3 question

THE WAGE IMPACT OF THE MARIELITOS:

A REAPPRAISAL

GEORGE J. BORJAS*

This article brings a new perspective to the analysis of the wage effects of the Mariel boatlift crisis, in which an estimated 125,000 Cuban refugees migrated to Florida between April and October, 1980. The author revisits the question of wage impacts from such a supply shock, drawing on the cumulative insights of research on the economic impact of immigration. That literature shows that the wage impact must be measured by carefully matching the skills of the immi- grants with those of the incumbent workforce. Given that at least 60% of the Marielitos were high school dropouts, this article specifically examines the wage impact for this low-skill group. This analysis over- turns the prior finding that the Mariel boatlift did not affect Miami’s wage structure. The wage of high school dropouts in Miami dropped dramatically, by 10 to 30%, suggesting an elasticity of wages with respect to the number of workers between 20.5 and 21.5.

T he study of how immigration affects labor market conditions has beena central concern in labor economics for nearly three decades. The sig- nificance of the question arises because of the policy issues involved and because how labor markets respond to supply shocks can teach us much about how labor markets work. In an important sense, examining how immigration affects the wage structure confronts directly one of the funda- mental questions in economics: What makes prices go up and down?

David Card’s (1990) classic study of the impact of the Mariel boatlift sup- ply shock stands as a landmark in this literature. On April 20, 1980, Fidel Castro declared that Cuban nationals wishing to move to the United States could leave freely from the port of Mariel, and around 125,000 Cubans

*GEORGE J. BORJAS is the Robert W. Scrivner Professor of Economics and Social Policy at the Harvard Kennedy School, a Research Associate at the National Bureau of Economic Research (NBER), and a Program Coordinator at the Institute for the Study of Labor (IZA). I am particularly grateful to Alberto Abadie and Larry Katz for very productive discussions and for many valuable comments and suggestions. I also benefited from the reactions and advice of Josh Angrist, Fran Blau, Brian Cadena, Kirk Doran, Richard Freeman, Daniel Hamermesh, Gordon Hanson, Larry Kahn, Alan Krueger, Joan Llull, Joan Monras, Panu Poutvaara, Jason Richwine, Kevin Shih, Marta Tienda, Steve Trejo, and two referees. Additional results and copies of the computer programs used to generate the results presented in this article are available from the author at [email protected].

KEYWORDs: immigration, wage inequality, local labor market conditions, employment effects of migration/immigration, wage determination model

ILR Review, 70(5), October 2017, pp. 1077–1110 DOI: 10.1177/0019793917692945. � The Author(s) 2017

Journal website: journals.sagepub.com/home/ilr Reprints and permissions: sagepub.com/journalsPermissions.nav

quickly accepted the offer. The Card study was one of the pioneering attempts to exploit the insight that a careful study of natural experiments, such as the exogenous supply shock stimulated by Castro’s seemingly ran- dom decision to let people go, can help identify parameters of great eco- nomic interest. In particular, the Mariel supply shock would let us measure the wage elasticity that shows how the wage of native workers responds to an exogenous increase in supply.

Card’s analysis of the Miami labor market, and of comparable labor mar- kets that served as a control group or placebo, indicated that nothing much happened to Miami despite the very large number of Marielitos. Native wages did not go down in the short run as would have been predicted by the textbook model of a competitive labor market. Card’s study has been extremely influential, both in terms of its prominent role in policy discus- sions and its methodological approach.1

During the 1980s and 1990s, a parallel (but non-experimental) literature attempted to estimate the labor market impact of immigration by essentially correlating wages and immigration across cities (Grossman 1982; Altonji and Card 1991). These spatial correlations are problematic for two reasons: 1) immigrants are more likely to settle in high-wage cities, so that the endo- geneity of supply shocks induces a spurious positive correlation between immigration and wages; and 2) native workers and firms respond to supply shocks by resettling in areas that offer better opportunities, effectively diffus- ing the impact of immigration across the national labor market.2

Card’s Mariel study is impervious to both of these criticisms. The fact that the Marielitos settled in Miami probably had little to do with pre-existing wage opportunities, and much to do with the fact that Castro suddenly decided to allow the boatlift to occur and that many of the Cuban Americans who organized the flotilla lived in South Florida.3 Similarly, the short-run nature of Card’s empirical exercise, looking at the impact of immigration just a few years after the supply shock, means that we should be measuring the short-run elasticity, an elasticity that is not yet

1Many other studies examine exogenous supply shocks and are clearly influenced by the Card frame- work; see Hunt (1992), Carrington and de Lima (1996), Friedberg (2001), Angrist and Kugler (2003), Saiz (2003), Borjas and Doran (2012), Glitz (2012), Pinotti, Gazzè, Fasani, and Tonello (2013), and Dustmann, Schoänberg, and Stuhler (2017).

2Beginning with Altonji and Card (1991), the endogenous settlement of immigrants across localities has been addressed by using the geographic sorting of earlier waves of immigrants to predict the sorting of the new arrivals. The validity of this instrument, however, depends on whether economic conditions in local labor markets persist over time; see Jaeger, Ruist, and Stuhler (2016).

3The geographic clustering of the large Cuban community in Miami probably had little to do with the relative economic opportunities offered by the Miami labor market, and much to do with the fact that the flights operated by Pan American Airways that carried almost all of the early refugees out of Cuba had a single destination, the Miami airport. The 1980 census indicates that 50% of Cuban immigrants lived in the Miami metropolitan area. Both the 1990 and 2000 censuses report that more than 60% of the Cuban immigrants who likely were part of the Mariel influx still resided in the Miami metropolitan area.

1078 ILR REVIEW

contaminated by labor market adjustments and that economic theory pre- dicts to be negative.

This article provides a reappraisal of the evidence of how the Miami labor market responded to the influx of Marielitos. The article is not a replication of earlier studies. Instead, I approach and examine these questions from a fresh perspective, building on what we have learned from the 30 years of research on the labor market impact of immigration. One crucial insight from this research is that any credible attempt to measure the impact must carefully match the skills of the immigrants with the skills of the incumbent workforce. Borjas (2003), in the study that introduced the approach of cor- relating wages and immigration across skill groups in the national labor market, found a significant negative correlation between the wage growth of specific skill groups, defined by education and age, and the size of the immigration-induced supply shock into those groups.

The Marielitos were disproportionately low-skill workers; approximately 60% were high school dropouts and only 10% were college graduates. At the time, about a quarter of Miami’s existing workers lacked a high school diploma. As a result, even though the Mariel supply shock increased the number of workers in Miami by 8%, it increased the number of high school dropouts by almost 20%.

The unbalanced nature of this supply shock suggests that we should look at what happened to the wage of high school dropouts in Miami before and after Mariel. Remarkably, this trivial comparison was not made in Card’s (1990) study and, to the best of my knowledge, had not been conducted before the first draft of this article was released.4 By focusing on this very specific skill group, we obtain an entirely new perspective of how the Miami labor market responded to an exogenous supply shock.

Data

The migration of large numbers of Cubans to the United States began shortly after Fidel Castro’s communist takeover on January 1, 1959. By the year 2010, more than 1.3 million Cubans had emigrated.

