This site has moved to
The posts below are backup copies from the new site.

December 8, 2007

David Card on Immigration and U.S. Cities

David Card on "How Immigration Affects U.S. Cities". Here is the abstract, introduction, and conclusion:

How Immigration Affects U.S. Cities, by David Card, June 2007 [via]: Abstract In the past 25 years immigration has re-emerged as a driving force in the size and composition of U.S. cities. This paper describes the effects of immigration on overall population growth and the skill composition of cities, focusing on the connection between immigrant inflows and the relative number of less-skilled workers in the local population. The labor market impacts of immigrant arrivals can be offset by outflows of natives and earlier generations of immigrants. Empirically, however, these offsetting flows are small, so most cities with higher rates of immigration have experienced overall population growth and a rising share of the less-skilled. These supply shifts are associated with a modest widening of the wage gap between more- and less-skilled natives, coupled with a positive effect on average native wages. Beyond the labor market, immigrant arrivals also affect rents and housing prices, government revenues and expenses, and the composition of neighborhoods and schools. The effect on rents is the same magnitude as the effect on average wages, implying that the average “rent burden” (the ratio of rents to incomes) is roughly constant. The local fiscal effects of increased immigration also appear to be relatively small. The neighborhood and school externalities posed by the presence of low-income and minority families may be larger, and may be a key factor in understanding native reactions to immigration.

Introduction The U.S. is once again becoming a country of immigrants. Immigrant arrivals – currently running about 1.25 million people per year – account for 40% of population growth nationally, and a much larger share in some regions.[1] The effects of these inflows are controversial, in part because of their sheer size and in part because of their composition. Something like 35-40% of new arrivals are undocumented immigrants from Mexico and Central America with low education and limited English skills (Passel, 2005). Although another quarter of immigrants – from countries like India and China – are highly skilled, critics of current immigration policy often emphasize the presumed negative effects of lower-skilled people in the overall economy (e.g., Rector, Kim and Watkins, 2007). Moreover, even the most highly skilled immigrants are predominately non-white, contributing to the growing presence of visible minorities in the U.S. population.

The size and composition of immigrant inflows is a special concern in the nation’s largest metropolitan areas, where most immigrants live. To illustrate this point Table 1 presents data from the most recent Current Population Surveys on the 17 largest metropolitan areas (with populations of 2 million or more).[2] The average share of immigrants in these cities is nearly 27% – two times the level in the nation as a whole – although the fraction varies from a low of 8% in Philadelphia and Detroit to a high of 35% in Los Angeles and Miami. Large cities also have a disproportionate share of “second generation” Americans: people born in the U.S. with at least one foreign-born parent. This group represents about 20% of the population in the 17 largest cities (versus 11% in the nation as a whole), with a range between 8% (in Atlanta, Minneapolis, and Philadelphia) and 25% (in Los Angeles).


The final column of Table 1 shows the overall share of minorities (i.e., people who are either non-white or Hispanic) in each major city.[3] As illustrated in Figure 1, the fraction of minorities rises one-for-one with the share of Asian and Hispanic first- and second-generation immigrants. Thus, the “diversity” of a city’s population is intimately related to the immigrant presence in a city.


This paper summarizes some of the main effects of immigration on major U.S. cities. I begin by examining the effect on overall population growth. Immigrants, like natives, are drawn to expanding cities, making it harder to draw inferences about the causal effect of immigrant inflows on population growth. Unlike natives, however, immigrants are particularly attracted to cities with historical enclaves of earlier immigrants. Exploiting this immigrant-specific “pull factor” I infer that immigrant arrivals increase the local population, with only limited outflows of other groups.

I then consider the effects of immigration on the composition of a city’s workforce. Immigrants as a whole are less skilled than natives, so in the absence of selective outflows more immigrants tend to reduce the average skills of the local population. Although the native populations of certain high-immigrant cities – like San Francisco and Boston – are relatively skilled, I show that offsetting native population flows tend to be small, so on average cities with more immigration have a larger share of lower-skilled workers.

I then turn to the implications of this unbalanced population growth for local incomes and the structure of the local economy. Despite the impact of immigration on the skill composition of the local labor market, there is only a small effect on the structure of relative wages.[4] The wage gap between the lowest-skilled natives (who are in most direct competition with immigrants) and natives at the middle of the skill distribution is 3-5 percentage points wider in high immigrant cities like New York and Los Angeles than in low-immigrant cities. The gap between the wages of the highly-skilled natives and those in the middle is also somewhat wider in high-immigrant cities.

An equally important issue is the effect of immigration on average wages of native workers. Even after controlling for city size effects, human capital spillovers, and the possibility that immigrants are drawn to cities with stronger local economies, the evidence suggests a positive effect. Taken together with the rather small magnitude of the relative wage effects, it appears that immigration exerts a modestly positive effect on the labor market outcomes of most natives. This parallels the conclusion reached in simulations of the national impacts of immigration by Ottaviano and Peri (2006), and in a recent empirical study for California by Peri (2007).[5]

Immigrants also affect the well-being of urban residents through a variety of other non labor market channels. One mechanism is the housing market: research by Saiz (2003, 2006), Susin (2001), and Greulich, Quigley, and Raphael (2004) suggests that rents are higher in cities with increased immigration, although Greulich, Quigley, and Raphael (2004) find that the mean ratio of rent-to-income (the “rent burden”) among natives is unaffected by immigrant inflows. In fact, the estimated impacts of immigration on average rents in the studies by Saiz (2006) and Greulich, Quigley, and Raphael (2004) are about the same magnitude as the estimated impacts on average native earnings. This parallel shift explains the absence of an effect on the average rent burden and suggests that net housing market effects on natives are small.

