Journal Issue

United States of America: Selected Issues

International Monetary Fund
Published Date:
July 2004
  • ShareShare
Show Summary Details

II. Domestic and Global Perspectives of Migration to the United States14

1. Recent economic developments and policy proposals have focused attention on the macroeconomic and other effects of migration to the United States. Recent proposals for significant immigration reform include the Administration’s plan to launch a new temporary worker program that would give legal status to unauthorized immigrants in the United States (Orrenius, 2003; Meissner, 2004). At the same time, some analysts have related the recent slowdown in the growth of the labor force and the increase in outsourcing of services offshore to a decline in immigration from the high levels of the 1990s, owing both to the weaker economy and security concerns after September 11.

2. In addition to output growth, immigration trends may have implications for the long-run sustainability of entitlement programs and the U.S. fiscal position. Higher immigration may at least partly offset the decline in the fertility rate that, together with increases in longevity, is expected to place old age retirement and health care programs under growing financial pressure. Nonetheless, questions still arise whether the net financial impact of immigration on the U.S. government budget is negative or positive. Several studies have concluded that, on average, immigrants and their descendants contribute more in terms of tax revenues than they absorb via higher government outlays. Although the order of magnitude is typically small, the beneficial impact on the federal budget increases with the share of high-skilled immigrants (Storesletten, 2000, and Lee and Miller, 2000).15

3. U.S. immigration patterns also have significant effects on the rest of the world. It is well known that migration can contribute significantly to the economic and social progress of developing countries through a number of channels (Grieco and Hamilton, 2004). While remittances are the most visible link between migrants and their country of origin, the transfer of skills, education and training can also play a large role, despite the risk that source countries can be deprived of their most important talent (“brain drain”). Based on the 2004 Commitment to Development Index, compiled for 21 advanced countries by the Center for Global Development, the United States ranks above average mainly as a result of an open policy toward immigration (Figure 1).

Figure 1.Commitment to Development Index

Source: Center for Global Development.

4. This chapter seeks to assess the contribution of immigrants to the U.S. economy. First, a “direct” contribution is calculated as the value-added produced by foreign-born residents in the United States, amounting to around 10 percent of GDP during 1994–2003. Second, industry-level data is examined to test the extent to which foreign-born workers contribute to U.S. productivity beyond their direct effect on value added. The analysis finds evidence for such indirect contributions in sectors requiring relatively low skill levels.

5. In addition, this chapter reviews the impact of migration to the United States on source countries. This is done first by considering remittances of U.S. immigrants to their country of origin. In addition, this paper calculates a measure of source countries’ Gross Migration-Corrected Product, taking into account income received by all workers born in a country regardless of current residence. This measure illustrates the significant positive impact that migration to the United States has had on developing countries.

A. Immigrants in the U.S. Labor Market

6. Immigrants have contributed strongly to U.S. labor force growth over the past decade.16 In 2003, foreign-born persons represented 11 percent of U.S. population, accounting for around 15 percent of the U.S. labor force. Although immigrants represent a larger share of the workforce in a number of other countries—e.g., Australia and Canada—this proportion increased rapidly in the United States between 1995 and 2002. Indeed, foreign-born workers account for half of total U.S. labor force growth during this period (Figure 2).

Figure 2.Immigrant Share of Total Labor Force

Sources: Current Population Survey (CPS); and Fund staff calculations.

7. There are significant differences between skill, education, and other labor market characteristics of immigrants and native-born workers in the United States:

  • Job characteristics. Immigrants are over-represented in low-skill jobs; that is, the proportion of foreign-born workers in these jobs is much higher than their overall share in the U.S. labor force (Figure 3a). This concentration largely reflects the strong influx of low-skilled workers from Latin America, with immigrants from other countries being over-represented in occupations that require high and medium skills.

  • Skills distribution. Education levels of foreign-born workers tend to concentrate at the two extremes of the skills distribution (Figure 3b). Immigrants are over-represented in the “elementary” and “high education without diploma” levels, as well as among workers with “higher education”. Latin American immigrants represent about half of all workers with elementary schooling only. European and Asian immigrants are more heavily represented at the higher education level.

  • Job mismatch. Foreign-born workers are more likely than domestic workers to hold a lower-skilled job for a given level of education. Around 1.7 percent of highly educated foreign workers (i.e., with BA/BS and higher degrees) work in low-skill occupations, which is more than double the share of native workers.