4Card (1990, table 7) reported labor market outcomes for the subsample of black high school drop- outs but does not report any other pre- or post-Mariel differences for the least-educated workers. Monras (2014) examined wage trends in the sample of workers who have at most a high school diploma; his evi- dence is suggestive of the findings reported in this article. Finally, three months after this study was released as an NBER working paper (Borjas 2015), Peri and Yasenov (2015) re-examined the data for high school dropouts and concluded that Card’s results stood the test of time. The Peri and Yasenov study, however, looked at wage trends that were contaminated by various sampling decisions. They exam- ined the pooled earnings of men and women, but ignored the changing gender composition of the workforce; included non-Cuban Hispanics in the analysis, even though many of those Hispanics were immigrants who arrived after 1980; and included teenagers aged 16 to 18, even though most of these teenagers were enrolled in high school and were classified as high school dropouts because they did not yet have a diploma. Borjas (2016) documented the biases created by these methodological choices.

WAGE IMPACT OF THE MARIELITOS 1079

The first large-scale data set that precisely identifies an immigrant’s year of arrival is the 2000 decennial census. Prior to 2000, the census microdata reported the year of arrival in intervals (e.g., 1960–1964). I merged data from various censuses and the American Community Surveys (ACS) to construct a mortality-adjusted number of Cuban immigrants for each arrival year between 1955 and 2010.5 For example, I used the 1970 census to estimate the number of Cubans who arrived in the United States between 1960 and 1964, and then used the detailed year-of-migration information in the 2000 census to allocate those early immigrants to specific years within the five-year band. Figure 1 shows the trend in the number of Cubans migrating to the United States.

Several patterns emerge from the time series. First, it is easy to see the immediate impact of the communist takeover of the island. In 1958, only 8,000 Cubans migrated to the United States. By 1961 and 1962, 52,000 Cubans were migrating annually.6 The Cuban Missile Crisis abruptly stopped the flow in October 1962, and it took several years for other escape routes to open up. By the late 1960s, the number of Cubans moving to the United States was again near the level reached before the Missile Crisis.

Figure 1. Number of Cuban Immigrants, by Year of Migration, 1955–2010

Notes: The specific year of migration (through 1999) is first reported in the 2000 census. Counts are adjusted for mortality and out-migration by using information on the number of arrivals provided by the 1970 through 1990 censuses; see the text for details. The 2000–2008 counts are drawn from the pooled 2009–2011 American Community Surveys (ACS), and the 2009–2010 counts are drawn from the 2012 ACS.

5In principle, the calculation also adjusts for potential out-migration by Cuban immigrants. I suspect, however, that the number of Cubans who chose to return is trivially small (although a larger number might have migrated elsewhere).

6Full disclosure: I am a data point in the 1962 flow.

1080 ILR REVIEW

The huge spike in 1980, of course, is the Mariel boatlift supply shock. Between 1978 and 1980, the number of new Cuban immigrants increased 17-fold, from 6,500 to 110,000. The figure shows yet another spike in 1994 and 1995, coinciding with the period of what Angrist and Krueger (1999) erroneously labeled the ‘‘Mariel Boatlift That Did Not Happen.’’7 The cen- sus data clearly indicate that somehow the phantom Cubans from that boat- lift ended up in the United States, making this supply shock a Little Mariel. Although the number of Little Marielitos pales in comparison to the number of actual Marielitos, it is still quite large; the number of migrants arriving in 1995 was similar to that of the early Cuban waves in the 1960s. A steady increase in the number of Cuban migrants since the early 1980s is also evi- dent. By 2010, about 40,000 Cubans were arriving annually.

One last detail is worth noting. A disproportionately large number of the Marielitos ended up residing in the Miami metropolitan area: 62.6% resided in Miami in 1990 and 63.4% still resided there in 2000.

The empirical analysis initially uses the 1977 to 1993 March Supplements of the Current Population Surveys (CPS).8 These surveys report a person’s annual wage and salary income as well as the number of weeks worked in the previous calendar year. Much of the analysis will be restricted to men aged 25 to 59, who report positive values for wage and salary income, weeks worked, and usual hours worked.9 The age restriction implies that a worker’s observed earnings are unaffected by transitory fluctuations that occur during the transi- tions from school to work and from work to retirement. Similarly, the restric- tion to working men ensures that wage trends are not distorted by the entry of large numbers of women into the workforce in the 1970s and 1980s.10

The 1977 to 1993 period analyzed throughout much of the article is selected for two reasons. First, although the March CPS data files are avail- able since 1962, the Miami metropolitan area can be consistently identified only in the 1973 to 2004 surveys. Beginning with the 1977 survey, the CPS began to identify 44 metropolitan areas (including Miami) that can be used in the empirical analysis.11 Second, my analysis of wage trends stops with

7In 1994, Castro again toyed with the idea of permitting a boatlift, but the United States preemptively responded and rerouted many of the potential migrants to the military base in Guantanamo. Angrist and Krueger (1999), writing before the release of the 2000 census, argued that this seemingly phantom flow had an adverse effect on Miami’s black unemployment rate, raising questions about how to inter- pret the evidence from the Mariel boatlift that did happen. The 2000 census indicates that many of the rerouted immigrants eventually made their way to the United States.

8The March surveys are known as the Annual Social and Economic Supplements (ASEC). The data were downloaded from the Integrated Public Use Microdata Series (IPUMS) website on May 9, 2016.

9I also exclude persons who reside in group quarters, have a negative sample weight, or have outlying hourly wage rates (less than $1.50 or more than $40 in 1980 dollars). This last restriction roughly excludes workers in the top and bottom 1% of the earnings distribution.

10The sensitivity of the evidence to including working women is discussed below. 11The Miami-Hialeah metropolitan area is not identified at all before 1973 and is combined with the

Fort Lauderdale metropolitan area after 2004. The 1973–1976 surveys identify only 34 metropolitan areas, and one of them (New York City) is not consistently defined throughout the period; the Nassau- Suffolk metropolitan area is pooled with the New York City metro area in 1976.

WAGE IMPACT OF THE MARIELITOS 1081

the 1993 survey to avoid contamination from the Little Mariel supply shock of 1994 and 1995.

The CPS did not report a person’s country of birth before 1994, so I can- not measure the wage impact of the Mariel supply shock on the native-born population. I instead examine wage trends among non-Hispanic men (Hispanic background is determined by a person’s answer to the Hispanic ethnicity question), a sample restriction that comes close to isolating Miami’s native-born population at the time. The 1980 census, conducted days before the Mariel supply shock, reported that 40.7% of Miami’s male workforce was foreign-born, with 65.1% of those immigrants born in Cuba and another 11.2% percent born in other Latin American countries. The restriction to non-Hispanics also ensures that the post-Mariel wage trends are not skewed by a composition effect created by the large post-1980 wave of new Hispanic immigrants into many local labor markets.12

The dependent variable will be the worker’s log weekly earnings, for which weekly earnings are defined as the ratio of annual income in the pre- vious calendar year to the number of weeks worked.13 I use the Consumer Price Index (CPI) for all urban consumers to deflate the earnings data (1980 = 100). For expositional consistency and unless otherwise noted, whenever I refer to a particular year hereafter, it will be the year in which earnings were actually received by a worker, as opposed to the CPS survey year.14

It is important to document what we know about the skill distribution of the Marielitos. As noted earlier, the Mariel supply shock began a few days after the 1980 census enumeration, which means the first large survey that contains a large sample of the Marielitos is the 1990 census. Nevertheless, a few CPS supplements conducted in the 1980s provide information on a very small sample of Cuban immigrants who arrived at the time of Mariel.

Table 1 reports the education distribution of the sample of adult Cuban immigrants who arrived in 1980 (or in 1980–1981, depending on the data set) and who were enumerated in various surveys sometime between 1983 and 2000. The calculation includes the entire population of Marielitos (work- ers and non-workers, as well as men and women) who were at least 18 years old as of 1980.

