Statistics on U.S. Immigration: An Assessment of Data Needs for Future Research (1996)

This chapter summarizes the workshop session that provided details on the characteristics and magnitude of trends in U.S. immigration.

The workshop discussion of demographic trends in immigration began with trends in the average number of immigrants (Figure 1). The annual numbers of new entrants reached their highest volume during the first two decades of this century. As a result of the passage of the National Origins Quota Act in 1924, the Great Depression of the 1930s, and an unfavorable immigration climate during World War II, immigration dropped to one-tenth of the record-setting levels during the next 25 years. Specifically, the number of entrants decreased from over 700,000 per year during the first 20 years of the century to less than 70,000 per year from 1925 to 1945 (Bean, 1992:4). After this lull and continuing for the 50 years since 1945, legal immigration has been moving steadily upward, reaching levels by the late 1980s and early 1990s similar to the all-time highs set in the early part of this century. And if the legalizations of previously illegal aliens resulting from the 1986 Immigration Reform and Control Act (IRCA) are included in the totals, the recent levels exceed all previous highs.

Figure 1

Average number of immigrants and percentage foreign-born of U.S.  population, by decade, 1910–1990.

The major features of post-World War II immigration include: (1) increasing numbers of illegal aliens entering the United States, (2) increasing numbers of refugees and those seeking asylum, (3) substantially increasing numbers of immigrants from Third World countries (mostly from Asia and Latin America) and concomitant declines in the fraction of immigrants of European origin, (4) increasing proportions of immigrants admitted on the basis of occupational and skills criteria rather than national origin or family connection criteria, and (5) enormously increasing numbers of people admitted for short periods of time on so-called nonimmigrant visas (students, tourists, and business travelers). This last phenomenon is not technically considered to be immigration, but it nonetheless carries considerable consequences for immigration. It often is a prelude to subsequent entry, and its facilitation by the Immigration Act of 1990 is viewed by many observers as a compromise in the face of proposals to allow even higher levels of business-related immigration than were actually written into the legislation.

In the mid-1970s, growth in real wages began to level off, unemployment rose as the country experienced a recession, and calls for immigration reform began to emerge (Bean et al., 1987). Frequently these consisted of restrictionist outcries against the new immigration, often stated in the form of unsubstantiated claims about the pernicious nature of immigration and its harmful effects on the country. During the 1980s, a body of social science research emerged that found little basis for the claims that immigration was generating substantial negative

economic and social effects (Bean et al., 1988). In fact, the research tended to show that immigrants were assimilating socially and economically within a reasonable period of time, were not exerting very large labor market effects on the wages and unemployment of natives, and were not consuming more in the way of public benefits than they were paying in taxes. To be sure, because the composition of immigrant origins was changing, questions were being raised about whether the skill levels of immigrants were actually declining, both within and across countries of origin (Borjas, 1990). The general conclusion that emerged from the research on this period was that immigration did not appear to be generating much in the way of large effects, either positive or negative (for a recent discussion, see Espenshade, 1994).

Almost all of the research cited above, however, has been based on data collected during the 1970s or, in one or two instances, during the early 1980s. Only recently have studies appeared that analyze data from the mid-and late 1980s. (Edmonston and Passel, 1994, contains a collection of empirical analyses based on more recent data.) The question that remains unanswered is whether similar effects would be observed during periods of greater immigration and slower economic growth. Given that these are precisely the conditions that have emerged in recent years, the issue of the country's capacity to absorb immigration continues to be a significant policy question. In the past, the implications of immigration have most frequently been addressed in terms of population growth and much less frequently in terms of economic growth (Borjas and Tienda, 1987).

The slowdown in the growth rate of the civilian labor force, in combination with the continuing increase in the growth rate of immigration, poses challenges for both analysts and policy makers (see Table 1). These challenges derive from circumstances that are currently even more severe than the above-mentioned rates of change for the decade of the 1980s. Since 1989, virtually no growth has occurred in the number of employed people in the United States. Over roughly the period 1989 to 1992, total legal immigration increased (over the previous three years) by about 140 percent, although non-IRCA-related immigration has increased by only about 7 percent. Whereas new research based on more recent and perhaps better data for local labor markets may continue to show that immigration exerts relatively negligible labor market effects, under conditions of slow economic growth the public may well conclude that employment growth has not created sufficient jobs for both natives and new immigrants. That may well be the case even if new evidence indicates that proportionately more jobs are being created with than without immigration.

