This paper attempts to develop estimates of emigration of the foreign-born population at the national level based on the numbers of foreign-born persons enumerated in the 1980 and 1990 censuses. The methodology employed here is a residual technique in which the counted population is subtracted from the expected population to obtain the amount of emigration and emigration rates. The total amount of emigration of the foreign-born population was about 195,000 per year during 1980-1990. The new estimate, which is about 47 percent higher than the current level of 133,000 per year, is consistent with the recent increase in immigration. Of the 1980 foreign-born stock, the highest emigration rate was for the cohort of 1970-1979 (19 percent), followed by the 1960-1969 (9 percent) and the before 1960 cohorts (7 percent). The age pattern of emigration shows a declining trend by age for both males and females. The pattern is similar for all races.

The emigration rates for non-Hispanic Whites and the Asian and Pacific Islander population hovered around 10 to 11 percent. The emigration rates for Blacks and Hispanics were 14 percent and 7 percent, respectively. The low emigration rate for Hispanics was probably because of the same reasons for which Hispanic immigration is high (e.g., relative difference in economic and social conditions at origin and destination countries). The high rate for Blacks was because of high rate for the 1970-1979 cohort and relatively larger proportion of Blacks who came recently.

Disclaimer

The views expressed are attributable to the author and do not necessarily reflect the views of the U. S. Bureau of the Census.

If you have any questions concerning this report, please e-mail a message to pop@census.gov. Include the name of this report and author in the body of the message.

1.  INTRODUCTION

2.  METHODOLOGY

3.  DATA EVALUATION

4.  ESTIMATES OF EMIGRATION

5.  SUMMARY AND CONCLUSIONS

REFERENCES

The Bureau of the Census assumes a fixed amount of emigration of foreign-born persons, 133,000 per year, for its current population estimates and projections (e.g., Day, 1993; Hollmann, 1993). The distribution of the total amount of emigration by age, sex, and race is based on the 1980 census distribution of foreign-born persons who entered during 1975- 1979. The Bureau's current assumed level is an extrapolation of emigration levels developed by Warren and Peck (1980) for 1960- 1970, and by Warren and Passel (1987) for 1970-1980. The Warren- Peck estimates were based on analyses of the foreign-born populations counted in the 1960 and 1970 censuses and the number of immigrants that entered during 1960-1970. The Warren-Passel estimates were based on analyses of alien registration data collected by Immigration and Naturalization Service (INS) for 1965-1980. Neither of these studies, however, produced emigration rates by detailed demographic characteristics (e.g., age, sex, race, country of birth and year of entry) that could be used in simulating emigration in the future. In a more recent study, Woodrow (1991) estimated emigration levels for the 1980s, using data from multiplicity sample surveys. Although the Woodrow estimates contain the Bureau's current assumed level, they are less valuable for calculation of rates or distribution by characteristics because of small samples of respondents with emigrant relatives.

Are the current emigration assumptions sustainable for the rest of the 1990s and beyond? Can we keep assuming that the amount and characteristics of emigrants will not vary with changes in the amount and characteristics of recent immigrants, especially their age and country of origin? The Bureau's population estimates and projections and census coverage estimates are likely to be biased to the extent the level and characteristics of emigrants are changing. To ensure the quality of population estimates and projections and census coverage estimates, development of new estimates of emigration, or better yet, emigration rates, is warranted.

This paper attempts to develop estimates of emigration of the foreign-born population at the national level based on the numbers of foreign-born persons enumerated in the 1980 and 1990 censuses. The methodology employed here is a residual technique in which the counted population is subtracted from the expected population to obtain the amount of emigration and emigration rates. The paper first develops 10-year emigration rates (and levels) of foreign-born persons who were resident in 1980 (stock data) and then uses these rates to obtain emigration levels of those who came during 1980-1990 (flow data). This is the first time that emigration rates have been produced by detailed characteristics, such as age, sex, race, country of birth and year of entry.

The next section--Section 2--describes the residual technique employed in the analysis. It also focuses on the grouping of countries by race-ethnicity, and the selection of life table survival rates which are essential in producing the expected 1990 population. Section 3 evaluates the quality of census data on the foreign-born population from the perspective of its utility for estimating emigration. Section 4 presents the estimated numbers of emigrants and emigration rates. The last section summarizes the main findings and discusses the development of alternative sets of emigration levels.

Residual Technique

Both the 1980 and 1990 censuses collected information on foreign-born persons by detailed characteristics, including their year of entry. The residual technique employed in this analysis starts with two sets of foreign-born populations from these censuses on a cohort basis. The first or initial population is the foreign-born that came to the United States before 1980 and was enumerated in the 1980 census. This population is carried forward to 1990 to produce an expected 1990 population. The second or final population is the foreign-born that also came to the United States before 1980, but was enumerated in the 1990 census. If the population remains closed, that is, nobody ever dies or moves in or moves out, then the final population should be equal to the initial population. However, mortality always takes its toll and international migration not only occurs but is increasing. Note that by fixing the period of entry at "before 1980" we have eliminated the component "moving in", and are left with the components "moving out" or emigration and mortality. The final population then can be represented as follows:

editor's note: dates are subscripted

where P₁₉₉₀ = the foreign-born population that came before 1980 and was enumerated in 1990, P₁₉₈₀ = the foreign-born population that came before 1980 and was enumerated in 1980, D_1980-1990 = the number of deaths that occurred to P₁₉₈₀ during 1980-1990, and E_1980-1990 = the number of foreign-born persons who left during 1980-1990. By rearranging equation (1), we get:

or,

where S₁₉₉₀ is the expected or survived number of foreign-born persons after adjusting for mortality. S₁₉₉₀ can also be obtained by directly multiplying P₁₉₈₀ by 10-year life table survival rates; this is further discussed later in this section.

Equation (3) gives us the amount of emigration during 1980-1990 out of the initial stock, P₁₉₈₀, that is, the foreign-born population that came before 1980 and was enumerated in 1990. The 10-year emigration rate is obtained by (E_1980-1990/P₁₉₈₀) and the yearly emigration rate by 1/10*(E_1980-1990/P₁₉₈₀). These rates can be estimated by age, sex, race, year of entry and country of birth of the foreign-born population.

Equation (3) cannot be used directly to estimate emigration of the foreign-born population that came during 1980-1990 for it depends on expected (or initial) and final populations. Note that the foreign-born population that came during 1980-1990 and was enumerated in the 1990 census is the final population, not the initial one. We can estimate the initial population from the following equation which is similar to equation (1):

where P^*₁₉₉₀ = the foreign-born population that came during 1980-1990 and was enumerated in 1990 (stock data), P_1980-1990 = the foreign-born population that came during 1980-1990 and was subject to both mortality and emigration (flow data), e = emigration rate, and s = survival rate. By simplifying equation (4), we get:

or,

Equation (6) will hold true if s > e. If P^*₁₉₉₀, s and e are known, we can estimate P_1980-1990 by using equation (6) and the amount of emigration during 1980-1990 by P_1980-1990*e.

Note that both e and s pertain to the average number of years that P_1980-1990 was expected to stay during the 1980-1990 period. Because P_1980-1990 was not directly available, we estimated the average length of stay by using the proportions of P₁₉₉₀ that entered during the periods 1980-1981, 1982-1984, 1985-1986 and 1987-1990, and the length of the interval from the mid-point of each period to April 1, 1990. The average length of stay was estimated by age, sex and race.

Also note that both e and s refer to the initial population's average age at time of entry. The average age at entry during 1980-1990 can be estimated by subtracting the average length of stay from the average age of the final population in 1990.

Limitations

The emigration estimates developed with these models are subject to the following assumptions: (1) both the 1980 and 1990 foreign-born population data were accurately enumerated or had the same rate of coverage, (2) information on year of entry and country of birth was accurate and consistently reported in both censuses, and (3) the life tables selected for estimating the number of deaths are appropriate. No direct evidence on the extent of coverage of the foreign-born population in the United States censuses exists. Although it is possible that the coverage of the foreign-born may differ by country of birth, it is resonable to assume that for a particular country the extent of coverage would remain the same over time. As for year of entry, we are not aware of any systematic research that says the information collected was inaccurate. For both 1980 and 1990, the foreign-born populations with unknown countries of birth were allocated to known countries. Finally, the life table survival rates selected for estimating the number of deaths seem to be reasonable (see the discussion on life table selection).

Grouping of Countries by Race

Tabulated 1980 census data on the foreign-born population, in age-sex-period of entry detail, are available for 40 mutually exclusive major countries or areas: 10 areas are from North and Central America, 5 from South America, 13 from Europe, 9 from Asia, one from Africa (Africa as a whole), and 2 from Oceania. Tabulated 1990 census data on the foreign-born population, in demographic detail, are available for 111 countries of birth. For comparison purposes, we recoded the 111 countries to the 40 areas available from 1980.

