III. Wage Inequality and Labour Market Institutions

III. Wage Inequality and Labour Market Institutions F. Globalization, Impact of Immigration Plan 1. Globalization and De-industrialization 2. Changes in Immigration Flows and the Simple Model 3. Local Labour Market Studies 4. Economy-wide/National Studies a. Differences-in-Differences Analysis b. Imperfect Substitution across Skill Group

1. Globalization and De-industrialization Increased globalization arising from reductions in barriers to trade and reduced costs to international economic transactions is a popular culprit for rising inequality in the labour markers of industrialized countries. Increased trade with developing countries is viewed as a driving force behind the decline in manufacturing (de-industrialization) in the U.S. in particular and the move towards a service economy, which in turn is seen as a leading cause of poor labour market performance of less skilled workers (Wood, 1995). International capital mobility, reduced costs of technology transfer, and greater outsourcing opportunities may increase the elasticity of product demand facing workers, making their employment more sensitive to shocks. Prospects of foreign competition may erode the workers bargaining power, and reduce the extent to which internal labor markets (intra-firm markets) insulate them from product market and labor market shocks (e.g. Borjas and Ramey, 1995).

The number of immigrants relative to native-born workers has risen and in the U.S. in particular, these immigrants are increasingly low educated and are well placed to fill low-skill services jobs, such as cleaning, gardening, baby-sitting, etc. Effectively, immigration and trade have increased the effective labor supply of less skilled workers in the U.S., what are the distributional consequences? We have seen some empirical analyses of the impact of trade, where the tractability of the outsourcing of intermediate inputs limits the analysis. Now let s focus on immigration.

2. Changes in Immigration Flows and the Simple Model The simplest model of immigration assumes that immigrants and residents (or natives) are perfect substitutes in production, that is, they have the same types of skills and are competing for the same jobs. In this case, immigration shifts the supply curve rightwards. As a results the wage falls from w 0 to w 1, and total employment increases from N 0 to E 0, but at the lower wage w 0, there is a decline, from N 0 to N 1, in the number of residents who work. The assumption that resident workers and immigrants are perfect substitutes is questionable. It may be that immigrants and native workers are not competing for the same types of jobs. If immigrants and residents can complement each other in the labour market, an increase in the number of immigrants raises the marginal product of residents, shifting up the demand curve for native born workers. This lead to a higher native wage w0 w1 and an increase in the native employment N 0 N1.

Source: Borjas (1996)

Source: Borjas (1996)

This suggests a simple way of determining whether immigrants and natives are complements or substitutes in production. o If they are substitutes, the earnings of native workers should be lower if they reside in labour markets where immigrants are in abundant supply. o If they are complements, native earnings should be higher in those labour markets where immigrants tend to choose. Thus determining the distributional impact of immigration on wages hinges on findings which groups of natives and immigrants are substitute for each other and which groups of natives and immigrants are complements for each other. This will also have implications for wage inequality, if immigrants are largely substitutes for low-skilled workers but complements for high skilled workers, the effect of immigration would be to widen wage inequality. In both Canada and the United States, there have been large changes over the 20 th century of the source countries of immigrants, away from Western and Southern European countries, towards Eastern Europe and Asia for Canada, and Mexico and Latin America for the United States.

There has also been a change in the skill-mix of the immigrant populations, which varies by country. For example, more than a third of immigrants from India to the U.S. in 2000 held a professional or doctorate degree, while less then 2% of Mexican immigrants did. Immigrants also tend to concentrate in specific states/provinces, such as the large states of California, Texas, Florida and New York in the U.S. and Ontario in Canada. o 51% of immigrants to Canada now locate in the greater Toronto area Many western European countries, in particular Spain, Germany and the UK, have also switched from being sending countries to being countries receiving foreign (non-eu) immigrants. o But only Spain had, before the current crisis, a relatively high immigration rate.

Figure 17 Number of immigrants and immigration rate in Canada, 1900 to 2006 450,000 400,000 350,000 number of immigrants immigration rate (per 1,000) Number of immigrants Immigration rate 60 50 300,000 40 250,000 200,000 30 150,000 20 100,000 50,000 10 0 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 0 Sources: Statistics Canada, 2006, Report on the demographic situation in Canada 2003 and 2004, Statistics Canada Catalogue number 91-209-XIE; and Citizenship and Immigration Canada. Statistics Canada Catalogue number 91-003-X - 20 -

Changes in the source country of Canadian in Immigrants (from StatCan) Immigration to Canada by Source Region: 1955-2006 300,000 250,000 200,000 150,000 100,000 50,000 0 1955 1958 1961 1964 1967 1970 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 Britain USA Other europe Asia West Indies Other