  • Sectoral distribution. Foreign-born workers are over-represented in the agriculture, construction and mining, and manufacturing sectors, and under-represented in service sectors (Figure 3c). This picture is somewhat skewed, however, by the concentration of Latin American immigrants in sectors with low skill requirements. Excluding immigrants from Latin America, foreign workers are over-represented in the manufacturing and services sectors, and under-represented in agriculture as well as construction and mining.

Figure 3.Labor Market Characteristics of Immigrant Workers

Sources: Current Population Survey; and Fund staff calculations.

8. Immigrants’ relative income levels reflect differences in skill and education. Between 1994 and 2003 the share of labor income received by immigrants has on average been almost equal to their share in the U.S. population. However, the aggregate result masks a large gap between immigrants from Latin America and those from other countries (Figure 3d). As a result of their concentration in medium-and high-skill jobs, immigrants from Asia and Europe (including Australia, Canada, and New Zealand) have higher per capita-incomes than native workers, while immigrants from Latin America receive about one-third less than workers born in the United States.

B. Recent Immigration Trends

9. Family-related migration has accounted for the majority of permanent U.S. immigration since the mid-1990s, but temporary labor-related migration is becoming more important.17 Between 1998 and 2001, around 70 percent of permanent immigration occurred by way of family reunion, compared with only 15 percent on the basis of employment. As shown in Figure 4, the employment-based share of permanent immigration into the United States is lower than in other countries, including Australia, Canada, and New Zealand, whose immigration policies are oriented to a greater degree at meeting skill shortages. However, the bulk of U.S. immigration—roughly two-thirds—is accounted for by temporary immigrants, whose share has more than doubled between 1994 and 2001 (OECD, 2004). Between 1998 and 2001, temporary workers and trainees (including specialty occupations, agricultural workers, professional workers and intercompany transferees) made up about 40 percent of immigrants with time-limited visa status, up from around 30 percent in the mid-1990s.

Figure 4.Permanent and Long-Term Immigration Flows in Selected OECD Countries

Source: Organization for Economic Cooperation and Development, 2004.

10. The rapid increase in temporary immigration has been partly a response to labor supply pressures emerging in the 1990s. In particular, the annual quota for H-1B visas (which are granted to professional and skilled workers for a maximum of six years) was raised from 65,000 in the mid-1990s to 195,000 over 2000–03, and the 7 percent ceiling on the proportion of visas going to nationals of any given country was lifted.18 The increase in visas for seasonal workers (both in the agricultural and other programs) over the second half of the 1990s also points to a response to labor shortages in relatively low-skill occupations.

11. A recent proposal from the U.S. administration aims to increase U.S. reliance on temporary immigrant workers programs (Meissner, 2004).19 This contrasts with other countries, including Australia, Canada, and New Zealand, that have made permanent immigration subject to a point system emphasizing immigrants’ “employability” by placing more weight on criteria such as age, education, skills, and work experience.20

12. The effectiveness of selective immigration policies is difficult to assess. For one, the cost of administering the immigration process rises with the level of detail among the selection criteria (Doudeijns and Dumont, 2003). Moreover, despite the differences in policies, the relative labor force participation of immigrants is higher in the United States than in countries that have based their immigration polices on employment criteria, such as Australia and Canada (Figure 5). In addition, the average educational level of immigrants into the United States, excluding immigrants from Latin America, is comparable to that of Canada and Australia (Antecol and others, 2001).

Figure 5.Ratio of Foreign and Native-Born Participation Rates

Source: Organization for Economic Cooperation and Development, 2004.

13. Population aging and an anticipated shortage of highly-skilled workers are likely to boost the demand for immigrant labor in the future. Although population aging is less severe in the United States than in other industrialized countries (the United States is one of the few industrial countries where the working-age population is not expected to decrease in the next 50 years), the U.S. Department of Commerce has warned that the United States is not immune to labor supply shortages, especially at higher skill levels (USDOC, 1997). At the same time, the 2004 BLS Occupational Handbook predicts that the U.S. economy will generate jobs for workers of all levels of education and training over the 2002–2012 period. Significant job growth is expected to take place among professional occupations in the IT, health, and education sectors, and half of the 20 fastest-growing jobs would require at least a bachelor or associate degree. However, the largest increase in jobs is expected to be in occupations requiring less formal education and training.

C. Immigrants’ Contribution to U.S. GDP

14. In calculating immigrants’ contribution to U.S. GDP, it is convenient to assume that immigration does not affect incomes of U.S.-born residents. A large body of research has investigated the economic impact of immigration by looking at its effect on wages. Reviewing this literature, Friedberg and Hunt (1995) and Hanson and others (2001) note that, despite the popular belief that immigrants have a large adverse impact on wages and employment opportunities of the native-born population, the empirical support for this conclusion is at best inconclusive.21 The lack of a significant impact of immigration on incomes of native residents would suggest that these effects can be largely ignored in measuring immigrants’ contribution to GDP.