Estimation of emigration rates should be based on large enough data sets to ensure the stability of the rates while still reflecting some regional or race-ethnic patterns. For our purposes, we divided the 40 countries into 4 race groups as follows:

Although Other North and Central America and Other South America contain countries which are non-Hispanic, they were grouped with Hispanic countries in the 1980 tabulations. This grouping seems to be justifiable based on the 1990 census data on Hispanic origin (see Table A1).

Life Table Selection

Ideally, one should use average survival rates obtained from the 1979-1981 and 1989-1991 decennial life tables for surviving the 1980 foreign-born to 1990. Although the 1979-1981 life tables are available by race, they are not available for non- Hispanic Whites, Asian and Pacific Islanders or Hispanics. The National Center for Health Statistics has not yet published the 1989-1991 life tables by race. For internal use, the Census Bureau's population projections staff developed life tables for all races in 1990, and for Hispanics in 1980. Using these tables and assuming that the life tables for Whites in 1979-1981 would approximate those for non-Hispanic Whites and Asian and Pacific Islanders in 1980, we calculated the average survival rates in 1985 (see Table A2).

As expected, the 1985 rates lie between the 1980 and 1990 rates for most age groups. As shown by their ratios, the 1990 survival rates are slightly higher than 1985 rates. The ratios are generally higher for older age groups, reflecting greater reduction in mortality in older ages. The ratios are very high for Asian and Pacific Islanders in ages 55 and above. The disproportionately higher ratios for this population may originate directly from the life table assumptions. Recall that we took the 1979-1981 White life tables for Asian and Pacific Islanders in 1980. Because White survival rates are usually lower than the Asian and Pacific Islanders (see 1990 survival rates for these groups), the average of the 1990 and 1980 rates was lower than expected and hence the ratio of 1990 to 1985 was higher than expected.

The anomalies in the 1985 survival rates led us to choose 1990 rates over 1985. Our choice was also guided by another reason. Theoretically, the estimated total number of deaths is the sum of deaths in the United States and deaths after emigration. This means that part of the deaths should be added back to the number of emigrants. If we do not do this adjustment, the number of deaths would be overestimated and emigration underestimated, especially at older ages. Because we are not aware of any other robust procedure to accomplish this adjustment, we opted for using higher survival rates as provided by the 1990 schedule.

Table 1 shows the differences between the expected and the counted foreign-born populations for the selected 40 countries or areas of birth. Ideally, under the given assumptions, the difference (1990 expected minus 1990 counted) should be positive for each country because it is a measure of the amount of emigration. The difference is positive for 31 countries (for both males and females). The positive difference (for both sexes) is 0-5 percent for 5 countries, 5-10 percent for 5 countries, and 10 percent or more for 21 countries. For 9 countries, however, the differences are negative (either sex) and implausible. Of the countries with negative differences, 6 are Hispanic (Mexico, Dominican Republic, El Salvador, Guatemala, Colombia and Peru), 2 Black (Haiti and Trinidad and Tobago), and one Asian (Other South and East Asia). The distribution of the countries according to the percent difference between the expected and the counted numbers of foreign-born persons is given below:

In general, the differences between the expected and counted populations are higher (or positive) for the countries in Western Europe, Africa, Oceania, Middle East and Oriental Asia and lower (or negative) for those in Latin America. The majority of positive differences indicate that both the 1980 and 1990 census data on the foreign-born population were consistent for a large number of countries. But why is the difference negative for many countries in Latin America? Examination of the numbers by age and sex (not shown) reveals that the differences for these countries are even more inconsistent (negative) by age.

Such inconsistencies may arise from a number of sources: (1) differential undercoverage of the foreign-born population in the 1980 and 1990 censuses, (2) misreporting of year of entry and nativity, and (3) overestimation of mortality. As explained in the previous section, we can initially rule out the possibilities (2) and (3). As for undercoverage, there is no direct evidence available to explain the inconsistent differences between the expected and counted foreign-born populations. However, the negative differences for 9 countries could result if coverage of the population born in those countries was more complete in 1990 than in 1980.