The crucial implication of the table is that the Mariel supply shock con- sisted of workers who were very unskilled, with a remarkably large fraction

12The restriction to non-Hispanics implies that few of the low-skill workers in the sample are foreign- born. According to the 1980 census, only 18.7% of non-Hispanic high school dropouts in Miami were foreign-born (as compared to 6.4% outside Miami). By contrast, 55.1% of the non-Hispanic dropouts in Miami were black (as compared to 17.6% outside Miami). The key native-born group potentially affected by the Marielitos, therefore, was the low-skill African American workforce.

13I replicated the analysis using the log hourly wage as an alternative measure of a worker’s income. Because the measure of usual hours worked weekly in the March CPS is noisy, the results are similar but less precisely estimated.

14For example, a discussion of the earnings of workers in 1985 refers to the data drawn from the 1986 March CPS.

1082 ILR REVIEW

being high school dropouts.15 Despite the variation in sample size and the almost 20-year span in the surveys examined in the table, the fraction of Marielitos who lacked a high school diploma hovers around 60%. Table 1 also shows that a small fraction of these immigrants were college graduates (around 10%).

It is insightful to compare the education distribution of the Marielitos with that of the incumbent workforce. The last row of Table 1 shows that 26.7% of labor force participants in the Miami metropolitan area were high school dropouts. Miami’s workforce was remarkably balanced in terms of its skill distribution, with 20 to 30% of workers in each of the four education groups.16

Table 2 summarizes what we know about the magnitude of the Mariel supply shock. There were 176,300 high school dropouts in Miami’s labor force just prior to Mariel (out of a total of 659,400). According to the 1990 census, 60,100 Cuban workers migrated (as adults) either in 1980 or 1981. If we make a slight adjustment for the small number who entered the country in 1981, Mariel increased the size of the labor force by 55,700 persons, of which almost 60% were high school dropouts.17 Although the Mariel supply shock increased Miami’s workforce by 8.4% and increased the number of the most-educated workers by 3 to 5%, the number of low-skill workers rose

Table 1. Education Distribution of Adult Marielitos

Years of education

Sample \ 12 12 13 2 15 � 16 Sample size

Marielitos April 1983 CPS 57.9 25.6 3.5 13.1 31 June 1986 CPS 55.2 28.0 6.4 9.6 31 June 1988 CPS 58.7 26.1 4.4 10.9 46 1990 Census 64.8 15.8 12.9 6.5 4,234 1994 CPS-ORG 61.4 20.5 9.8 8.3 143 2000 Census 59.9 20.0 12.7 7.4 3,301

Miami’s pre-existing labor force 1980 Census 26.7 28.4 26.0 18.8 32,971

Notes: The statistics are calculated in the sample of persons born in Cuba who migrated to the United States at the time of Mariel and were 18 years old in 1980. In the April 1983 CPS and 2000 census, the Marielitos are identified as persons born in Cuba who migrated to the United States in 1980. In all other samples, the Marielitos are identified as Cubans who entered the country in 1980 or 1981. The pre- existing labor force of Miami includes both natives and immigrants.

15The skills of the Marielitos may have been further downgraded upon arrival, as found in Dustmann, Frattini, and Preston (2013), so that even those immigrants with a high school diploma were competing with the least-educated workers in the pre-existing Miami workforce.

16The pre-existing workforce includes all labor force participants in Miami, regardless of where they were born or their ethnicity.

17The 2000 census indicates that 92.8% of the Cubans who immigrated in either 1980 or 1981 actually entered the country in 1980. Note that the supply shock was probably slightly larger than indicated in Table 2 because the calculation does not account for mortality through 1990.

WAGE IMPACT OF THE MARIELITOS 1083

by a remarkable 18.4%. Moreover, this supply shock occurred almost over- night. The Coast Guard reports that the first Marielitos arrived in Florida on April 23, 1980, and that more than 100,000 refugees had been admitted by June 3, 1980 (Stabile and Scheina 2015).

Descriptive Evidence

The very low skills of the Marielitos indicate that we should perhaps focus our attention on the labor market outcomes of the least-educated workers in Miami to establish a first-order sense of whether the supply shock had any impact on Miami’s wage structure. The literature sparked by Borjas (2003) supports the importance of matching the immigrants to correspond- ing native workers by skill groups. Educational attainment is a skill category that would seem to be extremely relevant in an examination of the Mariel supply shock.

Any empirical study of the impact of Mariel runs into an immediate data problem: The number of workers enumerated by the CPS in the Miami labor market is small, introducing a lot of random noise into the calcula- tions. The top panel of Table 3 reports the number of men who satisfy the sample restrictions and who were enumerated in the Miami metropolitan area between survey years 1977 and 1993. The average number of observa- tions in each pre-1990 March CPS was 96.5, and the average number of high school dropouts was 19.5. The sample size drops dramatically with the 1991 survey, when the number of non-Hispanic men sampled in Miami falls abruptly (by nearly a third), and the number of high school dropouts some- times drops to the single digits. The empirical analysis reported below will effectively pool at least three years of the CPS (either by manually aggregat- ing the data or by estimating impacts for three-year intervals) to enable a more precise calculation of the effect.

I begin by carrying out the most straightforward calculation of the poten- tial wage impact that uses all the available data. In particular, I simply calcu- late the average log weekly wage of high school dropouts in Miami each

Table 2. Size of the Mariel Supply Shock

Education group Size of Miami’s labor force in 1980 (1000s)

Number of Marielitos in labor force (1000s)

Percentage increase in supply

High school dropouts 176.3 32.5 18.4 High school graduates 187.5 10.1 5.4 Some college 171.5 8.8 5.1 College graduates 124.1 4.2 3.4 All workers 659.4 55.7 8.4

Notes: The pre-existing number of workers in Miami is calculated from the 1980 census; the number of Marielito workers (at least 18 years old at the time of Mariel) is calculated from the 1990 census, and a small adjustment is made because the 1990 census reports the number of Cuban immigrants who entered the country in 1980 or 1981.

1084 ILR REVIEW

year between 1972 and 2002, the period for which the March CPS has a consistent time series for the Miami metropolitan area.18 Figure 2 illustrates the wage trend and the 95% confidence band around the mean, using a three-year moving average to smooth out the noise in the time series. The

Table 3. Number of Observations in the Miami-Hialeah Metropolitan Area

March CPS May CPS and CPS-ORG

Survey year All workers High school dropouts All workers High school dropouts

A. Men 1977 104 23 66 16 1978 101 26 66 12 1979 94 22 216 56 1980 88 17 245 55 1981 93 18 237 51 1982 93 20 209 39 1983 93 27 212 50 1984 93 18 209 48 1985 101 16 135 26 1986 99 15 231 36 1987 91 16 234 46 1988 102 18 264 37 1989 102 17 247 37 1990 98 16 223 38 1991 77 4 191 24 1992 84 10 177 20 1993 84 10 154 15 B. Men and women 1977 190 36 124 24 1978 175 42 117 24 1979 190 40 427 95 1980 180 30 450 88 1981 179 30 447 84 1982 175 37 414 73 1983 157 45 410 88 1984 170 28 420 77 1985 188 27 249 42 1986 205 31 512 59 1987 199 30 496 75 1988 213 37 545 71 1989 198 36 504 77 1990 197 35 475 82 1991 171 17 441 60 1992 178 18 398 44 1993 173 19 348 38

Notes: The table reports the number of observations for the Miami-Hialeah metropolitan area of non- Hispanic workers, aged 25–59, who reported positive annual wage and salary income, positive weeks worked, and positive usual hours worked weekly (in the March CPS), or positive usual weekly earnings and positive usual hours worked weekly (in the May/ORG data).

