants of other nationalities; only 2 percent of the workers were non-Latino immigrants and U.S. natives).

This closely resembles the ethnic composition of the work force in the Mexican-owned businesses represented in our 1996 San Diego sample. They are more dependent on foreign-born labor (especially undocumented Mexicans) than Anglo-owned businesses in San Diego; they recruit almost exclusively through referrals from their immigrant employees; and they pay lower wages than U.S.-born employers (but not sub-minimum wages). Surprisingly, labor turnover rates in immigrant-owned businesses are lower than in U.S. native-owned firms in San Diego, which suggests the importance of a workplace culture in which ethnic solidarity, ease of communication (in Spanish), and family ties provide incentives for stability.

Similarly, in Hamamatsu and other Japanese cities where Latin American Nikkeijin have clustered since the late 1980s, ethnic-enclave businesses - restaurants, supermarkets, travel agencies, banks, and other service-providers - catering to immigrants are making it easier and more attractive for them to settle permanently in these places, and they are preferentially hiring immigrant workers (Tsuda, forthcoming).

The Impact of Business Cycles

As immigrants become a preferred labor source, we would expect employers to do more to retain them, even in a recessionary economy. What we found, in both San Diego and Hamamatsu, is that most employers who had been using foreign-born labor before the recession of the 1990s tried to retain as many of those workers as possible, despite the poor business conditions. In San Diego, 63 percent of our sample firms had laid off not a single foreign-born worker during the 1990s recession. In Hamamatsu, only 28 percent of our sample firms had laid off some foreign workers during the recession, or had declined to renew the expiring work contracts of foreign employees, or had returned foreign workers to the brokers from whom they had obtained them.

While there was some shedding of foreign labor as a consequence of the 1990s recession, employers in both San Diego and Hamamatsu continued hiring foreign workers during the recession. Seventy-two percent of our San Diego sample firms reported that there had been no net change in the proportion of immigrants in their work force from 1991-1996; 36 percent of them had put new immigrant workers on their payrolls during this period. In Hamamatsu, too, we found an impressive degree of stability in the demand for foreign labor, despite the severity and duration of the Japanese recession. In nearly half (49 percent) of the Hamamatsu firms, the proportion of foreigners in the work force had remained stable or even increased during the five years preceding our survey. Forty-one percent of the Hamamatsu employers had put new foreign workers on their payrolls during the recession. Taken together, the data from San Diego and Hamamatsu suggest that the business cycle has little effect on the demand for immigrant labor in these cities.

Moreover, the recession apparently did little to expand the supply of native-born labor for the firms we studied in San Diego and Hamamatsu. In San Diego, only one out of five of our sample firms reported that it had been easier to find U.S.-born workers to fill low-level jobs in their company during the recession. A much higher proportion (63 percent) of the Hamamatsu employers reported that they had found it relatively easier to recruit native-born workers during the recession. Even so, over one-third (34.7 percent) of the Hamamatsu companies reported that it was still difficult to find native workers for production jobs during the recession, especially young Japanese.

When asked what they would have to do to attract more native-born workers, nearly half (49 percent) of our San Diego employers replied, "raise wages!"; but only 20 percent of the Hamamatsu employers believed that raising wages would increase their access to native-born workers. A plurality (26 percent) of the Hamamatsu employers believed that the most potent recruitment incentive for native Japanese would be improved working conditions. But most of these firms could not improve their working conditions significantly without prohibitively large capital investments, and even so, many undesirable (so-called "3-K") jobs would remain in these companies.

Our findings from San Diego (and to a lesser extent, from Hamamatsu) are consistent with the hypothesis that the use of immigrant labor has become largely independent of the business cycle. Neither is it accurate to portray immigrant labor simply as a counter-cyclical "cushion" that employers can expand or contract almost at will, to buffer their firms against recessions or seasonal fluctuations in product demand. The impressive stability of the immigrant work force in that San Diego firms that we surveyed both in 1983 (immediately following the 1981-82 recession) and in 1996 (following the recession of the early 1990s) belies this notion. The typical employer of immigrants in San Diego clearly does not treat them as highly disposable workers, relative to native-born employees. Employers certainly can and do use immigrants as a flexible labor source, but in San Diego this normally takes the form of using immigrant workers on night shifts and for overtime work to respond to peaks in demand. Lay-offs of non-temporary, full-time immigrant employees during slack periods are no more prevalent than among native-born workers. The same pattern seems to obtain in Hamamatsu firms, especially those that hire their foreign workers directly rather than obtaining them through professional labor brokers.