To investigate the possibility that the 1990 coverage was higher (or 1980 coverage was lower) for the countries with inconsistent data, we examined the legalizations data under the Immigration Reform and Control Act (IRCA) of 1986. We hypothesized that the countries with negative differences were major sources of undocumented aliens who were hard to count in the 1980 census but relatively easier to count in the 1990 census because of adjustment of legal status. As Table 2 shows, the 9 countries with negative differences for either sex (Mexico, El Salvador, Guatemala, Peru, Colombia, Dominican Republic, Haiti, Trinidad and Tobago, and Other South and East Asia) were the source of 89 percent of the 2.5 million IRCA adjustees who became permanent residents during 1989-1991. Another 3 percent of the aliens adjusting status were from the 5 countries with differences of less than 5 percent each (Cuba, Ecuador, Jamaica, India, Philippines). Mexico alone accounted for about 75 percent of the IRCA adjustees whereas Europe as a whole accounted for only 1.3 percent. Examination of the relationship between the difference and the proportion of the 1990 foreign-born population that is IRCA adjustees also shows that large proportions are generally associated with negative differences.

How are we to estimate emigration for countries with inconsistent data? One procedure would involve adjusting the foreign-born population by the extent of undercoverage and then estimating emigration. Another procedure would estimate a schedule of emigration rates based on countries with consistent data and then apply these rates to extrapolate emigration for other countries. We adopted the latter procedure because the former involves making too many assumptions about the extent of undercoverage.

Table 3 shows the countries or areas selected for estimating emigration rates by race. The selection process involved the estimation of standard errors of the differences between the expected and counted foreign-born populations (not shown). We excluded the countries that had statistically significant negative differences for either sex (except Jamaica). These countries or areas are labelled "Non-Selected Countries" in Table 3. The selected countries together were origins for more than three-fourths of the 14 million foreign-born persons resident in 1980. Race-wise, the selected countries were origins for 41 percent of Hispanics, 100 percent of non-Hispanic Whites, 36 percent of Blacks, and 78 percent of the Asian and Pacific Islander population.

As mentioned in the preceding section, we first estimated emigration for the selected countries that had consistent data on the foreign-born population and then for non-selected countries with inconsistent data. We present these in order.

Emigration to Selected Countries

The estimates of emigration for the selected countries were developed in two steps: (1) the estimates for foreign-born persons who were resident in 1980 (stock), and (2) the estimates for those who came during 1980-1990 (flow). The 1980 foreign- born stock was the basis for the estimation of emigration rates during 1980-1990. We first present the estimates for the stock followed by those for the flow.

Emigration For Those Who Came Before 1980

From among the 10.6 million foreign-born persons who had come to the United States from the selected countries before 1980, about 1,198,000 or 11 percent emigrated during 1980-1990 (see Table 4). Of these emigrants, 634,000 came during 1970- 1979, 221,000 during 1960-1969, and 343,000 before 1960. The emigration rates were 19 percent for the cohort of 1970-1979, 9 percent for 1960-1969, and 7 percent for before 1960. The rates are consistent in that the highest rate was for the most recent period of entry and the lowest for the most distant past. The rates for males and females were very close; each shows a declining trend by period of entry.

The emigration rates by race also seem to be consistent. Overall, the highest emigration rate was for Blacks (20 percent), followed by the Asian and Pacific Islander population (16 percent), non-Hispanic Whites (11 percent) and Hispanics (8 percent). The emigration rate for Blacks was consistently high in all three periods of entry. The rate for Hispanics was consistently low in all three periods of entry.

For all races, the age-specific emigration rates by sex were estimated for the three periods of entry: 1970-1979, 1960-1969, and before 1960. Table 5 shows the age-specific emigration rates by sex for those who had entered during 1970-1979. The age- specific rates for males are plotted in Figure 1 and those for females in Figure 2. Irrespective of race and sex, the emigration rates of recent immigrants show a similar pattern: high at younger ages, low during adolescence, high again during college life, low again during working life, high again during retirement and low again after retirement. Despite the fluctuations associated with life cycles, the age pattern of emigration shows a declining trend by age.

Emigration for Those Who Came During 1980-1990

As explained in section 2, in order to estimate the amount of emigration for those who came during 1980-1990, we need to estimate the following: (1) average length of stay during 1980- 1990, (2) average age at time of entry, (3) survival rate, (4) emigration rate, and (5) the initial foreign-born population that was at risk of death and emigration. The average length of stay is the weighted average of the duration to April 1, 1990 from the middle points of the years 1980-1981, 1982-1984, 1985-1986, and 1987-1990. The weights are the proportions of foreign-born persons who came during these years and were counted in the 1990 census. As shown in Tables A3 and A4, the average length of stay varies by age, sex, and race. For example, for Hispanic males, the length is 5 years for the age group 20-25 in 1990 and 6 years for the age group 25-30 in 1990 (see Table A3). For Hispanic females, however, the length is 4 years for the age group 20-25 in 1990 and 5 years for the age group 25-30 in 1990 (see Table A4). For non-Hispanic White males and females, the length is 4 years for the age group 25-30 in 1990. Tables A3 and A4 also show the average age at time of entry of the initial population. This age was estimated by subtracting the average length of stay from the middle point of age groups in 1990.