These circumstances raise issues for the perceived legitimacy of the social contract. Specifically, what kinds of people are perceived as having access to what appears to be an increasingly scarce supply of jobs? This issue is particularly acute when illegal aliens are one source of labor seeking access to jobs. Many observers have argued that an integral part of the country's social contract concerning immigration consists of an implicit agreement that good-faith efforts

to control illegal aliens are the price paid for the continuation and growth of a moderately expansionist legal immigration policy (Bean and Fix, 1992; Schuck, 1990). The current social contract may have been rendered even more fragile by the apparent failure of the 1986 legislation to curb at all, or to slow more than briefly, the flow of illegal aliens to the United States (Bean et al., 1990; Espenshade, 1992). Another exacerbating circumstance is that consistent evidence continues to mount that immigration exerts its harshest labor market effects on other immigrants (Borjas, 1990). In what is currently an exceedingly difficult labor market, previously arrived immigrants themselves may come to constitute yet another source of pressure to reform immigration policy.

Some workshop participants argued that the challenge that arises from what are likely to be ever more insistent calls for policy reform is to chart a course that avoids policies based primarily on narrow attitudes of either an anti-immigrant or a pro-immigrant nature. Rather, the task is to develop an immigration policy that gives recognition not only to the realities of a difficult domestic labor market, but also to the fact that the immigration policies of developed countries increasingly have environmental, developmental, and foreign policy implications. Stated differently, participants raised questions about the need for U.S. immigration policies to develop in a context that also considers the country's trade, development, and foreign policies.

The changing circumstances in which U.S. immigration policy is made—and a broader context for the need for data—have two major implications for the collection of immigration-related data. First, much of what is currently known about labor market behavior and the effects of immigrants may turn out to be period-dependent. Period effects may prove as important, or more important, than other factors in explaining patterns of socioeconomic incorporation and labor market effects of immigrants. Although the collection of longitudinal data would be useful in providing a basis for estimating dynamic models of immigrant behavior, hypotheses of period dependence need to be built into such models. Second, data collection efforts in the United States need to be coordinated with those in important immigrant-sending countries, especially Mexico. Almost all

current research on U.S-Mexican migration, regardless of whether it uses U.S. or Mexican data, suffers from selectivity problems, because a portion of the relevant study population is residing in the other country at the time of the data collection. A coordinated data collection effort, with longitudinal follow-up of respondents, that occurs simultaneously in both countries would help to solve this problem.

Three recent trends in immigration influence the requirements for immigration data. First, during a period of large-scale international migration, the only measures of the number of illegal aliens are indirect. Although estimates of the foreign-born population can be of some help, they cannot facilitate specific estimates of illegal aliens, especially for annual in-and out-flows. Improved data are required in order to have adequate measures of this population.

Second, data about immigrants are incomplete at best. The decennial census showed more than 20 million foreign-born people in 1990 and could have missed about 1 to 2 million. The 1983, 1986, 1988, and 1989 Current Population Surveys collected data on country of birth for respondents and their parents, which are useful for tracking the growth in the number of foreign-born people; these data have been consistent with decennial census estimates. About 8.7 million foreign-horn people in 1990 said that they entered the United States after 1979; this number is broadly consistent with other estimates, but there are some unknowns. For example, we do not know the proportion of Special Agricultural Workers who received legal status in 1986 and stayed in the United States until 1990. We also lack data sources on the potential impact of this program on immigration. We do not know about emigration from current data on the foreign-born population and their children.

Third, illegal immigration is likely to persist in the future. Martin and his associates (Martin and Taylor, 1991) have suggested that illegal flows will increase in the future because of the continued need for farmworkers and other low-wage workers in the United States. ² Thus, the foreign-born population can be expected to increase, and its relative proportion in the U.S. population will also increase.

Current Population Survey data could be improved in three ways. First, it is critical to obtain information on nativity, including country of birth and date of immigration for respondents and their parents. ³ Such nativity questions would

offer the requisite information for studying foreign-born people and their children—the two immigrant generations of the population, called the foreign-stock population in demography. Second, increased sample sizes are required for analysis of nationality groups and immigrants classified by labor force characteristics. Current Population Survey data can often be pooled to increase sample sizes, but even larger sample sizes would be required to provide sufficient cases for analyzing generational patterns for specific nationality groups. Because of the high cost of expanding the Current Population Survey on a regular basis, workshop participants suggested the possibility of an enhanced sample size for the Current Population Survey on a periodic basis, perhaps once every three to five years. Third, race and ethnicity information is needed on the Current Population Survey questionnaire and could be collected at minimal cost. The 1994 redesign of the Current Population Survey includes more detailed self-reported ethnic identification. Analysis must rely on country-of-origin data to serve as a proxy indicator for ethnic identity. No specific ethnic identity is available for native-horn people on the Current Population Survey.