Source: CIC

Source: Aydemir and Borjas (2006) 56 Canada: Figure 2. Trends in the immigrant/emigrant share for male workers, by education and country Immigrant share 0.4 0.3 0.2 0.1 Some college Post-graduate degrees High school graduates College graduates High school dropouts 0 1960 1970 1980 1990 2000 Year United States: 0.5 Immigrant share 0.4 0.3 0.2 College graduates High school dropouts Post-graduate degrees 0.1 Some 0 High school graduates college 1960 1970 1980 1990 2000 Year Mexico: 0.3 High school graduates Emigrant share 0.2 0.1 High school dropouts, 9-11 Some college High school dropouts, 0-8 0 College graduates 1960 1970 1980 1990 2000 Year Notes: The trend lines for Canada and the United States give the fraction of the workforce that is foreign-born; the trend line for Mexico gives the fraction of the Mexican workforce that emigrated to the United States.

Source: Bourbadat and Lemieux (2007) Canada Table 1(a): Sample Means, Males Cdn born 1981 2001 Cdn Immigrant born Immigrant Log weekly wage 6.66 6.73 6.66 6.64 Canadian experience 17.7 16.1 19.7 16.8 Foreign experience 6.4 6.4 Age 35.9 41.3 39.3 43.4 Schooling Less than HS 0.403 0.371 0.222 0.212 High School degree 0.215 0.130 0.247 0.189 Trade Certificate 0.159 0.204 0.171 0.139 Post-secondary 0.106 0.137 0.185 0.180 Bachelors' degree 0.075 0.084 0.118 0.157 Post-graduate 0.040 0.073 0.056 0.123 Years of schooling 11.8 12.2 13.5 14.0 Married 0.688 0.807 0.669 0.776 Language English only 0.617 0.804 0.634 0.816 French only 0.159 0.027 0.136 0.023 Bilingual 0.224 0.142 0.230 0.135 Neither fr. nor eng. 0.000 0.027 0.000 0.025 Mother tongue English 0.379 0.275 French 0.037 0.031 Country of Origin UK and US 0.258 0.147 FR,IT,GER,NET,POR,GRE 0.326 0.171 "USSR", POL, CZE 0.088 0.072 Other Europe 0.087 0.054 Asia 0.129 0.368 Africa 0.029 0.058 S-C America 0.071 0.119 Rest of world 0.011 0.010 CMA 0.509 0.772 0.616 0.894 Province Quebec 0.318 0.144 0.296 0.125 Ontario 0.360 0.557 0.368 0.583 Manitoba 0.046 0.032 0.043 0.026 Saskatchewan 0.042 0.013 0.037 0.008 Alberta 0.115 0.095 0.128 0.092 British Columbia 0.120 0.159 0.126 0.167 Number of Observations 82218 20678 124620 30615

3. Local Labour Market Studies Research on immigration has exploited the economic clustering of immigrants and use differences across labour markets to identify the impact of immigration. Comparing native earnings in cities with a high fraction of immigrants (NY, LA) with cities where immigrants are a relatively small fraction of the population. This type of analysis treats cities or regions as closed economies, which may be a strong assumption in many cases. For example, the estimating equation in Altonji and Card (1991) is Wˆ X b f c e where Wˆ Nj Nj Nj j is the adjusted labour outcomes of native workers N in city j, that is the average wage is purged from age and education effects across cities (i.e. are SMSA dummies from a first-stage regression) and f j is the fraction of immigrants in the city and e Nj is the error term. Nj Wˆ Nj

Issues with this type of analysis: a. endogenous location choice, immigrants endogenously cluster in cities with thriving economies b. inflows of immigrants may result in the outflows of native labor or capital not captured by partial equilibrium Solutions to the first problem: a. Instrumentation of the fraction of immigrants in the city with the stock of existing immigrants, the so-called enclave effect b. First-differencing eliminates city fixed effects. Using this combined strategy, Altonji and Card (1991) find that an increase of 0.01 in the fraction of immigrants reduce the wages of less-skilled natives by 1.2 percent. The use of the enclave instrument has been pursued in the recent literature using longer lags and country-specific flows. In the case of Mexican migrants, Munshi (2003) the use of rainfall in the origincommunity (collected from local weather stations) as an instrument for the size of the migrant network at the destination.

Source: Altonji and Card (1991) Table 7 Effects of Immigration on Four Groups of Less-Skilled Natives Pooled Sample (standard errors in parentheses) Cross sectional First-Dif ferenced 1970 1980 1980-1970 1980-1970 IV Outcome Variable: 1. Labor Force! Population -.173 (.086) -083 (.049).080 (.083) -.102 (.122) 2. Employment! Population -.240 (.074) -.054 (.060).404 (.097).085 (.144) Employment/ Labor Force.109 (.036).019 (.040).461 (.077).231 (.113) 4. Fraction Worked Last Year -.161 (.063) -.158 (.050).090 (.084) -.246 (.125) 5. Log Weeks Worked -.191 (.078) -.088 (.061).232 (.132).142 (.193) 6. Log Earnings! Week.467 (.165).018 (.112) -.262 (.228) -1.205 (.342) equations include the average education and age of the subgroup in the SMSA (with subgroup specific slopes and Intercepts), as well as total population in the SMSA. The sample size is 424. "Esti.ated by instrumental variables. The change in the fraction of immigrants in the SMSA is instrumented with the fraction of immigrants In 1970 and its square.