18The calculation of the average log weekly wage for each year weighs each individual observation by the product of the person’s sampling weight times the number of weeks worked in the previous calendar year.

WAGE IMPACT OF THE MARIELITOS 1085

figure also shows the trend for similarly educated non-Hispanic men work- ing outside Miami. Note that this simple exercise does not adjust the CPS data in any way (other than taking a moving average), so it provides a trans- parent indication of what happened to low-skill wages in pre- and post- Mariel Miami.

Despite the similarity in wage trends between Miami and the rest of the country prior to 1980, and despite the small sample size for the Miami met- ropolitan area, it is obvious that something happened in 1980 that caused the two wage series to diverge in a statistically significant way. Before Mariel, the log wage of high school dropouts in Miami was about 0.1 log points below that of workers in the rest of the country. By 1985, the gap had widened to 0.4 log points, implying that whatever caused the divergence had lowered the relative wage of low-skill workers in Miami by about 30%. The low-skill wage in Miami fully recovered by 1990, only to be hammered again in 1995, perhaps coincidentally with the time of the Little Mariel supply shock. By 2002, the wage gap between high school dropouts in Miami and elsewhere had returned to its pre-Mariel normal of about 0.1 log points.

Of course, Miami’s distinctive wage trend may not appear quite as distinc- tive when contrasted with what happened in other specific cities. The com- parison of Miami to the aggregate U.S. labor market may be masking part of the variation that influences particular localities and that disappears when averaged out. It may be important, therefore, to create a control group of comparable cities unaffected by the Mariel supply shock to

Figure 2. Log Wage of High School Dropouts, 1972–2003

Source: Data are drawn from the March CPS files. Notes: 95% confidence band. Log weekly wage is a three-year moving average of the average log wage of high school dropouts in each geographic area.

1086 ILR REVIEW

determine if the wage trends evident in Miami were attributable to macro- economic factors that affected other similar communities as well.

Beginning with the 1977 survey, the March CPS identifies 43 other metro- politan areas that can be combined in some fashion to construct a sort of placebo. Card (1990: 249; emphasis added) describes the construction of his control group as follows:

For comparative purposes, I have assembled similar data . . . in four other cities: Atlanta, Los Angeles, Houston, and Tampa-St. Petersburg. These four cities were selected both because they had relatively large populations of blacks and Hispanics and because they exhibited a pattern of economic growth similar to that in Miami over the late 1970s and early 1980s. A comparison of employment growth rates . . . suggests that economic conditions were very similar in Miami and the average of the four comparison cities between 1976 and 1984.

It is important to emphasize that the four cities in the Card placebo were chosen based partly on employment trends observed after the Mariel supply shock. In other words, if Mariel worsened employment conditions in Miami, the Card placebo is comparing the poorer outcomes of workers in Miami to the outcomes of workers in cities where some other factor worsened their opportunities as well. It is preferable to exogenize the choice of a placebo by comparing cities that were roughly similar prior to the treatment, rather than being similar after one of them was injected with a very large supply shock.

The panels of Figure 3 illustrate the wage trends in Miami and several potential placebos between 1976 and 1992.19 The top panel shows that the log wage of high school dropouts declined dramatically after 1980 when compared to what happened in the cities that make up the Card placebo. Of course, trends in absolute wages reflect many factors that are specific to local labor markets, so these ups and downs may capture idiosyncratic shifts that affected all workers in Miami. The Mariel supply shock, however, specif- ically targeted the least-educated workers, and the bottom two panels of the figure show that the relative wage of high school dropouts in Miami—relative to either college graduates or high school graduates—also declined drama- tically after Mariel, and also recovered by 1990. In sum, the wage trends observed in Miami consistently indicate that the economic well-being of the least-educated workers in Miami took a downward turn shortly after 1980, reached its nadir around 1985–1986, and did not recover fully until 1990.

As noted above, the cities in the Card placebo do not make up a proper control group because they were chosen, in part, so that post-Mariel employment conditions in the placebo cities resembled those in Miami. To determine the set of cities that had comparable employment growth prior to

19The calculation of wage trends in Figures 2 and 3 differs slightly. To calculate the standard error of the (moving) average in Figure 2, I computed the average by pooling three years of micro data. The data in Figure 3, which are used in the regression analysis reported below, calculate the mean log wage for each survey and then take a simple moving average of those means.

WAGE IMPACT OF THE MARIELITOS 1087

Figure 3. Trends in the Wage of Low-Skill Workers in the March CPS, 1977–1992

A. Log weekly wage of high school dropouts

B. Log wage of high school dropouts relative to college graduates

C. Log wage of high school dropouts relative to high school graduates

Notes: Figures use a three-year moving average of the average log wage of high school dropouts, high school graduates, and college graduates in each specific geographic area.

1088 ILR REVIEW

Mariel, I pooled the 1977 and 1978 surveys of the CPS and also pooled the 1979 and 1980 surveys. I then used the pooled surveys to calculate the log of the ratio of the total number of workers in 1979–1980 to the number of workers in 1977–1978.20 Column 1 of Table 4 reports the employment growth rate for each of the 44 metropolitan areas, ranked by the growth rate.

Miami’s pre-Mariel employment conditions were quite robust, ranking 6th in the rate of employment growth. Note that all the cities that made up the Card placebo had lower growth rates than Miami had between 1977 and 1980. The average employment growth rate in those four cities (Atlanta, Los Angeles, Houston, and Tampa-St. Petersburg) (weighted by the city’s employment) was 6.9%, less than half the 15.3% growth rate in Miami.

I use the rankings reported in Table 4 to construct a new placebo, which I call the ‘‘employment placebo,’’ by simply choosing the four cities that were most similar to Miami prior to 1980. Specifically, the employment pla- cebo consists of the four cities (Anaheim, Rochester, Nassau-Suffolk, and San Jose) ranked just above and just below Miami.

Figure 3 clearly shows that the relative decline in the wage of low- educated workers in Miami is much larger when we compare Miami to cities that had comparable employment growth than to the cities that made up the Card placebo. Between 1979 and 1985, for instance, the wage of high school dropouts in Miami relative to the Card placebo fell by 0.25 log points (or 22%), but the decline was 0.43 log points (35%) when compared to the cities in the employment placebo. This difference is not surprising; by con- struction, the comparison of post-Mariel economic conditions in Miami to that of cities where employment conditions are also poor inevitably hides some of the impact of the Marielitos.

I also constructed an alternative ‘‘low-skill placebo’’ by choosing the four cities that had similar pre-Mariel growth for low-skill employment. Column 2 of Table 4 reports the rate of employment growth for high school dropouts. Miami also ranked 6th by this metric. Coincidentally, two of the cities with similar low-skill employment growth are in the Card placebo (Los Angeles and Houston; the other two are Gary and Indianapolis). Figure 3 shows that the post-1980 Miami experience was also unusual when compared to this placebo, with a wage drop of 30% (which lies in between the effects implied by the Card and employment placebos).

The choice of a placebo obviously plays a crucial role in determining the magnitude of the wage impact of Mariel. An element of arbitrariness occurs in how a placebo is created, so the researcher’s choice of a particular pla- cebo can exaggerate or attenuate the wage effect. For example, 123,410 potential four-city placebos can be created from 43 metropolitan areas. It

20A person is employed if he or she works in the CPS reference week. The 1980 survey, collected in March, is not affected by the supply shock, as the Marielitos did not begin to arrive until late April.