15. A first step is to estimate the value-added produced by foreign-born residents in the United States. Such an estimate is obtained by extrapolating from the March Current Population Survey (CPS) the share of total income earned by workers born outside the United States to U.S. GDP.22 This approach indicates that the contribution of immigrants to U.S. GDP has increased steadily from about 10 percent to 13 percent between 1994 and 2003. Indeed, immigrants’ contribution to U.S. income is only slightly below their head count representation, a reflection of the small aggregate difference in skills and average income between immigrants and native workers reported in the previous section.

16. Income-based estimates, however, might not fully capture immigrants’ contribution to GDP. Discrepancies could be caused both by the presence of external effects in the production process and by the possibility that immigrants may be paid less than their marginal product. Two types of external effects can be identified. The first is an intra-industry effect, with immigration contributing to industry TFP growth by affecting the efficiency of human capital in that sector. The second type of externality caused by immigration is an inter-industry effect illustrated by standard Heckscher-Ohlin models of international trade theory. To the extent that they change relative factor endowments, migration flows could affect aggregate productivity by inducing changes in a country’s industrial structure.23

17. The existence of external effects is examined within an industry-growth accounting framework. In particular, the existence of intra-industry external effects is tested by analyzing the relationship between industry-specific productivity, using Solow residuals, and the growth of immigrant labor input for the period 1994–2000. The existence of inter-industry externalities is analyzed by running the same regression at the aggregate level (for the period 1982–2000). The benchmark regression is:

where SR denotes the Solow residual, Δ log differences, and M immigrants’ labor input. The Solow residuals were estimated under standard assumptions, that is, all industries are assumed to have constant returns to scale and all input factors are paid their marginal product, in which case the Solow residuals equal TFP growth.24 As demonstrated in the Appendix, if external effects from immigration were present, the Solow residuals would be affected by the growth in immigrant labor input, and the coefficient α1 from zero.25

18. The results suggest the presence of external effects of immigration particularly in low skill sectors. Coefficients are estimated using both a fixed effects model (where industry-specific effects are captured by industry dummies) and a model in first differences (where they are eliminated). In both models, the coefficient on ΔM is significantly positive for agriculture, and the fixed effects model also yields a positive result for the food, beverage, and tobacco sector and the trade sector (Table 1). Overall, the results suggest that externalities may exist in sectors requiring relatively low skills, which attract a relatively large share of immigrant labor, whereas little evidence of externalities is found in sectors that have attracted relatively high-skilled immigrants. Running the regression at the aggregate level indicates some evidence of external effects for the economy as a whole, with the immigration coefficient significantly different than zero in some of the regressions, depending on the number of controls included (Table 2, previous page).26

Table 1.Cross-Industry Solow Residual Regressions1
Fixed EffectsFirst Differences
Food, Beverage and Tobacco0.870.56
Electronics and Machinery-0.03-0.29
Other services-0.15-0.21

Independent variable is the Solow residual (see text for details).

Robust t-statistics in brackets.Note: *,**,*** indicate significance at the 10 percent, 5 percent, and 1 percent level, respectively.

Independent variable is the Solow residual (see text for details).

Robust t-statistics in brackets.Note: *,**,*** indicate significance at the 10 percent, 5 percent, and 1 percent level, respectively.
Table 2.Aggregate Business Sector Solow Residual Regressions1

Independent variable is the Solow residual (see text for details). Robust t-statistics in brackets.

Note: *,**,*** indicate significance at the 10 percent, 5 percent, and 1 percent level, respectively.

Independent variable is the Solow residual (see text for details). Robust t-statistics in brackets.

Note: *,**,*** indicate significance at the 10 percent, 5 percent, and 1 percent level, respectively.