A second type of solution of the endogenous location problem uses quasi-natural experiments In an influential study, Card (1990) used the Mariel Boatlift as a case study of the impact of an exogenous shift in supply on a local labour market (Miami). With this study, Card introduced the differences-in-differences methodology comparing the outcomes (wages and unemployment) of various demographic groups in Miami and in a counterfactual city (made up of Altanta, Houston, Los Angeles and Tampa) before and after the boatlift. This study shows that the influx of Mariel immigrants had virtually no effect on the wage rates of less-skilled non-cuban workers and little effect on the employment rates. Why did immigration not have an effect in the Mariel experiment or at least not a detectable one? 1. Card argues that the ability of Miami s labour market to rapidly absorb the Mariel immigrants was largely owing to its adjustment to other large waves of immigrants in the two decades before the Mariel Boatlif.

Source: Angrist and Krueger (1999)

Source: Angrist and Krueger (1999)

Perhaps the posited 7% percent increase in the unskilled labour force was not actually that large: Only 50% of the 125,000 Cubans settle in Miami, 50% of those were women whose labour force participation rates were probably not higher than that of the natives. 2. Another possibility is that this is not a high-powered test. As shown in Angrist and Krueger's Figure 1, variables such as employment, unemployment, wage levels, bounce around from year to year, especially in small samples, so we may have less ability to detect small effects than we might assume. Angrist and Krueger perform an event study like Card s study using the Mariel Boat Lift that Didn t Happen, when in the summer of 1994, tens of thousands of Cubans were diverted to Guantanamo Bay by the Clinton administration. The data comes perilously close to showing a significant rise in the black unemployment rate in Miami by 6.3 percentage points (t = 1:70) [between 1993 and 1995, the black unemployment rate rose by 3.6 percentage points in Miami and fell by 2.7 percentage points in comparison cities). 3. A third possibility is a reduction in native inflows to Miami.

From 1970 to 1980, the Miami population grew at 2.5% per year while the rest of Florida grew at 3.9%. After April 1, 1980, the growth rate in Miami slowed to 1.4% per year, while the rest of the state declined to only 3.4%. Notably, Miami still received a disproportionate share of new immigrants in this time. Thus, it appears that natives and older immigrants may have been deterred from migrating to Miami. 4. Also, there may be other margins of adjustment such as endogenous technological choice (Beaudry and Green, 2003). In the presence of an abundant supply of less-skilled workers, Lewis (2004) has argued that that computer use was lower in Miami (23%) than in other comparison cities (+7%). 5. A final possibility is that Miami cannot realistically be treated as an autarkic labour market. This final criticism sparked a lively debate between proponents and critics of local area studies analyses as a proper setting to learn about the impact of immigration of labour markets.

Source: Angrist and Krueger (1999) The boatlift that never happenned (1994)

Source: Lewis (2004) Table 5 Explaining 1984 Computer Use in Miami and Comparison Cities (1) (2) (3) (4) (5) (6) (7) In Comparison City 0.047 0.023 0.003 0.010-0.013 0.010 0.009 (0.023) (0.028) (0.031) (0.002) (0.001) (0.005) (0.007) In Card (1990) Comp- 0.075 0.052 0.058 0.051 0.022 0.042 0.041 arison City (0.024) (0.028) (0.032) (0.002) (0.002) (0.003) (0.006) Pre-Mariel (1980) Share 19.079 11.715 7.832 8.954 Computer Ops, this MA (5.743) (4.836) (3.552) (3.706) Constant (Miami Average) 0.231 0.256 0.282 0.181-2.295-0.939-0.515 (0.022) (0.027) (0.030) (0.030) (0.224) (0.301) (0.260) Cubans Excluded? NO YES YES YES YES YES YES Other Hispanics Excluded? NO NO YES YES YES YES YES Ed, Age, Gender Controls? NO NO NO NO YES YES YES Occupation Dummies? NO NO NO NO NO YES YES Industry Dummies? NO NO NO NO NO NO YES R 2 0.002 0.001 0.003 0.005 0.107 0.330 0.379 N 8,592 8,451 7,634 7,634 7,634 7,634 7,634 Source: 1984 October CPS and 1980 Census of Population (PUMS). Standard errors in columns (4) - (7) take account error correlation among observations in the same city group (Miami and each of the two comparison groups). Cincinnati, Cleveland, Minneapolis, Rochester, Pittsburgh, Greensboro, Nassau-Suffolk, Riverside, Chicago. Atlanta, Houston, Tampa, Los Angeles. Fully interacted by gender: dummies for 1-18 years of education, quartic in age, and an interaction of years of education and quadratic in age. Results are robust to other specifications.