WAGE IMPACT OF THE MARIELITOS 1089

might be reasonable to expect a huge dispersion in the estimated wage effect of the Marielitos across the 123,410 potential comparisons. I will report below the distribution of estimated wage impacts across all potential

Table 4. Rates of Employment and Wage Growth before Mariel

Rank Metropolitan area

Employment growth:

all workers

Employment growth: high school dropouts

Wage growth: high school dropouts

1 San Diego, CA 0.194 0.067 20.127 2 Greensboro-Winston Salem, NC 0.182 20.063 20.336 3 Kansas City, MO/KS 0.179 0.052 20.178 4 Anaheim-Santa Ana-Garden Grove, CA 0.162 0.257 0.068 5 Rochester, NY 0.153 20.172 0.019 6 Miami-Hialeah, FL 0.153 0.086 0.012 7 Nassau-Suffolk, NY 0.151 0.056 20.047 8 San Jose, CA 0.137 0.130 0.138 9 Albany-Schenectady-Troy, NY 0.130 0.065 0.065 10 Boston, MA 0.121 20.100 20.024 11 Milwaukee, WI 0.121 20.006 0.056 12 Indianapolis, IN 0.115 0.071 20.048 13 Seattle-Everett, WA 0.110 20.079 20.051 14 Norfolk-Virginia Beach-Newport News, VA 0.103 0.052 0.110 15 Philadelphia, PA/NJ 0.102 20.033 0.012 16 Newark, NJ 0.092 20.116 20.124 17 Tampa-St. Petersburg-Clearwater, FL 0.083 0.068 0.120 18 Denver-Boulder-Longmont, CO 0.082 20.139 0.027 19 Houston-Brazoria, TX 0.078 0.090 0.003 20 Sacramento, CA 0.078 0.152 20.034 21 Dallas-Fort Worth, TX 0.076 0.062 20.050 22 Portland-Vancouver, OR/WA 0.071 20.074 20.020 23 Riverside-San Bernardino, CA 0.071 20.017 0.254 24 Atlanta, GA 0.069 20.087 0.014 25 Cincinnati-Hamilton, OH/KY/IN 0.063 0.038 20.032 26 Washington, DC/MD/VA 0.061 0.028 0.090 27 Detroit, MI 0.060 20.099 0.010 28 Fort Worth-Arlington, TX 0.058 20.006 20.059 29 Los Angeles-Long Beach, CA 0.056 0.075 20.128 30 Columbus, OH 0.048 20.324 20.004 31 Buffalo-Niagara Falls, NY 0.039 0.040 20.144 32 Chicago-Gary-Lake, IL 0.025 20.082 20.017 33 St. Louis, MO/IL 0.019 20.060 20.060 34 Bergen-Passaic, NJ 0.015 20.051 0.007 35 Baltimore, MD 0.012 20.108 20.001 36 Minneapolis-St. Paul, MN 0.007 20.050 20.012 37 Cleveland, OH 0.001 20.071 20.014 38 New York, NY 0.000 20.146 0.074 39 Pittsburg, PA 20.013 20.111 0.123 40 Birmingham, AL 20.020 20.172 20.124 41 San Francisco-Oakland-Vallejo, CA 20.027 20.200 20.105 42 Gary-Hammond-East Chicago, IN 20.029 0.119 0.042 43 New Orleans, LA 20.046 20.313 20.017 44 Akron, OH 20.110 20.351 20.005

Notes: Rate of employment growth is the log ratio of average employment in 1979–1980 to average employment in 1977–1978, calculated from the 1977–1980 survey years of the March CPS. Rate of wage growth is the difference in the (age-adjusted) log weekly wage between 1978–1979 and 1976–1977.

1090 ILR REVIEW

four-city placebos and show that Mariel had a negative impact regardless of which placebo is chosen.

Alternatively, one can employ the synthetic control method developed by Abadie and Gardeazabal (2003) and Abadie, Diamond, and Hainmueller (2010). The method essentially searches across all potential placebo cities and derives a weight that combines cities to create a new synthetic city. This synthetic city is the one that best resembles the pre-Mariel Miami labor mar- ket along some set of pre-specified conditions. Unlike the hands-on method I used to create the employment and low-skill placebos, the synthetic con- trol allows the construction of the synthetic city to be based on several char- acteristics. The synthetic control approach seems to limit the researcher’s ability to make arbitrary decisions about what the proper placebo should be. Note, however, that an element of arbitrariness still transpires. The researcher must specify the vector of variables that should be comparable between Miami and the placebo in the pre-treatment period. As I show below, different choices of control variables lead to different estimates of the wage impact.

Initially, I construct the synthetic city by using three such control vari- ables: the rate of employment growth in the four-year period prior to Mariel (i.e., the variable used to define the employment placebo); the con- current rate of employment growth for high school dropouts (the variable used to define the low-skill placebo); and the concurrent rate of wage growth for high school dropouts. The last column of Table 4 shows that the low-skill market in Miami also had robust wage growth prior to Mariel, rank- ing 13th in the country. Figure 3 illustrates the wage trends in the ‘‘city’’ that makes up the synthetic control. The post-1980 Miami experience dif- fered markedly from that of the synthetic control.

It is interesting to examine the weights attached to each city by the syn- thetic control method. When looking at the log wage of high school drop- outs, the method assigns the largest weights to Anaheim (a weight of 0.40), Rochester (0.20), San Diego (0.18), and San Jose (0.06). The ranking of those cities in Table 4 indicates that the synthetic control method consis- tently selects metropolitan areas that had robust labor markets prior to Mariel.

Finally, I can establish that the steep drop in the low-skill wage in post- Mariel Miami was a very unusual event. The average wage of male high school dropouts in Miami fell by about 37% between 1976–1979 and 1981– 1986. We can calculate the comparable wage change in every other metro- politan area for all equivalent time periods between 1976 and 2003 to see if the Mariel experience stands out.21 If 37% wage cuts happen frequently in local labor markets, it would be harder to claim that Miami’s experience had much to do with the Marielitos. Perhaps something else was going on—a

21Garthwaite, Gross, and Notowidigdo (2014) conducted a similar exercise to examine the distribution of the impact of an experiment in health insurance availability on employment lock.

WAGE IMPACT OF THE MARIELITOS 1091

something else that other cities experience often enough at different times—that just happened to coincide with the timing of Castro’s decision.

To assess how Miami’s post-Mariel experience compares to that of the entire distribution of wage changes, I calculated the wage change between every single pre-treatment period t (1976–1979, 1977–1980, . . ., 1993– 1996) and the corresponding post-treatment period t# (1981–1986, 1982– 1987, . . ., 1998–2003). Specifically, I pooled the March CPS data for the four years in each pre-treatment period and the six years in each post- treatment period, and calculated the average log wage in each city–period permutation. To duplicate the Mariel experiment, I skip a year between each four-year pre-treatment span and each six-year post-treatment span. The exercise generates a total of 774 possible events outside Miami (43 met- ropolitan areas and 18 potential treatment years between 1980 and 1997).

The top panel of Figure 4 illustrates the frequency distribution of all observed changes in the wage of male high school dropouts outside Miami. Between 1976–1979 and 1981–1986, the log wage of high school dropouts

Figure 4. Distribution of Pre- and Post-Mariel Wage Changes, 1976–2003

Source: Data are drawn from the March CPS files. Notes: Pre-treatment period is four years; post-treatment period is six years; and year of the treatment is excluded from the calculation.