D. An Immigration-Adjusted Measure of National Income

19. In theory, migration is likely to raise real wages in source countries. Migration would be expected to raise per capita-income for the remaining residents by reducing labor supply and causing wages to be bid up, with the effect depending on how the changes in labor supply affect the source country’s industry mix (see Hanson and others, 2001). However, the effect on overall income would also depend on the impact on human capital in the source country. Migration could have a particularly negative impact on residents’ income if migrants were highly skilled relative to the rest of the population (brain drain). However, such a process could also provide current residents with stronger incentives to accumulate human capital (brain gain).27

20. Data on remittances provide a possibility of measuring a direct monetary benefit for source countries of migration to the United States. Immigrants, particularly from developing countries, are known to send a substantial amount of remittances to their families and relatives. Accounting for these transfers would likely increase a source country’s welfare above the level indicated by its national accounts. For a variety of reasons, however, remittances are extremely difficult to measure. In particular, official figures fail to capture informal transfers, and thus are likely to underestimates the size of remittances. Estimates of net workers remittances and migrants’ capital transfers based on the IMF Balance of Payment Statistics show that remittances per U.S. immigrant are above per capita GDP for several Latin American countries, compared to only around 15 percent of per capita GDP for Mexico (Figure 6, previous page).28 However, the result for Mexico may be an underestimate, given that one in every two Mexican immigrants may be undocumented, according to the March 2002 Current Population Survey as well as census data. A higher share of undocumented immigrants could imply that a larger share of remittances to Mexico may flow through informal channels and thus be more difficult to measure.

Figure 6.Selected Countries: Remittances per Migrant and GDP per Capita

Sources: IMF Balance of Payment Statistics; and Fund staff calculations.

21. A different way of capturing the benefit from migration to the United States consists in adding immigrants’ income to the country-of-origin Gross National Income (GNI). This is equivalent to measuring national income on the basis of country-of-origin rather than residency (Ueda, 2002). In the case of Mexico, for example, a nominal “Gross Migration-Adjusted Product” (GMP) can be calculated by summing the income of Mexican migrants to the United States, derived from the CPS, to Mexico’s GNI adjusted for remittances and differences in cost of living:

where REM stands for remittances, Ex for the US/Mexican Peso exchange rate, and PPPEx for the purchasing power parity-adjusted exchange rate. Nominal GMP is converted into real U.S. dollars using the U.S. GDP deflator.

22. GMP calculations for a range of countries show that immigration to the United States has provided significant benefits to a large set of developing countries over the last decade.29 Mexico’s real GMP has been an average 11 percent higher than real GNI between 1994 and 2002 (Figure 7). Real GMP grew at an annual 4¼ percent on average over this period, compared to average GNI growth of 2½ percent. The results also show that citizens of Latin American countries have benefited the most from migration to the United States as the GMP/GNI ratio is quite large for many of these countries, especially from Central America.30 For other geographical regions, similarly large differences are found only for Philippines and Vietnam. By contrast, migration-related benefits for India and China appear almost negligible, despite the large number of U.S. immigrants originating from these countries.

Figure 7.GMP/GNI Per Capita

Sources: CPS; and Fund staff calculations.

E. Conclusion

23. Immigration flows appear to have contributed greatly to the growth of the U.S. labor force, even if only a small share of legal permanent entry is labor-related. Data on education, skills, industry concentration and income reveal a highly polarized composition of migration flows, with migration from Latin America concentrated in low education and low skill occupations and migrants from elsewhere tending to have relatively high levels of education.

24. Immigrants have accounted for about one-tenth of U.S. income growth in recent years. Their contribution to U.S. growth could possibly be even higher, owing to the presence of positive spillovers especially in sectors requiring low skills that have a relatively high presence of foreign-born workers.

25. Both remittances data and an immigration-adjusted measure of source country-income indicate that migration to the United States has had a significantly positive effect for a range of developing countries. Despite the difficulties in capturing transfers from immigrants to their country of origin, remittances per immigrant in the United States are above GDP per capita for several Latin American countries. Adding back income earned by foreign workers in the United States to their source country’s gross national income confirms that citizens of a large set of developing countries have benefited from migration opportunities to the United States in recent years.

APPENDIX I The Standard Solow Residual

The production function can be represented as:

where Y is value added, A denotes total factor productivity (TFP), K represents capital services, and L represents labor input. Taking time derivatives, this can be written as

where a tilde denotes the proportional growth rate:

Assuming that the production function is based on a constant-returns-to-scale (CRS) technology and that factors of production receive the marginal product of labor, the value-added Y is entirely paid to labor under wages and salaries, and to capital as return on investments. Hence, the labor share of value added is

and the capital share is

Denoting with (1- α) the labor share of income, the Solow residual (SR) coincides with TFP growth:

One digression from the standard assumptions is that there are externalities from immigrants’ labor, M. This case can be represented through a production function like the following:

Taking time derivatives, this can be written as

Assuming that both capital and labor receive their marginal products, the SR can be expressed as TFP growth plus a factor proportional to immigrants’ labor growth:

Assuming that η—the elasticity of the external effect—is constant, the SR is affected by growth in immigrants’ labor inputs as well as TFP growth.