Given the age at entry and the length of stay, we estimated the life table survival rates and emigration rates. As before, we used the 1990 life tables for the estimation of survival rates. For emigration, we assumed that the rate for those who came during 1970-1979 would apply to those who came during 1980- 1990, with the length of stay being different (less than 10). To conform to the estimated length, we first converted the 10-year emigration rate into a one-year rate (see Table 6) and then multiplied it by the estimated length. The new rates and the estimated numbers of emigrants--by age and race--are given in Table A5 for males and Table A6 for females.

The total amount and percent of emigration for those who came from the selected countries during 1980-1990 are given in Table 7 (columns 3 and 4). About 384,000 people or 8 percent of the foreign-born population that came during 1980-1990 also emigrated during the same period. Both the amount and percent of emigration for the 1980-1990 cohort are less than the corresponding figures for the before 1980 cohort (see columns 4 and 5). The amount is less because of its smaller base; the percent is less because the base was at risk of emigration for a shorter period.

Total Emigration to Selected Countries

Emigration of those who came before 1980 and those who came during 1980-1990 sum to 1,582,000 for the decade or about 158,000 per year (column 1 in Table 7). The number of male emigrants was 799,000, with a rate of 11 percent. The number of female emigrants was 783,000, with a rate of 9 percent. The emigration rates for non-Hispanic Whites and the Asian and Pacific Islander population hovered around 10 to 11 percent. The emigration rate for Blacks was about 14 percent and for Hispanics only 7 percent.

The estimated yearly emigration of 158,000 is about 19 percent higher than the Bureau's current level of 133,000. Note that this estimate is only for the selected population. It doesn't include emigration for foreign-born persons who came from the non-selected countries listed in Table 3. The size of the foreign-born population from these non-selected countries was 3.4 million or 24 percent of the 1980 stock. Mexico alone was the country of birth of 2.3 million people who were resident in 1980. For another 1.1 million people, the countries of birth were El Salvador, Guatemala, and Peru. What was the size of emigration for people who came from these non-selected Hispanic and other countries listed in Table 3?

Emigration To Non-Selected Countries

Does the inconsistency observed between the expected and counted foreign-born populations from the non-selected countries tell us there was no emigration for these countries? The answer is no. We are then left with the possibilities that the emigration rates for the non-selected countries were higher, the same as or lower than those for the selected countries. To derive a "reasonable" schedule of emigration rates for non- selected countries, our recourse was to consider the determinants of emigration.

Determinants of Emigration

Research on the determinants of emigration is scant. In assessing the role of emigration on Social Security's financial status, Duleep (1994) listed four factors that he felt best accounted for the lower emigration rate of the current and future foreign-born populations. First, the overall relative attractiveness of the country of origin in terms of its economic, social, and political condition likely affects emigration (Liu, 1975). If a country is disproportionately less attractive than another (both compared to the United States) then people migrating from the less attractive country would be less likely to return home. Second, just as the proximity to family and friends encourages migration, the presence of family and friends in the place of origin discourages migration (Lansing and Mueller, 1967; Fabricant, 1970; Greenwood, 1969). It is possible that the greater the presence of family and friends in the United States, the greater the likelihood that immigrants' costs of migration (e.g., social, economic, and psychological costs) would be shared by friends and kin, lowering the probability of emigration. Third, immigrants admitted as refugees tend to emigrate at a lower rate because of fear of political persecution (Warren, 1979; Jasso and Rosenzweig, 1990). Finally, the lower the entry earnings of immigrant cohorts (adjusted for education), the lower is the subsequent rate of emigration. Research has shown that immigrants with low-entry earnings have high-earnings growth (Chiswick, 1978a, 1978b, 1979; Duleep and Regets, 1992, 1994a, 1994b, 1994c), which in turn encourages investment in U.S.-specific skills and capital, and ultimately lowers the propensity to emigrate.