Impacts of Public Policy Changes

What impacts have changes in the legal and public policy environment had on the utilization of foreign labor? In the case of San Diego, the most important change in the two years preceding our 1996 surveys of employers and immigrant workers had been the implementation of Operation Gatekeeper, a $300 million effort by the Border Patrol to fortify the San Diego County segment of the U.S.-Mexico border through manpower, hardware, and technology enhancements. More than 2,000 Border Patrol agents have been stationed in the San Diego sector to carry out Gatekeeper. By sharply increasing the probability of apprehension for illegal border crossers along the most fortified, westernmost segment of the U.S.-Mexico border (where 45 percent of illegal immigrant apprehensions all along the border occurred in the 1993 Fiscal Year), Gatekeeper has sought to force crossings into the much less hospitable, mountainous terrain in the eastern portion of San Diego County and into the desert of neighboring Imperial County. In theory, as the risk of apprehension increased, professional people-smugglers ("coyotes") raised their fees, and their clients were exposed to life-threatening physical hazards (including death from dehydration or hypothermia), prospective illegal migrants would be deterred from leaving their places of origin or, upon reaching the border and failing repeatedly to gain entry, would become discouraged and return home.

We were interested in how this intensified border enforcement effort had affected the actual supply of immigrant labor available to San Diego employers. We found that only 8 percent of the employers whom we interviewed had noticed any decrease in the number of immigrant workers seeking jobs at their company during the 15 months preceding our interviews -- a period that coincided with the implementation of Operation Gatekeeper. More than two-thirds (69 percent) of the employers had seen no change in the supply of immigrant labor, and 23 percent had noticed an increase in immigrant job applicants since "Gatekeeper" began.

Our observations and interviewing in street-corner labor markets where recently arrived undocumented migrants from Mexico often congregate, conducted from January 1996 through July 1997, revealed no shortage of job-seekers. Indeed, the most common complaint voiced by migrants seeking work as day-laborers is the persistent oversupply of Mexican labor - i.e., too much competition from other migrants, who continue to make their way to San Diego's labor markets despite the formidable new obstacles at the border.

While Gatekeeper has certainly raised the probability of apprehension at certain points along the border, illegal migrants are still crossing successfully in other areas. Among our sample of illegal migrant workers, those who entered the San Diego labor market before January 1995 made an average of 1.42 crossing attempts before gaining entry; those who entered between January and December 1995 made 1.18 attempts; and those who came to San Diego from January through June of 1996 made 1.63 attempts (mean statistics). Seventy-eight percent of the undocumented workers entering the San Diego labor market before January 1995 made it into the United States on their first try; 87 percent of those coming in 1995 (the first full year of Gatekeeper operations) made a single crossing attempt; and 81 percent of those entering in the first half of 1996 attempted entry only once. These findings, together with Border Patrol apprehension statistics, clearly indicate that as enforcement has been stepped up along the westernmost portion of San Diego County's border with Mexico, prospective illegal migrants have learned quickly to avoid the most heavily fortified segments of the border altogether and cross elsewhere (see Cornelius, 1997: 385-95; Gross, 1997).

We hypothesized that the highly publicized implementation of Gatekeeper could have a chilling effect on some employers and make the search for a job more difficult for undocumented workers entering the San Diego labor market, but our survey data do not support this expectation. In fact, job-seeking time in San Diego among undocumented migrants remained constant or declined during Gatekeeper's implementation period. The median number of job search days for undocumented migrants, excluding street-corner day laborers, remained constant at 7; the mean job search declined from 35.1 days for those arriving before 1995 to 26.8 days during 1995 and to 23.7 days in the first half of 1996. These figures refer to the amount of time needed to secure a steady job; they do not include street-corner labor market participants. Clearly, whatever chilling effect Gatekeeper may have had on the hiring of undocumented workers during its first 21 months of activity, it was more than offset by the increase in demand for labor as California's recovery from the early 1990s recession gathered momentum.

