Journal Issue: Opportunity in America Volume 16 Number 2 Fall 2006
The Economic Performance of the Children of Immigrants
It is widely perceived that, on average, the children of immigrants far outperform their parents in economic terms. This perception originated in the early studies of Barry Chiswick and Geoffrey Carliner that compared the earnings of various generations of U.S. workers at a particular time, such as in the 1970 decennial census.6 Table 1 summarizes the evidence for three such crosssections: the 1940 census, the 1970 census, and the pooled 1994–2003 Current Population Surveys (which, for convenience, I will refer to as the 2000 cross-section).7
Each of these cross-section data files allows the precise identification of two generations of Americans: the immigrant generation (those born abroad) and the second generation (those born in the United States with at least one parent born abroad). It is impossible to determine precisely the generation of the rest of the sample (those born in the United States with American-born parents), but they are typically referred to as “third-generation” Americans—an extremely broad group ranging from grandchildren of immigrants to descendants of the Mayflower pilgrims.
For each of the available data cross-sections, table 1 reports the (age-adjusted) percentage wage differential between the average worker in the first and third generations, as well as between the average worker in the second and third generations.8 In every cross-section, the second generation earns more than both the first and the third generation. In 1970, for example, second-generation working men earned about 14.6 percent more than men in the third generation, while the first generation earned 1.4 percent more. Even though trends in the relative wage of working women are potentially influenced by selection biases associated with rising female labor force participation, the intergenerational pattern for women is exactly the same. The typical second- generation working woman earned 10.1 percent more than women in the third generation, while the typical immigrant working woman earned about 6.8 percent more. In 1940 second-generation working men earned 17.8 percent more than the baseline third generation, while immigrants earned 5.8 percent more. In 2000 second-generation working men earned 6.3 percent more than the baseline third generation, while immigrants earned 19.7 percent less.
The wage superiority of the second generation workforce in each cross-section snapshot seems to imply—and has been interpreted as implying—that second-generation Americans earn more than both their parents and their children. A common explanation is that the children of immigrants are “hungry” and have the drive and ambition to ensure economic success in the U.S. labor market—and that this hunger is lost once the immigrant household becomes fully Americanized, by the third generation. If this interpretation were correct, concern over the relatively low skill level of the immigrants of the past three decades would be misplaced. If historical patterns were to hold in the future, the children of these immigrants would outperform not only their parents but the rest of the workforce in only a few decades.
It turns out, however, that the evidence summarized in table 1 does not necessarily justify this inference. After all, the family ties among the three generations identifiable in any one cross-section of data are very tenuous. It is biologically impossible for most second-generation workers in a particular crosssection to be the direct descendants of the immigrants in the same cross-section. For instance, working-age immigrants in the 2000 cross-section (most of whom arrived in the 1980s and 1990s) typically cannot have American-born children who are also of working age. Second-generation Americans of working age can be the descendants only of immigrants who have been in the country for at least two or three decades. Put differently, most of the second-generation workers in the 2000 cross-section are unlikely to be the children of the immigrant workers in the same cross-section. Because of skill differences across immigrant cohorts—and because some of these differences could easily be transmitted to their children—the wage gap between first- and second-generation workers in a cross-section does not correctly portray intergenerational social mobility.
In short, the fact that second-generation workers earn more than other workers at a particular time does not necessarily imply that they earn more than either their parents or their children. To calculate the improvement in economic status between the first and second generations, one must link the economic performance of parents and children, rather than compare the economic performance of workers in different generations at a particular time.
Making the correct intergenerational comparison requires tracking the immigrant population over time.9 For instance, the 1970 census provides information on the economic performance of immigrants in 1970, many of whom are the parents of the second-generation workers in the 2000 cross-section. Similarly, the 1940 census provides information on the economic performance of immigrants in 1940, who are, in turn, probably the parents of the second-generation workers in the 1970 census. Only by comparing the economic performance of immigrant workers in 1940 with that of second-generation workers in 1970—or that of immigrant workers in 1970 with that of the second generation in 2000—can one correctly determine the economic progress made by the children of immigrants.
To illustrate, consider again the wage information summarized in table 1. If one (incorrectly) used only the information provided by the 2000 cross-section, we would conclude that because second-generation workers earn 6.3 percent more than the baseline third generation and because first-generation workers earn 19.7 percent less than the baseline, second- generation workers earn 26.0 percent more than first-generation workers. A correct calculation, however, reveals much less intergenerational improvement. After all, the typical immigrant in 1970 earned 1.4 percent more than the typical third-generation worker. And the typical second-generation worker in 2000 (who is presumably the descendant of the immigrants in 1970) earns 6.3 percent more than the baseline. In short, the true intergenerational growth in relative wages was only on the order of 5 percent—rather than the 26 percent implied by the intergenerational wage differences observed in 2000.
Similarly, the 1970 census seems to imply that the children of immigrants earn 13.2 percent more than their parents (14.6 percent minus 1.4 percent). But the economic status of the parents of these second-generation workers can be observed only in the 1940 census, where the immigrants had a relative wage advantage of 5.8 percent. The intergenerational wage improvement between 1940 and 1970 is then on the order of 8.8 percent (14.6 percent minus 5.8 percent). Again, immigrant households saw less wage growth across generations than would be implied by looking at the intergenerational wage differentials in a single cross-section. Still, however, the 5 to 10 percent intergenerational wage growth between the first and second generations is a substantial improvement in economic opportunity.
The bottom panel of the table reports the wage differences among the generations after the data are also adjusted for differences in workers' educational attainment. For both working men and women, much of the intergenerational progress in the “raw” data disappears. Put differently, much of the progress between the first and second generations (which leads to the 5 to 10 percent wage increase in the second generation) can be explained by differences in schooling between these two generations, as the native-born children of immigrants go through the American education system.
Finally, the evidence summarized in table 1 reveals a second potentially important pattern. In particular, note that the age-adjusted relative wage of immigrants has declined steadily since 1940. In 1940, the typical immigrant working man earned 5.8 percent more than workers in the third generation; by 1970, this economic advantage had fallen to 1.4 percent; by 2000, it had become a sizable disadvantage of 19.7 percent. Although this decline in the relative economic status of the immigrant workforce has been well documented, the data in table 1 also show a concurrent (and much less studied) decline in the relative wage of the second-generation workforce. In 1940, the typical second-generation working man earned 17.8 percent more than workers in the baseline third generation; by 1970, the wage advantage had fallen to 14.6 percent, and by 2000, to 6.3 percent. In short, the relative economic status of second-generation workers has been falling over time—just as that of the foreignborn workforce has been.
Put differently, the second-generation workers in the 2000 cross-section (whose parents made up the immigrant workforce in 1970) have a lower economic status than the second- generation workers in the 1970 crosssection (whose parents made up the immigrant workforce in 1940). If these historical trends continue for the next few decades, the forecast for the economic performance of the children of today's wave of immigrants could be bleak. Today's immigrants have a substantial wage disadvantage of 19.7 percent. If the intergenerational jump in relative earnings between the first and second generation is between 5 and 10 percent, their children will experience a 10 to 15 percent wage disadvantage around 2030. This remarkable turnaround in the economic status of the second generation highlights an important historical insight: the long-term trend in the relative economic performance of the second-generation workforce tracks that of the first generation, but with a generational lag.10