Prepared by Roberto Cardarelli (WHD) and Kenichi Ueda (RES).

Lee and Miller (2000) also find that the costs will be much heavier for states and local areas that receive many incremental immigrants, while states with few immigrants should reap the advantages of reduced federal and social security taxes without bearing the local costs of education and health care for immigrants.

Immigration data used in this chapter are from the March Current Population Survey, a multi-stage stratified sampling survey of about 60,000 households which is the source for official government employment statistics. Although the population weights used by the Survey are adjusted to control for undercount, the adjustment is based on the 1990 Census, and thus may not reflect both Census undercount and changes in illegal immigration trends since 1990. Demographers believe that the undocumented population in the United States is currently close to 10 million people (Meissner, 2004).

For an H-1B visa, the minimum qualification is a bachelor’s degree or higher in the worker’s specialty skill. The program is most commonly used in the information technology and computer industries. Almost half of those admitted on H-1B visas in the last six to eight years have been from India, with China the next largest source country.

The Fair and Secure Immigration Reform (FSIR) would grant temporary legal status to illegal immigrants working in the United States by releasing 3-year work cards to current undocumented workers who choose to register. They would be allowed to travel freely in and out of the United States, but would have to return to their native country when the permit expires (the permit would be renewable for only one additional 3-year term). Moreover, U.S. employers could advertise jobs on a new internet labor exchange, and if no native worker accepted, could go abroad and get guest workers, who would receive three-year renewable visas like those issued to unauthorized workers in the United States. The proposal would also increase immigration limits to accommodate part of the higher demand for temporary immigration visas available for employers who cannot find U.S. workers, currently 140,000 a year for workers and their families.

More recently, Germany has introduced a point system based on these models (OECD, 2004).

In international trade theory, the Rybczynski theorem argues that an increase in the labor-to-capital ratio would lead to a change in a country industry-output mix (towards more labor-intensive goods), rather than to changes in wages and profits. Empirical studies provide some support for this theorem (Borjas, 1987; Card 1990), while other work suggests that immigration may have led at most to a slight decrease in wages of unskilled native workers (Borjas, Freeman, and Katz, 1997; Greenwood, Hunt, and Kohli, 1997).

While profits are included in national income, only dividend, interest, and rental income are included in the CPS. Using the survey, therefore, would amount to excluding retained earnings that belong to equity owners. In order to include them into household income, retained earnings are estimated by applying the retained earningto-dividend ratio from firm-level data to dividend income in the CPS. U.S. firms’ retained earnings-to-dividend ratio is obtained from Worldscope data, a database which provides information on the balance sheet of almost all listed firms in the United States.

See Jones (1965) for an analysis. The inter-industry effect is a general equilibrium effect, rather than a pure technological externality. This chapter does not explicitly consider policy-related externalities of immigration, including those related to the impact of immigration on the tax and social security systems.

The Solow residual is estimated for 23 U.S. industries, using several industry data sources including the database used by Jorgenson, Ho, and Stiroh (2004) in their latest study on the U.S. productivity performance (see Chapter I of this paper). Labor input from immigrants is derived as the share of total hours worked that can be attributed to immigrants. Being based on the CPS, this estimate is likely to suffer from undercount of illegal immigration (see above). Moreover, the regressions are affected by any measurement error in the estimation of TFP growth. With CPS data on U.S. immigration starting only in 1994, labor input from immigrants for the aggregate specification has been proxied by the growth of Mexican remittances.

The set of control variables includes capital services and labor input growth to capture general externalities from labor and capital; year dummies to control for any aggregate shocks; and, in the fixed effect model, the lagged Solow residuals to control for serial correlation, the lagged level of capital and the level of labor.

These results appear consistent with Caselli and Coleman II (2000), who suggest that countries with the most efficient use of skilled labor and capital—e.g., the United States—also tend to make the least efficient use of unskilled labor.

Data on remittances are from all destinations, not only the United States. To reduce the extent of the bias the figure is limited only to Latin American countries, for which remittances from the United States are more likely to account for a vast majority of the total. Indeed, estimates of remittances from the United States recently released by the Inter-American Development Bank (IDB, 2004) are not very different from aggregate data on remittances for Latin American countries for 2003.

As is the case for Mexico, the GMP estimates presented here are only based on income from migrants to the United States, which probably captures the bulk of migrants’ income for Latin American countries but only a portion of the total figure for other countries.

Owing to lack of data, GDP rather than GNI is used for China, Colombia, Haiti, Peru, Poland, Russia, and Trinidad and Tobago.

Other Resources Citing This Publication