Gatekeeper has made would-be illegal entrants more dependent on professional people-smugglers ("coyotes"), while bidding up the fees they can charge. Among illegal immigrants in San Diego whom we interviewed, the proportion using the services of coyotes rose steadily as Gatekeeper was implemented. Forty-two percent of those who entered the United States most recently before January 1995 used coyotes; 44 percent of those who entered during 1995 did so; and 52 percent of illegals entering during the first six months of 1996 found it necessary to use coyotes. Fees paid to coyotes for assistance in evading the Border Patrol and transportation to a safe house or work site rose sharply, from a mean of $143 in the pre-Gatekeeper period to $490 in 1995.

Concentrated border enforcement operations like Gatekeeper (San Diego, October 1994-present), Operation Hold-the-Line (El Paso, Texas, September 1993-present), and Operation Rio Grande (McAllen, Texas, September 1997-present) appear to be failing as deterrents to illegal immigration, in the sense of persuading would-be migrants to stay home, causing those who fail repeatedly to gain entry clandestinely to give up, and keeping those who succeed from getting jobs in the United States. However, such operations may have the unintended consequence of prolonging the stays of undocumented Mexican migrants and raising the probability that they will settle permanently in the United States.

Every major U.S. government effort to tighten border enforcement or restrict access to the labor market for undocumented immigrants in the past fifteen years has aroused the "psychology of the closing door" and at least temporarily increased the costs and risks of illegal entry. The effect of such measures is clearly reflected in the behavior of undocumented Mexican migrants to the United States following enactment of the 1986 Immigration Control and Reform Act (IRCA), which included penalties against employers who "knowingly" hired unauthorized migrants. Among undocumented migrants from three Mexican sending communities studied by the author in 1988-1989, those who had gone to the U.S. most recently in the pre-IRCA period spent 16.7 months there, on average; unauthorized migrants who had gone to the U.S. since IRCA's enactment were spending 20.6 months there before returning to their home community (Cornelius, 1990: 241). Among the undocumented Mexican workers whom we interviewed in San Diego during the first half of 1996, 77 percent of those who had entered the San Diego labor market during the January-December 1995 period (Gatekeeper's first full year of operation) had already stayed in the United States longer than they had expected upon arrival.

Another key change in the public policy environment in the period leading up to our 1996 surveys in San Diego was the movement to deny basic social services to undocumented immigrants. California voters approved Proposition 187 some 14 months before our field interviewing began in San Diego County. While implementation of most of its provisions was blocked almost immediately by federal court rulings, the passage of the state ballot initiative by a 59-to-41 percent majority sent a strong signal to immigrants - legals and illegals alike - that the tolerance of the aging, Anglo residents who constitute the bulk of the participating California electorate for the current wave of predominantly Latino immigration was ending, and punitive measures were likely to be taken against them. Nevertheless, as its advocates readily acknowledged, Proposition 187 contained virtually nothing that would make undocumented immigrants less employable in California; the initiative was aimed at denying them social services. Hence, in our fieldwork we expected to find no significant labor-market impacts of Proposition 187.

The results of our survey of immigrant workers in San Diego County were largely consistent with this expectation. Only one out of five regularly employed workers had experienced any type of problem that they could attribute to Proposition 187; a considerably higher proportion (41 percent) of the street-corner workers whom we interviewed reported problems resulting from "la 187." Except among the street-corner workers, the Prop. 187-related problem experienced most frequently by our immigrant worker interviewees was increased discrimination and hostility toward them, of a generalized character (i.e., not limited to hiring discrimination).

Only 1.5 percent of the regularly employed immigrants reported that they had found it more difficult to get work because of Prop. 187; but 16 percent of the street-corner migrant laborers had found it harder. If the street-corner workers' perceptions are accurate, it is reasonable to hypothesize that some casual employers of undocumented workers (typically individual homeowners and small subcontractors in construction, painting, and landscaping) were frightened off - at least temporarily - by the media hype surrounding Prop. 187 and, more importantly, by their own misunderstanding of the measure. Nevertheless, for the vast majority of undocumented immigrants - even street-corner workers - it has been business-as-usual in San Diego's labor markets.