Since the imposition of restrictive entry quotas in the early 1920s, U.S. immigration issues have generally been of little concern to economists. Immigration quotas and later the effects of depression and World War II produced relatively low levels of immigration, compared with the levels of the late nineteenth and early twentieth centuries. During the 1950s, immigration again began to rise toward quota ceilings, but fertility levels were high and immigration was a relatively low proportion of overall population growth.

From 1920 to 1970, immigration contributed a small proportion of new labor force entrants. With lower numbers of immigrants, the foreign-born population aged and period mortality more than offset net immigration, with the consequence that the foreign-born population declined during the 50-year period. Following World War II, fertility levels dramatically increased and net additions to the U.S. population came principally from births, rather than from immigration. Women became important contributors to labor force participation during this period. What attention was directed at international migration issues during this half-century came mainly from economic historians, who focused on the earlier period of unrestricted flows, and from those interested in the brain drain—the flow of high-level professionals from poor to rich countries.

Although relatively little work has been done on the impact of immigrants, illegal aliens, and temporary farmworkers in the United States until the last 15 years, there is now much more interest in immigration among economists; one important issue is the impact of less-skilled immigrants on the U.S. economy. This interest has been kindled by the highly visible debates on U.S. immigration policy, focused in large part on the legislation enacted in the 1986 Immigration Reform and Control Act and the Immigration Act of 1990. Renewed attention is perhaps also due to the fact that immigration is again an important source of population growth. Important questions also concern the possible impact of immigrant workers on domestic wage rates and the possible displacement of native-horn workers from their jobs.

The focus of contemporary economic research literature has been toward the effects of unskilled immigration on the U.S. economy (Greenwood and McDowell, 1992). This orientation is due at least partly to a significant downward shift in the skill composition of legal immigrants entering the United States during the 1970s and to the presumably high and continued immigration of illegal aliens with few skills and little education.

Work on the economic contribution of immigration suggests that the aggregate impact of immigration on the U.S. economy is not large, for three reasons. First, immigration is large in absolute terms but small relative to the U.S. labor force. Excluding IRCA legalizations, recent U.S. immigration has consisted of annual flows of about 600,000 to 700,000. These relatively large numbers, however, are small (less than 0.3 percent) additions to the resident population compared with higher proportionate additions in the early 1900s. Second, immigration has offsetting effects. Immigrants may substitute for less-skilled workers, but they may also bring wealth and new, needed skills. Third, the effects of immigration become distributed and smoothed very quickly in the U.S. economy. Labor and capital are highly mobile and the effect of new immigrants, even if they shift labor demands in local areas, rapidly dissipate.

Immigration does evidence some negative effects, but the evidence is usually circumstantial. Some industries experience strong effects of immigration, with large-scale shifts in their labor force to recently arrived immigrants. For example, citrus pickers in Ventura County, California, and construction workers in Houston, Texas, appear to have been seriously harmed by labor market competition from illegal alien workers. Furthermore, offsets to negative consequences exist and often spread through the economy among other workers, employers, consumers, and regions.

Many of the available studies concluding that less-skilled immigrants have small but nonnegligible impacts on the least-skilled resident workers use data from 1970 and refer to immigration that occurred during the 1960s. Not only was immigration somewhat greater during the 1970s than during the 1960s, but it also shifted toward less-skilled migrants. The quantitative shift was due partly to increased admissions of refugees. The qualitative shift was due both to the

admission of refugees and to changes in U.S. immigration law that occurred in the 1960s and became effective late in the decade. The changed magnitude and composition of U.S. immigration require analysis of more recent data. Researchers need to replicate much of the earlier work with the latest data, such as those from the 1990 census.

Several improvements in immigration data are needed for the study of the economic aspects of immigration. First, Immigration and Naturalization Service data are of major interest for economic analysis. Major improvements that would be valuable include the addition of collected data on schooling and language ability, assets at the time of arrival in the United States, and income at the time of visa adjustment.