Emigration Rates for Non-Selected Countries

To extrapolate emigration rates for the non-selected countries, let us now turn to the application of the determinants of emigration to individual countries. Direct evidence on all the four factors of emigration noted above does not exist for these countries. Nonetheless, the available indirect evidence indicates that the emigration rates for the non-selected countries were lower than those for the selected countries. We discuss this with respect to race below.

Hispanic: The countries or areas that were selected for the estimation of Hispanic emigration rates are Argentina, Cuba, Dominican Republic, Colombia, Ecuador, Other North and Central America (major countries are Costa Rica, Honduras, Nicaragua, and Panama), Other South America (major countries are Bolivia, Chile and Venezuela) and Spain. The countries that were excluded from estimation or put in the "Non-Selected" category are Mexico, El Salvador, Guatemala, and Peru. As shown in Table 2, about 2.1 million aliens or 84 percent of the 2.5 million aliens who legalized under IRCA during the fiscal years 1989 to 1991 came from the non-selected Hispanic countries compared to about 0.1 million aliens or about 4 percent from the selected Hispanic countries. Although proximity was a factor, the enormous difference in the numbers of undocumented aliens (2.1 million versus 0.1 million) was probably because of less "attractiveness" of the non-selected countries. This validates our contention that the emigration rates for the non-selected countries would be lower than those for the selected countries.

The second emigration factor relates to the presence of family and friends in the United States. The Immigration Act of 1965 abolished the national origins quota system which favored the European countries, and established a system that would admit immigrants primarily on the basis of kinship ties. This change in federal laws coupled with changing economic opportunities in Europe was responsible for a major shift in the origin of immigrants--from European to Asian and Hispanic countries--in 1980. However, not each Hispanic or Asian country had the same share of total immigrants. As shown in Table 3, the non-selected Hispanic countries were the origins of 2.5 million foreign-born persons in 1980 compared to 1.8 million for the selected Hispanic countries. Admissions of family-based immigrants during the 1980s reveal similar disproportionate share of immigration. To the extent the presence of family and friends dilute the cost of staying away from one's home country, the emigration rate would be lower for the non-selected Hispanic countries.

The third factor refers to admissions of refugees and asylees. Two of the non-selected Hispanic countries--El Salvador and Guatemala--were sources of asylees during the 1980s. Inclusion of these countries in the non-selected category would have dampening effect on its emigration rate.

The last factor correlates the low-entry earnings leading to high-earnings growth and investment in U.S.-specific human capital with a low propensity of emigration. Although we did not examine data on earnings of immigrant cohorts, we can speculate that the average earnings of foreign-born persons from the non- selected Hispanic countries would be lower than that from the selected Hispanic countries. Our speculation follows from the influx of several million undocumented aliens and special agricultural workers from the non-selected countries, who are generally employed in low-paying jobs.

Black: Africa (as a whole) was included in the estimation of emigration rates for Blacks while Haiti, Jamaica, and Trinidad and Tobago were excluded (see Table 3). For the same reasons given for the non-selected Hispanic countries, the emigration rates for the non-selected Black countries (Haiti, Jamaica, and Trinidad and Tobago) would be lower than those for the selected Black countries (Africa).

Asian and Pacific Islander (API): The selected countries or areas for API emigration rates are China, India, Japan, Korea, the Philippines, and Other Oceania. The API area in the "Non- Selected" category is "Other South and East Asia," which includes major countries such as Vietnam, Cambodia, and Laos. The selected countries are relatively more attractive than the non- selected countries in terms of social, economic, and political conditions. Although China's political condition is different from Japan and Korea, the strict governmental policy and loyalty of the Chinese to their form of government places China in the group of attractive countries as far as emigration is concerned. Three of the non-selected API countries--Vietnam, Cambodia, and Laos--are economically disadvantaged and are major sources of Asian refugees. Following Duleep's (1994) conclusions, these considerations suggest that the emigration rates for the non- selected countries would again be lower than the selected countries from this region.

This discussion about the determinants of emigration and their specific application to individual countries confirms our original assertion that the emigration rates for the non-selected countries should be lower than those for the selected countries. But, how much lower? In the absence of any empirical basis, we shall take a middle ground assumption and establish emigration rates at 50 percent of the rates for the selected countries.

Emigration Estimates for Non-Selected Countries

Table 8 gives the levels of emigration during 1980-1990 for the non-selected countries based on assumed emigration rates (columns 6-10). The levels vary from zero for the assumption of zero emigration to 747,000 for the assumption of the same rate as for the selected countries. Our preferred assumption--one-half of the rate for the selected countries--produced a level of 367,000 for the decade. Of this amount, Hispanics accounted for 231,000, Blacks, 49,000, and the Asian and Pacific Islander population, 87,000. The amount of emigration for the Mexico-born population was 201,000.