1092 ILR REVIEW

in Miami fell by 0.463 log points (or 37%). Visually, we see that such a large wage drop was a rare event. The mean observed wage change across all city–period permutations was only about 20.16 log points. The Mariel expe- rience ranks in the first percentile of the distribution of all observed wage changes between 1976 and 2003 across all metropolitan areas. And the fre- quency distribution for the 1980 treatment year shows that the wage drop observed in Miami at the time of Mariel was the largest wage drop observed among all metropolitan areas.

Equally important, the bottom panel of Figure 4 reveals that more- educated workers in Miami did not experience a substantial wage drop. Although it has been claimed that perhaps high school dropouts and high school graduates are perfect substitutes and should be pooled to form the low-skill workforce, the Mariel data clearly contradict this conjecture.22 The mean wage change in the log wage of high school graduates across all city– period permutations in the years 1976 through 2003 was 20.060, and Miami’s Mariel experience ranked in the 43rd percentile. The value observed in the Miami metropolitan area at the time of Mariel was 20.080, ranking 37th out of the 44 metropolitan areas in the distribution for treat- ment year 1980.

In short, something unique happened to the economic status of high school dropouts in Miami in the early 1980s. The event that shocked the wage structure in Miami at the time of Mariel, whatever it happened to be, occurs rarely and its adverse consequences were narrowly targeted on work- ers who lacked a high school diploma.

Robustness of the Descriptive Evidence

Given the striking picture that the raw data imply about the labor market impact of the Marielitos, and given the very contentious debate over immi- gration policy both in the United States and abroad, it is important to estab- lish that the evidence is robust. I now address two distinct issues to evaluate the sensitivity of the results. First, was the decline in the wage of high school dropouts in the Miami of the early 1980s recorded by other contempora- neous data sets, such as the CPS Outgoing Rotation Groups (ORG)? Second, is the evidence robust to the inclusion of women in the analysis?

Results from the CPS-ORG

It is well known that wage trends recorded by the March CPS sometimes dif- fer from the comparable wage trends recorded by the CPS ORG. Unlike the March CPS, which reports annual earnings in the calendar year prior to the survey, the ORG gives a measure of the hourly wage for respondents who are paid by the hour and of the usual weekly wage for all other

22See the contrasting arguments of Ottaviano and Peri (2012) and Borjas, Grogger, and Hanson (2012).

WAGE IMPACT OF THE MARIELITOS 1093

workers. The ORG time series begins in 1979, so that the pre-treatment period contains only one year of data. Following Autor, Katz, and Kearney (2008) and Lemieux (2006), I extend the pre-treatment period back by using the roughly comparable May CPS supplements for the pre-1979 years.23

One key advantage of the ORG data is that they contain much larger samples, by construction, roughly three times the size of the March CPS. This advantage is somewhat neutralized, however, by the need to use the May CPS files to create a longer pre-treatment period. The May files, like the March CPS, are monthly surveys and both have equally small samples. In fact, as Table 3 shows, the number of high school dropouts enumerated in pre-Mariel Miami who satisfy the sample restrictions is smaller in the May CPS than in the March file. The presumed advantage of the ORG file is fur- ther neutralized because, in practice, the ORG does not triple the number of observations in the March CPS. The average ORG sample between 1981 and 1989 is only 2.2 times as large as the March file (an annual average of 41.1 compared to 18.3 observations). The fact that using the ORG does not triple the sample size indicates that some workers go missing in the ORG analysis.

Those workers disappeared because the March CPS and the ORG mea- sure different concepts of income, creating very different samples that can be used to analyze wage trends. The March CPS reports wage and salary income from all jobs held in the previous calendar year. The ORG mea- sures the wage in the main job held by a person in the week prior to the survey—if working. The exclusion of persons who happen not to be working in that particular week (but worked sometime during the year) leads to a noticeable decline in the potential number of observations in the ORG data. At the same time, however, the ORG does allow a more precise calcu- lation of the price of skills. I use the log hourly wage rate of a worker, defined as the ratio of usual weekly earnings to usual hours worked per week, as the dependent variable in the ORG analysis.

The top panel of Figure 5 uses all the available May/ORG surveys in which the Miami metropolitan area is consistently defined (1973 through 2003) and illustrates the 95% confidence bands for the trends in the aver- age log hourly wage of low-skill workers in Miami and in the rest of the country.24 As with the March CPS, the two wage series are roughly similar before Mariel. Miami’s low-skill wage then tumbled after 1980 and recov- ered by 1990. The figure also shows a significant wage drop in the mid- 1990s, coinciding with the arrival of the Little Marielitos.

23I use the pre-1979 May files and the 1979–2003 ORG files archived at the National Bureau of Economic Research.

24To increase sample size in the pre-treatment period, the January, February, and March samples of the 1980 ORG are added to the 1979 data, as those early months are unaffected by Mariel. The calcula- tion of the average log wage in the cell weighs each individual observation by the product of the person’s earnings weight times the usual number of hours worked weekly.

1094 ILR REVIEW

Before proceeding to further document the potential disparities in wage trends across the different surveys, I adjust the data for differences in the age distribution of workers in different time periods and in different metro- politan areas. I use a simple regression model to calculate the age-adjusted mean wage of a skill group in a particular market. Specifically, I estimate

Figure 5. Trend in the Wage of High School Dropouts in the CPS-ORG, 1973–2003

A. 95% confidence band

B. Average wage trends in Miami and placebos

Source: Data are drawn from the May/CPS-ORG files. Notes: Log hourly wage is a three-year moving average of the average log wage of high school dropouts in each geographic area in Figure 5A, and of the age-adjusted wage in Figure 5B.

WAGE IMPACT OF THE MARIELITOS 1095

the following individual-level earnings regression separately in each CPS cross-section:

log wirst = ur + Ai gt + e,ð1Þ

where wirst is the wage of worker i in city r in education group s at time t; ur is a vector of fixed effects indicating city of residence; and Ai is a vector of fixed effects giving the worker’s age.25 The fixed effects ur deflate the log wage for regional wage differences. The average residual from this regres- sion for cell (r, s, t) gives the age-adjusted mean wage of that cell. Unless otherwise specified, I use age-adjusted wages for the remainder of the article.26

The bottom panel of Figure 5 illustrates the wage trends for Miami and the various placebos over the 1976–1992 period. It is visually evident that something happened to the low-skill labor market in Miami in the early 1980s, particularly when Miami’s trend is compared to the employment, low-skill, or synthetic placebos. The use of the Card (1990) placebo in the ORG data tends to mask much of what went on in post-Mariel Miami.

For example, the wage of high school dropouts in Miami fell by 0.18 log points between 1979 and 1985. The comparable wage fell by 0.13 log points in the Card placebo, but by only 0.06 log points in either the employment or synthetic placebos. The use of the Card placebo would imply that Mariel lowered the wage of high school dropouts in Miami by only about 5%, while both the employment and the synthetic placebos would imply an impact of more than 10%.

Inclusion of Women

The number of observations available to examine wage trends in Miami before and after Mariel could be increased by including women in the anal- ysis. Female labor force participation was increasing very rapidly in the 1970s and 1980s, however, so wage trends are likely to be affected by the changing gender composition of the workforce as well as by the selection that marks women’s entry into the labor market. Moreover, female partici- pation increased differentially across cities. For example, the 1970 decennial census data reported that 35.3% of Anaheim’s workforce was female; by 1990, this fraction had risen by almost 10 percentage points to 45.1%. The increase in San Diego was similar, from 34.0 to 42.5%. In Miami, however, the increase was far smaller, from 40.2 to 42.1%.