Other studies [Hunt (1992): repatriates from Algeria; Carrington and Lima (1996): Portuguese repatriates from Africa] of the impact of large and unexpected immigrants inflows in other countries have seemed to confirm the absence of adverse effects on local labour market conditions. These studies provide examples of the differences-in-differences methodology: measuring the impact of immigration by comparing what happened in the treated local labour market with an untreated local labour market. The pre-2000 analyses lead Friedberg and Hunt (1995) to conclude There is no evidence of economically significant reduction in native employment. Most empirical analysis of the United States and other countries finds that a 10 percent increase in the fraction of immigrants in the population reduces native wages by at most 1%. (Page 42) 4. Economy-wide/National Studies Can we find sufficient variation in immigrants supply shocks in a national/general equilibrium context to identify the impact of labour market shocks by directly observing how those shocks affect some workers and not others?

a. Differences-in-Differences Analysis An example of a difference-in-difference analysis which considered the case with a larger impact on a national labour market is that of Friedberg (2001). Rachel Friedberg studies the impact of the massive immigration from the former Soviet Union: close to one million Russian immigrants, relatively high skilled, came to Israel between 1989 and 1990, increasing the 12 percent. She uses an approach that combines use of a natural experiment with a novel instrumental variable which exploits detailed data on immigrants occupations in their country of origin. Cross-occupation variation is thought not to be as sensitive to FPE (factor-prize equalization) than cross-city variation, since occupational mobility is more restricted and often requires a large investment in retraining.

Source: Friedberg (2001) THE IMPACT OF MASS MIGRATION eb s\ sb 85 sb sk s> EL EL sb 9i 32 95 9; 95 Year FIGURE I Immigration to Israel Note: Number of immigrants, including immigrating citizens, per month. Sources are Bank of Israel [I9991 and Israeli Central Bureau of Statistics [19971.

Source: Friedberg (2001) 1390 QUARTERLY JOURNAL OF ECONOMICS 30 18 0 1 2.5 98 93 55 80 66 WLS 97 92 FIGURE IV Israeli Wages and the Presence of Russians in the Occupation in Israel Note: W is the average log wage of Israelis in 1994. r is the ratio of Russians to Israelis in the occupation in 1994.

Source: Friedberg (2001) 1386 QUARTERLY JOURNAL OF ECONOMICS i housemal Other se locksmit electric - ~therma C d Workers social S pboprzet supervis iurlsts 0 I 0 transpdr I 2 ln(p) FIGURE I11 Number of Russians in the Occupation in Israel and Abroad Note: ln(r) is the log of the number of Russians in the occupation in Israel in 1994.1n(P) is the log of the number of Russians formerly in the occupation abroad. I 4 1 6

Source: Friedberg (2001) 1398 QUARTERLY JOURNAL OF ECONOMICS TABLE IV THEEFFECTOF IMMIGRATION NATIVEISRMLIWAGES: ALTERNATIVE S.MPLESAND SPECIFICATIONS First OLS stage 2SLS By skill group: Low skill -.799.352 6.08 (.241) (.147) (16.6) High skill,314.309.265 (.300) (.082) (.383) By age group: 18-34 -.299,141.913 (.129) (.043) (326) 35-65 -.359,200,670 (.098) (.028) (.370) Numerical immigration variable:.000583,267.00130 (.00022) (.0823) (.00263) Relative dependent variable: -,532,241.387 (.287) (.087) (1.44) The dependent variable is the change m the log wage of Israelis 1989-1994. "Low-skill" and "high-skill" occupations refer to those with average worker schooling in 1989of less than or equal to fourteen years and greater than fourteen years, respectively. Age-group regressions use indi\.idual-level data, with the same set of control variables as in Table 111. The "numerical immigration variable" specifications replace the ratios r andp with the numbers R and P. The "relative dependent variable" specifications use wage growth relative to U. S. occupational wage growth over the same period as the dependent variable. Skill-group, numericalregressor, and relative dependent variable regressions use occupation-level data. Standard errors are in parentheses.

So now the average wage of natives by occupation j (instead of cities) are the unit of observation where time t. r R / jt jt N jt W jt X t jt t r jt e is the ratio of Russian immigrants to Natives in occupation j at jt A potential endogeneity problem arises from the immigrants s choice of occupations: as in Altonji and Card (1991), this is solved with a time differencing (of order k) and an instrumental variables strategy, which uses the immigrant s occupation in Russia. Friedberg find very different OLS and IV estimates. The OLS estimates are negative, while the IV estimates are insignificant or positive! We would normally expect the opposite: immigrants would tend to endogenously select into occupations with growing wages and employment. o This simultaneity would attenuate the estimated negative impact of immigration on native wages or employment. By this logic, the IV estimates should be more negative than OLS estimates.