Another channel is through government revenues and expenditures. Simple “static” comparisons suggest that immigrants pay about $100 per capita less than natives in state, federal, and Social Security taxes, and receive about $600 per capita less in cash transfers. As has been recognized in a number of “inter-generational accounting” studies, such comparisons depend on the treatment of the native-born children of immigrants (who pay less taxes than other natives, and also receive more cash transfers). They also ignore the contributions of immigrants to local taxes (mainly property and sales taxes) and local expenditures (mainly on schools and other services). And perhaps most importantly, they ignore the general equilibrium effects of immigration on the earnings and program participation of the local native population.

To provide a broader perspective I consider the relationship between the fraction of immigrants in a city and various indicators of local fiscal conditions, including average earnings per capita (a simple measure of the “tax base”) and school enrollment rates. This analysis suggests that cities with more or less immigrants have similar (or slightly higher) earnings per capita, and similar “demographic burdens”.

Even apart from housing prices or fiscal concerns, immigrant families pose a potential externality on other urban residents through their impacts on school and neighborhood “peer groups.” While the monetary value is hard to quantify, existing research suggests that people value neighborhoods and schools with better-educated, higher-income, and non-minority neighbors and schoolmates. Indeed, my reading is that these peer group externalities may be a first-order concern among many urban residents.


VII. Summary and Conclusions

Immigration is changing the size and composition of major cities in the United States. Cities with larger immigrant inflows have experienced faster population growth, increased racial and ethnic diversity, and a rise in the share of lower-skilled workers. In many respects, these changes parallel the experiences of high-immigrant cities in the late 19th and early 20th Centuries (Perlmann, 2005). And, as in the earlier wave of immigration, they have led to extensive debates about the impacts on native labor market opportunities, housing prices, local government finances, and the overall quality of urban life.

This paper presents an overview of the evidence on the magnitudes of these effects. Building on the past two decades of labor economists’ research on immigration, most of this evidence is based on cross-city comparisons. The variation in the fraction of immigrants across different U.S. cities is enormous, providing a powerful research design. A key limitation is that immigrants are drawn to cities with better job prospects, potentially obscuring the negative impacts of their presence. The main solution that has been devised is to use the “pull” of city specific enclaves to isolate a component of immigrant inflows that is unaffected by local demand conditions. This approach has been supplemented in the literature by case studies of events like the Mariel Boatlift that provided large and arguably exogenous inflows to specific cities based on external political events.

More than two decades of research on the local labor market impacts of immigration has reached a near consensus that increased immigration has a small but discernable negative effect on the relative wages of low-skilled native workers (i.e., the ratio of low-skilled wages to wages in the middle of the skill distribution). Less is known about the impact on the average level of native wages, but theoretical reasoning, as well as the evidence presented here (and in a few other studies, mainly at the national level) suggests a small positive effect. Overall, I conclude that native wages are a little higher, on average, but a little more unequally distributed in a high immigrant city like Los Angeles than in a low-immigrant city like Philadelphia.

Moving beyond the labor market, economists have recently investigated the impacts of immigration on housing markets and on government expenditures and revenues. My reading of the existing housing market studies is that increased immigration leads to somewhat higher rents, with an effect that is widely diffused across the market (i.e., across units that are more or less likely to be rented by new immigrants), and about the same size as the effect on average wages. So far, studies of the fiscal impacts of immigration have not used a cross-city research design. Simple comparisons of some key indicators of local fiscal impact – including the local share of the population who are of working age, and per capita earnings – suggest that local fiscal impacts are small, although these conclusions may be refined in subsequent work. Taken together with the labor market evidence, it seems that the direct economic impacts of immigration on existing native residents of major U.S. cities are relatively small, and may well be positive.

Nevertheless, there is strong evidence that many U.S. natives prefer to live in neighborhoods and school districts with fewer minorities and more high-income/highly-educated residents. Newly arriving immigrants pose a “peer group” effect that may partially offset or even completely reverse any positive labor market impacts. One clear indicator of a reaction to this effect is the rise in measures of school segregation between white non-Hispanics and Hispanics in many large cities over the 1990s. My view is that such “peer effects” – whether driven by genuine concern about spillovers from neighbors or schoolmates, or by perceived threats to social or group identity – may well be the most important cost of increased immigration in many natives’ minds.

*Prepared for the conference “Unraveling the Urban Enigma: City Prospects, City Policies”, University of Pennsylvania, May 4 2007. I am grateful to Robert Inman, Christian Dustmann, and Michael Greenstone for helpful comments and suggestions.

[1] See U.S. Department of Commerce (2006).

[2] I use the term “cities” in this paper to refer to wider metropolitan areas rather than to specific political entities. As of 2003 the Federal government has defined a new set of metro areas – so-called Core Based Statistical Areas (CBSA’s) – that are similar to the Metropolitan Statistical Areas (MSA’s) and Consolidated Metropolitan Statistical Areas (CMSA’s) used in the past two decades. For the largest metro areas – like New York – the CBSA is comparable to the older CMSA, though typically smaller.

[3] Appendix Table 1 presents a more detailed comparison between the racial/ethnic breakdown of the first and second generation immigrants in each city.

[4] See Friedberg and Hunt (1995) for a survey of the literature up to the mid-1990s. More recent studies of local labor impacts include Card (2001), Orrenius and Zavodny (2006), Card and Lewis (2007), Dustmann, Fabbri, Preston, and Wadsworth (2003), and Glitz (2006).

[5] Ottavanio and Peri (2006) extend the model developed by Borjas (2003) to allow for imperfect substitution between immigrants and natives with similar age and education. In their simulations that allow for capital growth, wages of most natives are increased by immigration.

No comments: