As of 2004 California employed almost 30% of all foreign born workers in the U.S. and was the state with the largest percentage of immigrants in the labor force. It received a very large number of uneducated immigrants so that two thirds of workers with no schooling degree in California were foreign-born in 2004. If immigration harms the labor opportunities of natives, especially the least skilled ones, California was the place where these effects should have been particularly strong. But is it possible that immigrants
raised the demand for California's native workers, rather than harming it? After all immigrants have different skills and tend to work in different occupations then natives and hence they may raise productivity and the demand for complementary production tasks and skills. We consider workers of different education and age as imperfectly substitutable in production and we exploit differences in immigration across
these groups to infer their impact on US natives. In order to isolate the "supply-driven" variation of immigrants across skills and to identify the labor market responses of natives we use a novel instrumental variable strategy. Our estimates use migration by skill group to other U.S. states as instrument for migration to California. Migratory flows to other states, in fact, share the same "push" factors as those to California but clearly are not affected by the California-specific "pull" factors. We find that between 1960 and 2004 immigration did not produce a negative migratory response from natives. To the contrary,
as immigrants were imperfect substitutes for natives with similar education and age we find that they stimulated, rather than harmed, the demand and wages of most U.S. native workers.
Giovanni Peri
Department of Economics
University of California, Davis
One Shields Avenue
Davis, CA 95616
and NBER
gperi@ucdavis.edu

1 Introduction

In the year 2004, California was home to almost 30% of all foreign-born individuals working in the U.S.; in turn, these foreign-born represented roughly one third of the almost 15 million workers employed in California, two thirds of California workers without a high school degree and almost half of the California workers with a doctoral degree. Many U.S.-born Californians moved out of the state during the nineties and job competition from immigrants has sometimes been regarded as a key factor for this outflow. It stands to reason that if recent inflows of immigrants indeed crowded out the labor market options of U.S. natives, specifically the low skilled ones, then such an effect should have been particularly strong in California. But is it possible that immigrants actually lifted California’s wages, rather than harming them? After all, immigrants have different skills and tend to work in different occupations than natives; they could make natives performing complementary production tasks more productive, thus increasing the demand for those tasks! The present study analyzes the effect of the migratory inflow on the employment, population and wages of U.S. natives in California, using data from the decennial U.S. Censuses and from the American Community Survey spanning the period 1960-2004. Our approach combines elements of a ”general equilibrium” (more structural) approach to immigration and wages, as proposed in recent national studies (Borjas 2003; Aydemir and Borjas 2007; Ottaviano and Peri, 2006) with the study of employment and inter-state migratory response of native workers typical of the so called ”area approach” (exemplified in Card 1990, 2001, Lewis, 2005 and Borjas, 2006). As in Borjas (2003) we consider labor as a differentiated input in production and we model the interactions between workers with different education and age using a nested CES production function. As in Ottaviano and Peri (2006) we allow for imperfect substitution between native and foreign-born workers within an education-age group (due to differences in skills, occupational choices and job opportunities) and we estimate the elasticity of substitution between natives and immigrants. The degree of substitutability between these two groups is a key parameter to determine whether immigrants increase or depress the demand for native workers. For a large degree of substitutability between natives and immigrants, uneducated immigrants mostly depress the demand for uneducated natives and augment the demand for highly educated natives. However, for a smaller degree of substitutability between natives and immigrants, uneducated immigrants have a much smaller depressive effect on (and may even increase) the demand for uneducated natives while still increasing the demand for more educated natives. As pointed out by the critics (e.g. Borjas et al, 1997) of the ”area approach” when focussing on a state economy it is important to account for the fact that any labor market effect of immigration can be ”diffused” to other states by out-migration of native workers. We carefully account for this effect in our empirical analysis. Moreover, by focussing on California over time, we are able to use a new identification strategy that addresses the problems of unobserved demand shocks and measurement errors; these are often deemed responsible for biased estimates of the impact of immigration on the labor market outcomes of natives in a state (e.g. Aydemir and Borjas, 2006; Borjas, 2006). Specifically, we use immigration to the other U.S. states by skill group over decades as a reasonable instrument to proxy the ”push”-driven component of immigration to California by skill group and decade. While sharing the ”push” factors behind international migrations with California, the flows of foreign-born workers to other U.S. states are not affected by California’s specific pull factors (i.e. unobserved demand shocks). The intuition of such identifying assumption is as follows. Suppose that immigrants with a college degree were “pulled” to California in the 90’s by the boom in the high tech sector, which increased the demand for workers with their qualifications. This would be a “pull factor” specific to California. The same boom probably would also have attracted (or reduced the potential outflow of) natives in the same skill group. It could thus create a positive correlation between foreign immigration and natives’ migration and wages–even if foreign migrants compete with natives for the same jobs. Such pull, however, would not be shared by other states and hence the instrument would not capture it. On the other hand take the cases of the increased international mobility of the college-educated Chinese middle-class or the worsened job outlook for young uneducated workers in Mexico during the nineties. Both are “push factors” that could increase immigration of some age-education groups to California (a large receiver of Chinese and Mexican migrants) as well as to other states. Push factors generate more migrants to California as well as to other states and are not related to changes in California’s local demand for labor. Thus, how native employment responded to those immigration changes would correctly estimate how immigrants affect natives’ employment opportunities–for a given local demand. The second purpose served by our instrument is to reduce the measurement error bias. As the measure of immigrants to other states is based on large national samples (excluding California), the instrument is also largely exempt from the measurement error affecting state-level measures of immigration by skill due to the potentially small size of cells1. Hence, we have an instrument for the inflow of immigrants that is potentially uncorrelated (or weakly correlated) with California-specific labor demand shocks and the measurement errors while still correlated with the supply-driven shocks to the composition of immigrants to California. This instrumental variables strategy that we adopt is novel for the ”area” approach and it is an improvement on the ”national” approach (e.g. Borjas, 2003) in that, while we cannot rely on some natural experiment, we have a more sophisticated way of isolating supply-driven variations of immigrants going beyond the simple inclusion of education-age, education-year and age-year specific effects (which we still include). The rest of the paper is organized as follows: Section 2 presents the data and shows some statistics on the skills of foreign-born and recent immigrants to California. Section 3 presents the production function used as framework to estimate the effect of immigration on wages. The skill-structure defined in the production function is used in the empirical estimations. Section 4 presents the identifying strategy and shows the migration and employment responses of California’s native workers to immigration for the period 1960-2004. Section 5 1Recent work by Aydemir and Borjas (2006) analyzes the role of measurement error in generating potential bias in the estimates of the impact of immigrants based on local data. estimates the substitutability between U.S. and foreign born workers within education-age groups. Section 6 uses the estimated parameters to calculate the effects of immigrants on wages of natives (by education) for the 1990-2004 period. Section 7 concludes. 2 Immigration to California: A Look at the Data The data we use are from the integrated public use microdata samples (IPUMS) of the U.S. decennial Census and of the American Community Survey (Ruggles et al, 2005). In particular, we use the general (1%) sample for Census 1960, the 1% state sample, Form 1, for Census 1970, the 5% state sample for the Censuses 1980 and 1990, the 5% Census sample for year 2000 and the 1/239 American Community Survey (ACS) Sample for the year 2004. As those are all weighted samples we use the variable “personal weight” to construct all the average and aggregate statistics relative to California. We consider people aged 17 to 66 not living in group quarters, and we included them among the workers if they worked at least one week in the previous year and earned a positive amount in salary income. When using wage data, we converted the current wages to constant wages