She appeals to anecdotal evidence of complementarity between Russian and Isreali doctors to explain the positive impact. But this is not totally convincing. Occupations are not labour markets, and depending upon elasticities of substitution/complementarity, it is not entirely clear how flows into one occupation should affect wages in another. George Borjas is unconvinced and writes the paper The Labor Demand Curve is Downward Sloping b. Imperfect Substitution across Skill Groups Card and Lemieux (2001) introduced a nested CES production function with imperfect substitution across age-groups to estimate the impact of relative supplies on relative wages.

Borjas (2003) assumes that workers with the same education but different levels of experience are imperfect substitutes and distinguishes supply shocks between immigrants and natives by level of experience. Thus he implicitly assumes that within the same education-experience (effective experience) group, immigrants and natives are perfect substitutes. Based on this idea, Borjas conducts an analysis of the impact of immigration on native earnings in cells defined by decade (1960-2000), education (4 groups), and 5-year-potential-experience groups. Hence, there are 5 4 8 = 160 cells. Considering workers who have educational attainment i, experience level j, and are observed in calendar year t, he defines the measure of the immigrant supply shock for this skill group by M ijt p ijt M N, where M ijt gives the number of immigrants in cell (i j,t), and N ijt gives the corresponding number of natives. ijt ijt

His basic estimating reduced-form model is y p s x ( s x ) ( s ) ( x ) it ijt i j t i where s i is a vector of fixed effects indicating the group s educational attainment, x j is a vector of fixed effects indicating the group s work experience, and x j is a vector of fixed effects indicating the time period. So that the interactions ( si x j ) allows different experience profile by schooling level s ) allows the returns to schooling to vary over time ( i t ( x j t ) ( si x j i t allows the returns to experience to vary over time ) is omitted, that is the changes over time in the returns to experience by schooling level are assumed uncorrelated with the immigrant supplies. j i t j t ijt The labor market impact of immigration is identified using time-variation within education-experience cells.

Source: Borjas (2006) 59 Figure 1. The share of immigrants in the workforce High school dropouts High school graduates Fraction 0.1.2.3.4.5 0.1.2.3.4.5 Some college College graduates 0 10 20 30 40 0 10 20 30 40 Labor market experience 1960 1970 1980 1990 2000

If the returns to experience is increasing differentially over time, say the older not improving as much as the young, this may be a source of spurious correlation. Table III coefficients imply that a 10 percent increase in immigrant labor supply reduces native weekly earnings by 4.0 log points (0.572 0.7) [the conversion factor for the relative share p.1349], reduces the fraction of time worked by 3.7 log points, and reduces total native earnings by 6.4 log points. Notice also that the effect on log annual earnings is the effect on log weekly earnings plus the effect on log weeks. Thus, a 10 percent increase in immigrant labour supply must reduce native weeks worked by 6.4 4.0 = 2.4 log points. This is a large disemployment effect. This result also suggests that immigrant labor supply could also potentially be endogenous. Footnote 8 discusses this concern and gives results when immigrant LF participation is instrumented by immigrant population shares in the relevant education/experience groups. Borjas finds significant negative effects of the immigrant share in the skill group on weekly earnings, an effect that translates into a wage elasticity of -0.4.

Source: Borjas (2003) 60 A. Blacks Figure 2. Trends in employment rates, by race, education and experience High school dropouts High school graduates Fraction.5.6.7.8.9 1.5.6.7.8.9 1 Some college College graduates 0 10 20 30 40 0 10 20 30 40 Labor market experience 1960 1970 1980 1990 2000 B. Whites High school dropouts High school graduates Fraction.5.6.7.8.9 1.5.6.7.8.9 1 Some college College graduates 0 10 20 30 40 0 10 20 30 40 Labor market experience 1960 1970 1980 1990 2000

Source: Borjas (2003) 1348 QUARTERLY JOURNAL OF ECONOMICS TABLE III IMPACT OF IMMIGRANT SHARE ON LABOR MARKET OUTCOMES OF NATIVE EDUCATION-EXPERIENCE GROUPS Dependent variable Speci cation: Log annual earnings Log weekly earnings Fraction of time worked 1. Basic estimates 20.919 20.572 20.529 (0.582) (0.162) (0.132) 2. Unweighted regression 20.725 20.546 20.382 (0.463) (0.141) (0.103) 3. Includes women in labor force counts 20.919 20.637 20.511 (0.661) (0.159) (0.148) 4. Includes log native labor force as regressor 21.231 20.552 20.567 (0.384) (0.204) (0.116) The table reports the coef cient of the immigrant share variable from regressions where the dependent variable is the mean labor market outcome for a native education-experience group at a particular point in time. Standard errors are reported in parentheses and are adjusted for clustering within education-experience cells. All regressions have 160 observations and, except for those reported in row 2, are weighted by the sample size of the education-experience-periodcell. All regression models include education, experience,and period xed effects, as well as interactions between education and experience xed effects, education and period xed effects, and experience and period xed effects.