The bottom panel of Table 3 shows that including women would indeed increase the sample size of low-skill workers in Miami, from an average of 19.5 in each pre-1990 March CPS cross-section to an average of 34.5.

25I use seven age groups to create the fixed effects (25–29, 30–34, 35–39, 40–44, 45–49, 50–54, and 55–59).

26Note that the wage trends in the age-adjusted data implied by the March CPS look almost identical to the raw trends illustrated in Figure 3.

1096 ILR REVIEW

Because of the male–female wage gap (which may itself be changing over time), the differential rates of growth in female labor force participation across cities distort relative wage trends between Miami and any placebo. Therefore, an analysis that examines the pooled earnings of men and women must (at the very least) account for the changing gender composi- tion of the workforce. I adjust for this composition effect by running the fol- lowing wage regression in the pooled sample of men and women in each cross-section of the CPS:

log wirst = ur + Ai gt + Fi dt + e,ð2Þ

where Fi is a dummy variable set to unity if the worker is female. The resi- dual from this regression gives the age- and gender-adjusted wage of worker i in year t.

The two panels of Figure 6 illustrate the trends in the average adjusted wage observed in the pooled sample of men and women in both the March CPS and the ORG files. Adding women to the sample does not change the insight that something happened in Miami after 1980. It is important to stress, however, that the gender-adjusted wage trends are still contaminated by (unknown) differences across cities in the nature of the selection that motivates only some women to enter the labor market. The statistical diffi- culties associated with purging the data from this type of selection bias sug- gest that the most credible evidence of the Mariel impact on the price of skills is likely drawn from samples that examine the wage trends among prime-age men.27

Regression Results

To estimate the impact of the Mariel supply shock relative to the various placebos, I use the mean age-adjusted log wage of male high school drop- outs in city r at time t, denoted by log wrt . This wage becomes the depen- dent variable in a generic difference-in-differences regression model:

log wrt = ur + ut + b(Miami 3 Post-Mariel) + e,ð3Þ

where ur is a vector of city fixed effects; ut is a vector of year fixed effects; Miami represents a dummy variable indicating the Miami-Hialeah metropol- itan area; and Post-Mariel indicates if time t occurs after 1980.

27My analysis uses the sample of non-Hispanic men aged 25 to 59. It may seem that an ‘‘easy’’ way of increasing sample size is to include non-Cuban Hispanics in the calculations. This solution, however, is very problematic. There was a rapid increase in Hispanic immigration in the 1980s, so that many of the observations added to the sample would be Hispanics (mainly from Mexico) who migrated to the United States after Mariel. The 1990 census implies that if one were to add non-Cuban Hispanics to the sample, 52% of the new observations are immigrants who arrived in the 1980s. The presence of such a large number of post-1980 immigrants, who typically have very low entry wages, distorts the post-1980 wage trend in many of the placebo cities, such as Los Angeles, seriously contaminating any measurement of the wage impact of the Mariel supply shock and biasing it toward zero.

WAGE IMPACT OF THE MARIELITOS 1097

The regression uses annual observations between t = 1977 and t = 1992, but excludes 1980, the year of the supply shock.28 The cities r included in

Figure 6. Trends in the Wage of High School Dropouts in Pooled Sample of Men and Women, 1977–1992

A. March CPS

B. CPS-ORG

Notes: Figures use a three-year moving average of the age- and gender-adjusted average log wage of the pooled group of men and women in each specific geographic area.

28This time span enables me to estimate the identical regression model in both the March CPS and ORG samples.

1098 ILR REVIEW

the regression are Miami and the cities in a specific placebo. For example, if the Miami experience is being compared to that of cities in the employ- ment placebo, there would be five cities in the data, and each of these cities would be observed 15 times between 1977 and 1992, for a total of 75 obser- vations. The regression comparing Miami to the synthetic control is similar in spirit, but only two cities are in this regression: Miami and the synthetic city, for a total of 30 observations. To allow the wage impact of Mariel to vary over time (and to partially alleviate the problem of small samples), the post-Mariel variable in Equation (3) is initially a vector of fixed effects indi- cating whether the observation refers to the three-year intervals 1981–1983, 1984–1986, 1987–1989, or 1990–1992.

Table 5 presents the estimated coefficients in the vector b (and robust standard errors) for various specifications of the model.29 Consider initially the coefficients reported in Panel A, drawn from regressions estimated in

Table 5. Difference-in-Differences Impact of the Marielitos on the Wage of High School Dropouts

Sample/Variable Card placebo Employment placebo Low-skill placebo Synthetic control All cities

A. March CPS, men 1981–1983 20.204 20.290 20.180 20.257 20.191

(0.076) (0.073) (0.081) (0.077) (0.064) 1984–1986 20.368 20.454 20.301 20.458 20.393

(0.060) (0.059) (0.054) (0.075) (0.022) 1987–1989 20.328 20.303 20.228 20.293 20.286

(0.081) (0.072) (0.060) (0.070) (0.059) 1990–1992 20.026 20.056 0.039 20.188 20.003

(0.072) (0.123) (0.065) (0.155) (0.032) B. CPS-ORG, men

1981–1983 20.075 20.140 20.079 20.154 20.083 (0.026) (0.049) (0.024) (0.026) (0.012)

1984–1986 20.069 20.116 20.076 20.157 20.086 (0.057) (0.065) (0.054) (0.056) (0.046)

1987–1989 20.106 20.175 20.101 20.150 20.141 (0.036) (0.064) (0.035) (0.023) (0.022)

1990–1992 0.019 20.070 0.036 20.079 20.008 (0.041) (0.062) (0.038) (0.016) (0.023)

Notes: Robust standard errors are reported in parentheses. Data consist of annual observations for each city between 1977 and 1992 (1980 excluded). All regressions include vectors of city and year fixed effects. The table reports the interaction coefficients between a dummy variable indicating if the metropolitan area is Miami and the timing of the post-Mariel period. Regressions that use the Card, employment, or low-skill placebos have 75 observations; regressions that use the synthetic placebo have 30 observations; and regressions in the last column have 658 observations. All regressions are weighted by the number of observations used to calculate the dependent variable. The number of observations of the synthetic control is a weighted average of the sample size in the actual cities that make up the synthetic city.

29The presence of serial correlation in outcomes at the city level requires further adjustments for valid statistical inference, but clustered standard errors are downward biased when the data have few clusters (Cameron and Miller 2015).

WAGE IMPACT OF THE MARIELITOS 1099

the March CPS sample. The various columns of the table use the alternative placebos introduced in the previous section as well as an aggregate placebo composed of all other 43 metropolitan areas. The various rows show how the wage impact varies during the post-Mariel period. The variation in these coefficients presumably captures the wage effect as the Miami labor market adjusts and moves from the short to the long run.

Regardless of the placebo used, the wage effect immediately after Mariel is negative, indicating an absolute decline in the wage of low-skill workers in Miami. All the regressions also suggest that the wage impact was stronger between 1983 and 1986 than between 1981 and 1983. On average, it seems that the low-skill wage in the first six years after Mariel fell by at least 20 or 30%. Finally, the evidence indicates that the wage effect weakened after 1986 and essentially disappeared by 1990.

Panel B reports comparable coefficients from regressions estimated in the ORG data. Although the wage impact in the first six years after Mariel is again consistently negative, it is numerically smaller, hovering around 10%. The ORG regressions also suggest an eventual attenuation of the impact.