Total Emigration During 1980-1990

The total amount of emigration during 1980-1990 is the sum of the amount of emigration for the selected countries and that for the non-selected countries. As shown in Table 8 (columns 1- 5), the amount varies from 1,582,000 to 2,329,000, depending on emigration assumptions for the non-selected countries. According to the middle-ground assumption, about 1,949,000 foreign-born persons emigrated over the 1980-1990 period. Among total emigrants, there were 439,000 Hispanics, 878,000 non-Hispanic Whites, 131,000 Blacks, and 501,000 Asians and Pacific Islanders.

The new single-year estimate of emigration--195,000--is about 47 percent higher than the currently used level of 133,000 per year. This higher amount of emigration is consistent with related events in recent U.S. history: the continuous rise in legal immigration, admissions of refugees and political asylees, and the general amnesty given to the undocumented aliens and special agricultural workers under IRCA.

Emigration by Country of Birth

Table 9 shows the numbers of yearly emigrants for 40 major countries or areas of birth. These numbers were derived by multiplying the base foreign-born population that came from each of these countries (by age, sex, and year of entry) by its corresponding "model-based" emigration rate (e.g., Hispanic, non- Hispanic White, Black, and API rates). The estimated numbers were independently controlled to 101,000 male emigrants and 94,000 female emigrants for a total of 195,000.

Europe received the highest number of emigrants per year (63,736), followed by Asia (61,071), North and Central America (52,512), South America (9,035), Africa (6,796) and Oceania (1,850). The five countries that received more than 10,000 emigrants per year are Mexico (20,068), Canada (11,241), Germany (11,887), China (10,149), and the Philippines (11,242). The United Kingdom is the only other country that had close to 10,000 emigrants per year.

Applying a systematic and innovative approach, this research has developed new estimates of emigration of the foreign-born population from the United States during 1980-1990. The approach is systematic in that it uses successive cohorts of foreign-born persons counted in the 1980 and 1990 censuses to develop schedules of emigration rates. It is innovative because the countries with consistent data on the foreign-born population were grouped together (by race) to give "model" emigration rates, and because the emigration for countries with inconsistent data were derived from the model-based rates in conjunction with a framework on the determinants of emigration. The main findings of this research are:

1. The total amount of emigration of the foreign-born population was about 195,000 per year during 1980-1990. The new estimate, which is about 47 percent higher than the current level of 133,000 per year, is consistent with the recent increase in immigration.
2. The overall emigration rate during 1980-1990 was about 10 percent for the selected countries. The male emigration rate was 11 percent and the female rate 9 percent.
3. Of the 1980 foreign-born stock, the highest emigration rate was for the cohort of 1970-1979 (19 percent), followed by the 1960-1969 (9 percent) and the before 1960 cohorts (7 percent).
4. The age pattern of emigration shows a declining trend by age for both males and females. The pattern is similar for all races.
5. For the selected countries, the emigration rates for non- Hispanic Whites and the Asian and Pacific Islander population hovered around 10 to 11 percent. The emigration rate for Blacks was about 14 percent while that for Hispanics only 7 percent. The low emigration rate for Hispanics was probably because of the same reasons for which Hispanic immigration is high (e.g., relative difference in economic and social conditions at origin and destination countries). The high rate for Blacks was because of high rate for the 1970-1979 cohort and relatively larger proportion of Blacks who came recently.

It is worth noting that the estimates of emigration developed here are sensitive to the levels of mortality assumed for each race group. This is because in the residual technique emigration and mortality are the only two forces of attrition of a given stock of foreign-born population. We believe that we selected the best available life table survival rates to estimate the number of deaths and thereby the number of emigrants. We are carrying out illustrative calculations to assess the sensitivity of the emigration estimates for alternative life table survival rates.

Also, the ranges of emigration estimates (presented in Table 8) were based on subjective judgement, which was guided by the review of the determinants of emigration. Because both the 1980 and 1990 censuses collected data on the foreign-born population on sample basis, the estimated levels of emigration are further subject to sampling error. We plan to estimate this error and build confidence intervals around our point estimate of 195,000 per year; this will be done in the near future.