For greater comparability with local labor market studies, he also investigates a specification by state and finds a coefficient of -0.13 for the immigrant share, a much more believable result! o He argues that interstate flows of labour and capital tend to equalize opportunities for workers of given skills across regions. o Given the presence of substantial measurement error in the state education experience time cells, however, it would have been valuable if Borjas had instrumented labour supply with population shares as above. o As estimated, we cannot tell if the area studies estimates are much smaller because of arbitrage or attenuation. He also adjusts the immigrants experience given a weight of 0.4 to foreign experience and 1.6 to domestic experience. However, Borjas analysis of the short run in this and other papers uses the total immigration over a 20-year period on an economy that does not adjust its capital.

Source: Borjas (2003) LABOR MARKET IMPACT OF IMMIGRATION 1351 TABLE IV IMPACT OF IMMIGRANT SHARE ON NATIVE LABOR MARKET OUTCOMES, BY EDUCATION GROUP Dependent variable: High school dropouts High school graduates Some college College graduates At least high school graduates 1. Log annual earnings 21.416 22.225 20.567 1.134 21.184 (0.313) (0.622) (0.421) (0.436) (0.668) 2. Log weekly earnings 20.947 22.074 21.096 0.610 20.335 (0.164) (0.510) (0.461) (0.440) (0.612) 3. Fraction of time worked 20.086 0.393 0.567 0.300 21.040 (0.073) (0.251) (0.385) (0.499) (0.211) The table reports the coef cient of the immigrant share variable from regressions where the dependent variable is the mean labor market outcome for a native education-experience group at a particular point in time. Standard errors are reported in parentheses and are adjusted for clustering within experience cell (in the rst four columns) and within education-experience cells (in the last column). All regressions are weighted by the sample size of the education-experience-periodcell. The regressions reported in the rst four columns have 40 observations and include experienceand period xed effects. The regressions reported in the last column have 120 observations and include education, experience, and period xed effects, as well as interactions between education and experience xed effects, education and period xed effects, and experience and period xed effects.

Source: Borjas (2003) LABOR MARKET IMPACT OF IMMIGRATION TABLE V IMPACT OF IMMIGRANT SHARE ON LABOR MARKET OUTCOMES OF NATIVE STATE-EDUCATION-EXPERIENCE GROUPS 1353 Dependent variable: (1) (2) (3) (4) 1. Log annual earnings 20.115 20.276 20.253 20.217 (0.079) (0.053) (0.046) (0.068) 2. Log weekly earnings 20.124 20.217 20.203 20.183 (0.042) (0.039) (0.038) (0.050) 3. Fraction of time worked 20.038 20.100 20.078 20.119 (0.030) (0.015) (0.015) (0.021) Controls for: (State 3 period), (education 3 period), (experience 3 period), (state 3 education) xed effects Yes Yes Yes Yes (State 3 education 3 experience) xed effects No Yes Yes Yes (Education 3 experience 3 period) xed effects No No Yes Yes (State 3 education 3 period), (state 3 experience 3 period) xed effects No No No Yes The table reports the coef cient of the immigrant share variable from regressions where the dependent variable is the mean labor market outcome for a native state-education-experiencegroup at a particular point in time. Standard errors are reported in parentheses and are adjusted for clustering within state-educationexperiencecells. All regressions are weighted by the sample size of the state-education-experience-periodcell and include state, education, experience, and period xed effects. The regressions on log annual earnings or log weekly earnings have 8153 observations; the regressions on the fraction of time worked have 8159 observations.

Finally, Borjas perform some structural estimation using a 3-level nested CES where the marginal productivity condition implies that the wage for skill group (i, j, t) is ln w ijt ln Lt (1 KL ) ln Q t ( KL E )ln L t ln it ( E X ) ln L it ln ij ( X 1) ln L ijt where the subscripts are used to identify the expression for (1/1- ) with the relevant subscript. Assuming that the first three terms can be absorbed by period fixed effects, the fourth and fifth terms by interactions between education fixed effects and period fixed effets, and the sixth term by interactions between education fixed effects and experience fixed effects, then elasticity of substitution across experience groups is estimated with 1 ln wijt t it ij ln Lijt The elasticity of substitution across education groups is estimated following Katz and Murphy (1992) is found to be 1.3, the elasticity of substitution between capital and labour is assumed to be equal to 1 and simulations are performed using these estimates. X