Second-generation data are required in order to study an important group—the sons and daughters of immigrants. Understanding the nature of assimilation is important in order to understand the adjustment process for different nationality groups, and the second generation is a pivotal social group for studying adjustment. If immigrant quality (as evidenced by lower educational attainment and poorer job skills) has fallen, then there is a need to understand the generational differences. The 1990 census did not ask a generational question; it is absolutely essential to collect generational data on the 2000 census. The Current Population Survey has relatively small sample sizes for most special nationality groups, making even more critical the collection of data on the census. For example, only the Mexican-origin population in the Current Population Survey has observations sufficient for detailed study.⁵ Third-generation data would also be useful, but their compilation would require collecting data on the country of origin of grandparents, which would complicate questionnaire design considerably.

There is considerable discussion among economists about the uses of panel data for immigration studies. It is not sufficient to collect data on a single cohort of immigrants. To collect useful panel data, different cohorts must be followed over time—but there is debate among economists on cohort interpretations. Chiswick (1977, 1978, 1982), in his work with cross-sectional data on white immigrants, finds that the economic status of immigrants improves with duration

of residence. Concluding in his research that unmeasured differences exist between cohorts, Borjas (1983, 1986) argues that economic conditions at the time of entry may distinguish immigrant cohorts and that the quality of immigrants (in terms of education and job skills) may therefore vary. If a longitudinal survey of immigrants were feasible, available economic studies demonstrate that it would be critical to design a survey that periodically selects new immigrants to follow over time, noting immigrant status at entry into the United States and all subsequent changes in status and citizenship.⁶ Such a survey would, of course, increase the cost and complexity over the survey of a single immigrant cohort.

Another important debate in the economic literature concerns the economic adjustment of immigrants (Borjas and Tienda, 1987). Numerous methodological problems have emerged in the course of this debate. First, census groups differ over time. Even though researchers have attempted to compare synthetic age cohorts, it is clear that differential emigration and misclassification confound the analysis. Second, native-born white men are often used as a reference group, further challenging comparisons for trends over time as the reference group itself changes. Third, immigrant cohorts change over time, through mobility and international migration, and these changes add complexity to interpretation. Fourth, research suggests that emigration has a generally negative selection effect (successful immigrants depart the United States more often than other immigrants). Some effects observed in cohort studies using census data may be affected by selection for emigration. To address these problems in the research, immigrants should be followed individually through time, and the selection process should be known. Such a longitudinal study could also reveal more about the remittances and transfer payments that immigrants make.

Finally, two other types of data would improve immigration statistics: (1) a question on precise year of entry into the United States on the decennial census and (2) data on immigrant status at time of entry and at time of survey in order to inform analysis about immigrant changes.

One of the most important variables for studying the economic aspects of immigration is occupation. The study of the experiences of immigrants must include the examination of an immigrant's occupation at entry into the United States, the occupational mobility that he or she experiences over a lifetime of interactions in the U.S. labor market, and the occupations into which his or her children enter. Better data are especially needed on occupational mobility, distribution, and competition. The Immigration and Naturalization Service should compile distributions of immigrants who enter for family reunification and employment reasons, not just data on family members. The analysis of the impact of immigration on U.S. labor markets requires documentation of the occupational

concentration of immigrant groups upon arrival, changes in occupational concentration over time, and the occupational nature of competition for jobs among various immigrant and native groups.

The Immigration Act of 1990 tilts immigration selection toward employment characteristics. Although the United States has received more immigrants in the early 1990s than in previous years, they are a smaller proportion of the labor force and they vary by occupation. Immigrants are now concentrated in certain occupations, especially by countries of origin, and this occupational concentration can be extreme at the metropolitan level. Unfortunately, flow characteristics are noticeably absent from current data. For example, according to decennial census data, most Mexican immigrants work in only a few specific occupations, but annual immigration data collected by the Immigration and Naturalization Service lacks the detail on occupation needed to discern such occupational patterns for important flow information.

Aggregated, national estimates can be misleading indicators on which public opinion and understanding of metropolitan dynamics come to be based. The impact of immigration and illegal flows has varied greatly among metropolitan areas, generating public concerns that are not necessarily common nationwide.

For policy purposes, broad research questions should focus on several issues: defining occupational mobility for native-born people and immigrants in order to understand immigration effects; expanding the number of studies that focus on variations in employment conditions and the impact of immigration in metropolitan areas and regions of the country; conducting case studies in order to build up a stock of empirical effects; and sorting out the short-term and long-term effects of immigration. There may be short-term substitutability in employment but long-term adjustments in the labor market that differ in their effects. Research questions should also address the need to understand the relationship of immigration and economic cycles. There are potential fears of rising and falling employment perceived to be caused by immigration. Policy research must be very clear about the perceived and known cycles that the economy experiences.