Despite the limitations noted above, the new estimates of emigration or emigration rates should replace our current approach for estimating emigration in the 1990s. The reasons are:

1. The new approach shifts our estimation focus to the more methodologically correct basis of rates, which allows the level of emigration to vary as the population at risk (the foreign-born) varies;
2. The new estimates are more recent because they are based on the foreign-born populations enumerated in the 1980 and 1990 censuses;
3. The estimates are different (195,000 vs. 133,000) and more plausible, reflecting the rise in immigration;
4. The estimates are better because they are based on successive cohorts of foreign-born populations taken from countries with consistent data;
5. The estimates are detailed by demographic characteristics such as age, sex, race, and country of birth; and
6. We are able to document the details of our methodology.

Chiswick, Barry R. 1979. "The Economic Progress of Immigrants: Some Apparently Universal Patterns," in Contemporary Economic Problems, 1979. William Fellner (ed.). Washington, D.C.: American Enterprise Institute, pp. 359-399.

_____________. 1978a. "The Effect of Americanization on the Earnings of Foreign-Born Men," Journal of Political Economy, Vol. 86, No. 5 (October), pp. 897-922.

_____________. 1978b. "A Longitudinal Analysis of Occupational Mobility of Immigrants," Proceedings of the 30th Annual Winter Meeting, Industrial Relations Research Association. Barbara Dennis (ed.). Madison.

Day, Jennifer Cheeseman. 1993. Population Projections of the United States, by Age, Sex, Race, and Hispanic Origin: 1993 to 2050. U.S. Bureau of the Census, Current Population Reports, P-25-1104. Washington, D.C.: U.S. Government Printing Office.

Duleep, Harriet Orcutt. 1994. "Social Security and the Emigration of Immigrants," Social Security Bulletin, Vol. 57, No. 1 (Spring), pp. 37-52.

Duleep, Harriet Orcutt and Mark C. Regets. 1994a. "Admission Criteria and Immigrant Earnings Profiles." Program for Research on Immigration Policy. Discussion Paper No. PRIP-UI-30. Washington, D.C.: The Urban Institute.

_____________. 1994b. "Country of Origin and Immigrant Earnings." Program for Research on Immigration Policy. Discussion Paper No. PRIP-UI-31. Washington, D.C.: The Urban Institute.

_____________. 1994c. "The Elusive Concept of Immigrant Quality." Program for Research on Immigration Policy. Discussion Paper No. PRIP-UI-28. Washington, D.C.: The Urban Institute.

_____________. 1992. "Some Evidence on the Effect of Admission Criteria on Immigrant Assimilation," in Immigration, Language and Ethnic Issues: Canada and the United States. Barry Chiswick (ed.). Washington, D.C.: American Enterprise Institute, pp. 410-440.

Fabricant, Ruth A. 1970. "An Expectional Model of Migration," Journal of Regional Science, Vol. 10, No. 1 (April), pp. 13-24.

Greenwood, Michael J. 1969. "An Analysis of the Determinants of Geographic Labor Mobility in the United States," Review of Economics and Statistics, Vol. 51, No. 2 (May), pp. 189-194.

Hollmann, Frederick W. 1993. U.S. Population Estimates, by Age, Sex, Race, and Hispanic Origin: 1980 to 1991. U.S. Bureau of the Census, Current Population Reports, P25-1095. Washington, D.C.: U.S. Government Printing Office.

Jasso, Guillermina and Mark R. Rosenzweig. 1990. The New Chosen People: Immigrants in the United States. New York: Russell Sage Foundation.

Lansing, John B. and Eva Mueller. 1967. The Geographic Mobility of Labor. Ann Arbor: Survey Research Center, Institute for Social Research, University of Michigan.

Liu, Ben-Chieh. 1975. "Differential Net Migration Rates and the Quality of Life," Review of Economics and Statistics, Vol. 57, No. 3 (August), pp. 329-337.

Warren, Robert. 1979. "Alien Emigration from the United States: 1963 to 1974." Paper presented at the Annual Meeting of the Population Association of America, Philadelphia.

Warren, Robert and Jeffrey S. Passel. 1987. "A Count of the Uncountable: Estimates of Undocumented Aliens Counted in the 1980 United States Census," Demography, Vol. 24, No. 3 (August), pp. 375-393.

Warren, Robert and Jennifer Marks Peck. 1980. "Foreign-Born Emigration from the United States, 1960-1970," Demography, Vol. 17, No. 1 (February), pp. 71-84.

Woodrow, Karen A. 1991. "Emigration from the United States: Multiplicity Survey Evidence." Unpublished paper (August). Washington, D.C.: U.S. Bureau of the Census.