LABOR MARKET IMPACT OF IMMIGRATION Source: Borjas (2003) TABLE IX WAGE CONSEQUENCES OF IMMIGRANT INFLUX OF THE 1980S AND 1990S (PREDICTED CHANGE IN LOG WEEKLY WAGE) Education 1369 Years of experience High school dropouts High school graduates Some college College graduates All workers 1 5 20.065 20.021 0.004 20.035 20.024 6 10 20.101 20.027 0.001 20.042 20.029 11 15 20.128 20.036 20.009 20.059 20.041 16 20 20.136 20.033 20.011 20.055 20.039 21 25 20.108 20.025 20.008 20.049 20.033 26 30 20.087 20.023 0.000 20.049 20.029 31 35 20.066 20.022 0.001 20.050 20.027 36 40 20.044 20.013 0.008 20.056 20.022 All workers 20.089 20.026 20.003 20.049 20.032 The simulation uses the factor price elasticities reported in Table VIII to predict the wage effects of the immigrant in ux that arrived between 1980 and 2000. The calculations assume that the capital stock is constant. The variable measuring the group-speci c immigrant supply shock is de ned as the number of immigrants arriving between 1980 and 2000 divided by a baseline population equal to the average size of the native workforce (over1980 2000) plus the number of immigrants in 1980. The last column and the last row report weighted averages, where the weight is the size of the native workforce in 2000.

As shown in Table 9, using national data by Borjas (2003) also finds significant negative wage impact of immigrants (1980-2000) on less educated natives (-9%) and on average wages (-3%) in the short run. Aydemir and Borjas (2006) perform a similar analysis for Canada and find a similar average wage response to international migration. They also argue that the smaller estimates from the local labor market studies and from national studies using public use files (e.g. Bohn and Sanders (2005) for Canada and hand, Bonin (2005) for Germany) may be due to the attenuation bias coming from smaller sample size In contrast, the national labor market evidence indicated that wage growth was strongly and inversely related to immigrant- induced supply increases. But because Canada and the United Stated received a different mix of skilled immigrants, Aydemir and Borjas (2006) find that international migration has narrowed wage inequality in Canada and increased wage inequality in the United States.

Source: Aydemir and Borjas (2006) Figure 3 The immigrant supply shock in Canada, 1971-2001 A. High school dropouts B. Either high school or vocational degree C. Both high school and vocational degree 0.25 0.3 0.45 1971 Immigrant share 0.2 0.15 0.1 1986 2001 1971 Immigrant share 0.25 0.2 0.15 0.1 1986 1971 2001 Immigrant share 0.35 0.25 0.15 1986 2001 0.05 0 10 20 30 40 Years of experience 0.05 0 10 20 30 40 Years of experience 0.05 0 10 20 30 40 Years of experience D. Bachelor's degree E. Post-graduate degree 0.4 0.45 2001 Immigrant share 0.325 0.25 0.175 1986 2001 1971 Immigrant share 0.35 0.25 1986 1971 0.1 0 10 20 30 40 Years of experience 0.15 0 10 20 30 40 Years of experience Note: The immigrant share gives the fraction of the workforce that is foreign-born in a particular education-experience group.

Source: Aydemir and Borjas (2006) 61 Table 1. Relation between the immigrant/emigrant share and labor market outcomes. Log annual earnings Earnings outcomes Log weekly earnings Log monthly earnings Employment outcomes Fraction of weeks worked Labor force participation rate Weighted Regressions 1. Canada -0.617-0.507 --- -0.241 --- (0.246) (0.202) (0.108) 2. United States -0.845-0.489 --- -0.345 --- (0.472) (0.223) (0.075) Mexico 3. All workers --- --- 0.798 --- 0.058 (0.443) (0.044) 4. All workers, 1990-2000 --- --- 0.841 --- 0.062 (0.540) (0.048) 5. Urban workers --- --- 0.652 --- 0.065 (0.419) (0.055) Notes: Standard errors are reported in parentheses and are adjusted for clustering within education-experience cells. All coefficients are obtained from regressions weighted by the sample size used to compute the dependent variable. For Canada and the United States, the table reports the coefficient of the immigrant share variable from regressions where the dependent variable is the mean labor market outcome of native-born persons in an education-experience group at a particular point in time. For Mexico, the table reports the coefficient of the emigrant share variable from regressions where the dependent variable is the mean labor market outcome of Mexican stayers in an educationexperience group at a particular point in time. The regressions estimated in Canada have 240 observations; the regressions estimated in the United States have 200 observations; the wage regressions estimated in Mexico have 160 observations in rows 3 and 5, and 80 observations in row 4; and the labor force participation regressions estimated in Mexico have 120 observations in rows 3 and 5, and 80 observations in row 4. All regression models include education, experience, and period fixed effects, as well as interactions between education and experience fixed effects, education and period fixed effects, and experience and period fixed effects.