The Future Demand for Immigrant Labor
The U.S. economy and society are being transformed in ways that inevitably will increase the demand for immigrant labor. Growth in low-level service and retail employment has been particularly explosive in recent decades: hotels and motels; restaurants and fast-food outlets; car-washing, dry-cleaning, convalescent care, and domestic service. Many of the same industries (especially care for the elderly) are also poised for rapid growth in Japan.

Some of these industries (like fast food) are slowly becoming more automated; but many others have already reached a more-or-less permanent plateau of technological modernization and are unable to further reduce their labor requirements in this way. A simple example: Virtually all restaurants have installed automatic dishwashers, but those machines must still be loaded, unloaded, and maintained by someone - almost invariably an immigrant, in San Diego and other major California cities.

In some sectors of the U.S. economy, the utilization of foreign-born labor has been driven partly - or at least facilitated - by the rapidly expanding supply of immigrant workers. In California, this phenomenon of supply-driven employment of immigrants can be observed most easily in the domestic service and agricultural sectors. Since the 1970s there has been an exponential increase in the demand for private household service workers who provide housecleaning, gardening, child-care, and elder-care services. In a society where both parents now typically work outside the home, such workers are considered a necessity, not a luxury, even by middle-class families -- which is why this common source of employment for immigrants is virtually recession-proof. For example, census data show that twice as many gardeners and private household servants were employed in the Los Angeles region in 1990 than in 1980 (Waldinger, 1997: 457), a trend that has been documented in other major metropolitan areas (on Washington, D.C., see Repak, 1995).

The same rapid growth in employment has been observed in California's agricultural sector, where production of labor-intensive fresh vegetables, fruits, and nuts has been booming in the last two decades. According to agricultural experts, two factors are primarily responsible: changing consumer tastes (i.e., a preference for more fresh vegetables to maintain a healthier life-style), and a virtually inexhaustible reservoir of Mexican hand labor. Thus, employer demand and the Mexican immigrant labor supply have grown in tandem. Unquestionably, California's labor-intensive agricultural industry expanded more rapidly in the 1980s and 1990s than would have been possible in the absence of a plentiful supply - indeed, oversupply -- of low-cost immigrant workers (see Martin, 1996).

Consistent with the predictions of labor market segmentation theory, the vast majority of Mexican immigrants being added to the labor force in California in the 1970s, 1980s and 1990s appear to have been complements to - not substitutes for - native-born workers, including members of ethnic and racial minorities (Marcelli, 1997). The growth of the U.S. economy in the past quarter-century has created a large number of low-skill, so-called "bad jobs" that do not pay enough to sustain a middle-class standard of living. And because most of these jobs involve arduous manual labor, or at least menial and highly repetitive tasks, they are shunned by most young native-born workers entering the labor force. Most of those jobs have not disappeared, despite all the technological innovation, corporate downsizing, and relocation of production to East Asia, Mexico, and the Caribbean that has occurred in the last ten years or so. And there is no obvious reason why we should expect them to disappear in the foreseeable future.

Significantly, most of the firms in our San Diego and Hamamatsu surveys had already experimented with one or more alternatives to employing foreign workers. Sixty-seven percent of the firms studied in San Diego and 83 percent of the Hamamatsu firms had done so. Rationalization of production to reduce labor requirements was the most commonly pursued option, followed by mechanization, subcontracting, hiring more part-timers, hiring more native-born women (in the Hamamatsu firms), hiring more native-born elderly workers, and moving some production abroad.

How had these experiments worked? In San Diego, only 11 percent of the employers whom we interviewed reported that doing any of these things had actually helped them to reduce the proportion of foreign-born workers in their labor force. The vast majority of these firms anticipate the same or even a higher level of reliance on immigrant labor in the future; only 7 percent expect to reduce the proportion of immigrants in their work force in the foreseeable future. Hamamatsu employers evaluated the alternatives much more positively; 78 percent of them reported that adopting one or more of these options had been "effective" to some degree. However, as their responses to other questions show, these measures have not been effective enough to enable most of these companies to actually reduce their reliance on foreign labor. And when we asked Hamamatsu employers whether they expected to have a lower or higher proportion of foreigners in their labor forces in the foreseeable future, 39 percent foresaw higher reliance on foreign labor and 49 percent anticipated no change; only 12 expected to be able to reduce their current level of reliance on foreign labor. In short, like their San Diego counterparts, most of these Japanese companies seem solidly locked into the transnational labor market.