Source: Aydemir and Borjas (2006) 47 Table 2. Estimated wage impact of immigration, national-level analysis Stat. Can. 5/100 PUMF 1/100 1/1000 1/10000 Canada: 1. ˆβ -0.507-0.468-0.403-0.342-0.076-0.011 2. Standard error of ˆβ 0.202 0.196 0.189 0.180 0.191 0.200 3. Standard deviation of ˆβ --- 0.056 0.099 0.119 0.174 0.174 4. R 2 of auxiliary regression 0.967 0.965 0.960 0.953 0.845 0.590 Using large sample share * 5. β --- -0.505-0.501-0.499-0.466-0.384 * 6. Standard error of β --- 0.209 0.226 0.241 0.485 1.475 * 7. Standard deviation of β --- 0.049 0.093 0.126 0.405 1.353 Corrected coefficients: 8. Using mean immigrant share -0.520-0.531-0.524-0.638 1.174 0.044 9. Standard deviation of row 8 --- 0.064 0.132 0.241 15.647 1.652 10. Using mean of binomial error -0.521-0.538-0.590-0.689 0.192-0.400 11. Standard deviation of row 10 --- 0.065 0.151 0.265 7.693 14.139 12. USSIV method -0.519-0.515-0.520-0.525 0.482-0.486 13. Standard deviation of row 12 0.034 0.010 0.211 0.304 18.460 25.475 United States: 1. ˆβ --- -0.489 --- -0.476-0.347-0.082 2. Standard error of ˆβ --- 0.223 --- 0.225 0.247 0.279 3. Standard deviation of ˆβ --- --- --- 0.056 0.162 0.227 4. R 2 of auxiliary regression --- 0.974 --- 0.973 0.964 0.883 Using large sample share * 5. β --- --- --- -0.488-0.497-0.498 * 6. Standard error of β --- --- --- 0.228 0.291 0.631 * 7. Standard deviation of β --- --- --- 0.045 0.171 0.534 Corrected coefficients: 8. Using mean immigrant share --- -0.496 --- -0.506-0.642 5.794 9. Standard deviation of row 8 --- --- 0.060 0.320 89.464 10. Using mean of binomial error --- -0.496 --- -0.507-0.658-0.287 11. Standard deviation of row 10 --- --- --- 0.060 0.331 9.328 12. USSIV method --- -0.496 --- -0.496-0.503-5.420 13. Standard deviation of row 12 --- 0.040 --- 0.090 0.371 48.076 Notes: The coefficient ˆβ * gives the estimated wage impact of immigration; β gives the coefficient when the observed immigrant share is replaced by the immigrant share calculated from the largest file (i.e., the Statistics Canada file or the 5/100 U.S. Census). The R 2 of the auxiliary regression gives the multiple correlation of the regression of the immigrant share on all other explanatory variables in the model. The corrected coefficients use the methods described in the text to net out the impact of sampling error on ˆβ. All statistics reported in the table, except those referring to the Statistics Canada file and the 5/100 U.S. Census, are averages across 500 replications of random samples at the given sampling rate. The analysis of the Canadian labor market has 240 cells; the analysis of the U.S. labor market has 200 cells.

In Mexico, emigration rates are highest in the middle of the skill distribution and lowest at the extremes. So international migration has greatly increased relative wages in the middle of the Mexican skill distribution and lowered the relative wages at the extremes, thus decreasing wage inequality. Ottaviano and Peri (2006) use an approach similar to Borjas (2003) but accounts for actual year to year capital adjustment and different productive specialization (i.e. skills complementarities ) of native and immigrants. They finds much smaller wage loss for workers with no high school degree (-2% in the short run and -1% in the long run) and a positive average wage effect of immigrants on average native wages in the short and in the long run (+0.7 to +1.8%) as well as on workers with high school degree (+1% in the short run and +2% in the long run). Borjas himself gets different results showing a much smaller wage effect when he takes capital stock changes into account o Borjas and Katz (2007) immigration has a negative impact on high school graduates equal to -2.2% not accounting for capital adjustment and a positive impact of +1.1% (Table 11).

However, Ottaviano and Peri (2006) agree that the positive effect is unevenly distributed, so medium and highly educated natives gain more from immigration relative to less educated.

Basic readings: Borjas, G., R. Freeman, and L.Katz, How Much Do Immigration and Trade Affect Labor Market Outcomes?" Brookings Papers on Economic Activity, (1997) no. 1, 1-90. Borjas, G. The Labor Demand Curve Is Downward Sloping: Reexamining the Impact of Immigration on the Labor Market," Quarterly Journal of Economics 118 (November 2003), 1335-7. Friedberg, R. The Impact of Mass Migration on the Israeli Labor Market, Quarterly Journal of Economics, 2001, p. 1373-1408.