To easily illustrate the sensitivity of the regression evidence, the remain- der of the analysis uses a simpler form of the model in Equation (3), focus- ing on the years between 1977 and 1986 (excluding 1980). The short-run wage effect reported in Table 6 is simply the interaction between the indica- tor for the Miami metropolitan area and the indicator for a post-1980 obser- vation. Among men, the short-run impact ranges from about 10 to 30%. Table 6 also reports the analogous coefficient estimated using the age- and gender-adjusted wage in the pooled sample of men and women. Although these effects are weaker (between 5 and 20%), they are always negative and significant.

Table 6. Sensitivity of Estimates of the Short-Run Wage Impact, 1977–1986

Men Men and women

Placebo March CPS CPS-ORG March CPS CPS-ORG

1. Card placebo 20.280 20.074 20.189 20.040 (0.069) (0.031) (0.061) (0.024)

2. Employment placebo 20.358 20.131 20.220 20.069 (0.064) (0.044) (0.048) (0.033)

3. Low-skill placebo 20.269 20.078 20.182 20.040 (0.072) (0.028) (0.056) (0.022)

4. Synthetic control 20.343 20.155 20.175 20.075 (0.081) (0.028) (0.051) (0.024)

5. All cities placebo 20.279 20.086 20.184 20.036 (0.062) (0.023) (0.051) (0.020)

Notes: Robust standard errors are reported in parentheses. Data consist of annual observations for each city between 1977 and 1986 (1980 excluded). The table reports the interaction coefficients between a dummy variable indicating if the metropolitan area is Miami and if the observation is drawn from the post-Mariel period. See the notes to Table 5 for more details on the regression specification.

1100 ILR REVIEW

The ORG wage effects are about one-half to one-third the size of the cor- responding effects in the March CPS, but the standard errors imply that the difference is often not statistically significant. A large part of the difference in point estimates between the two data sets can be attributed to labor sup- ply effects and to measurement issues related to labor supply. For example, the short-run wage impact for men using all other metropolitan areas as a placebo is 20.279 (0.062) in the March CPS and 20.086 (0.023) in the ORG. The estimated coefficient in the March CPS would fall to 20.216 (0.059) if I used the log hourly wage rate instead of log weekly earnings as the dependent variable and the average log hourly wage in a cell was calcu- lated as in the ORG, using a weight equal to the sampling weight times the number of hours worked weekly. Finally, the coefficient would fall even more, to 20.167 (0.075), if the March sample was limited to persons who worked in the reference week (as in the ORG). In short, conceptual and measurement issues related to labor supply account for about half the dif- ference in point estimates.

It is also instructive to extend the synthetic control approach to estimate the distribution of short-run wage effects implied by an intriguing counter- factual exercise. What would the distribution of estimated effects look like if we acted as if each city had experienced a shock in 1980 and simply calcu- lated the pre- and post-wage change attributable to that imaginary supply shock relative to each city’s synthetic control?30 Would the wage effect esti- mated for the actual Mariel supply shock look all that unusual when com- pared to the entire distribution of hypothetical wage effects?

To be more specific, I again define the pre-treatment period as 1977 to 1979 and the post-treatment period as 1981 to 1986. Imagine that the city of Akron was hit by a phantom supply shock in 1980. We can then calculate the wage trends in Akron and in Akron’s synthetic control in the pre- and post-treatment periods and estimate the difference-in-differences regression model in Equation (3). Presumably, the wage effect resulting from this exer- cise should be near zero because Fidel Castro did not suddenly decide to relocate more than 100,000 Cubans to Akron in 1980. However, other (ran- dom) things may have happened in post-1980 Akron that we know nothing about and that may have changed the relative wage of low-skill workers in that city relative to its synthetic control.

I construct each city’s synthetic control by using the same control vari- ables introduced earlier: the city’s rate of employment growth and the con- current rates of employment and wage growth for the low-skill workforce. I then estimate the regression model in Equation (3) to calculate the short- run wage impact for each city relative to its synthetic control. Figure 7 illus- trates the distribution of the estimated effects on the log wage of male high school dropouts. Although there is a lot of dispersion across all the

30This exercise effectively extends the distributional analysis summarized in Figure 4 by contrasting what actually happened in city r with what happened in city r’s synthetic control.

WAGE IMPACT OF THE MARIELITOS 1101

hypothetical shocks, the mean effect is numerically equal to zero.31 Both the March CPS and the ORG imply that the wage effect induced by the real Mariel supply shock was the most negative wage impact observed during the period.

Figure 7. Distribution of Hypothetical Short-Run Impacts Relative to Synthetic Placebo, Assuming a Supply Shock Hits Each City in 1980

Notes: Pre-treatment period is three years; post-treatment period is six years. Wage effect is estimated from a difference-in-differences regression model that excludes 1980, the year of the treatment.

31The mean of the distribution is 20.018 in the March CPS and 20.014 in the ORG.

1102 ILR REVIEW

Although the regression results unambiguously indicate that the Mariel supply shock harmed low-skill workers, the overall evidence may not be con- sistent with the textbook model of factor demand. The March CPS data sug- gest that the adverse wage effect of the Marielitos initially increased over time before eventually disappearing. This finding is hard to square with the theoretical prediction that the wage effect should be largest right after the supply shock and would weaken as the capital stock adjusted over time.

There has been little research on the dynamics of immigration-induced supply shocks. But there has been much work on the dynamics of demand shocks. The presumption that wages are sticky downward is a common fea- ture in business cycle models. Many studies, such as those that examine the impact of oil shocks (Hamilton 1983), recognize that the largest effects of demand shocks do not happen immediately. It typically takes a few years for the adverse shock to reverberate through the labor market. In the Mariel context, it may also be that employers exploited sticky nominal wages dur- ing the high inflation of the early 1980s as a way of passing through the wage cuts. Given the well-documented lags between demand shocks and their consequences, we do not yet know if the dynamics of the wage data in post-Mariel Miami are consistent with the expected effects of a supply shock.32

We also do not fully understand why the relative wage of low-skill workers in Miami recovered after a decade (as illustrated in Figure 3). Economic theory implies that it is the average wage that will return to its pre-Mariel level if the production function is linear homogeneous (Borjas 2014). The relative wage effect will not go away unless there has also been a change in the relative quantities of low- and high-skill labor.

Equally important, the adjustments induced by the Mariel supply shock probably involved much more than the increase in the capital stock that plays the central role in the neoclassical model of labor demand. Within a month after the Mariel boatlift began, racial riots ravaged parts of Miami, leaving 18 people dead and 400 injured. The conditions on the ground were volatile, and the riots were the consequence of a long list of accumu- lated grievances. Nevertheless, one of the grievances was ‘‘the displacement of blacks by Cubans from jobs and other opportunities’’ (Vogel and Stowers 1991: 120).

The political and social upheaval ignited by Castro’s decision to open up the port of Mariel, as well as the increasingly important role that the Miami

32Another potential explanation for the delayed impact is that perhaps Miami continued to be hit by large supply shocks, which eventually subsided by the late 1980s. The data from the 1990 census, how- ever, is not consistent with this conjecture. After the entry of the Marielitos, there was a steady flow of low- skill immigrants into Miami, increasing their number by 9.1% in 1982–1984, by 8.6% in 1985–1986, and by 11.8% in 1987–1990. The Miami experience was not unusual. There was a surge in low-skill immigra- tion in the 1980s throughout the country, so that many of the cities that often end up in the placebos experienced similar supply shocks. In Anaheim, new arrivals increased the number of low-skill immi- grants by 11.1% in 1982–1984, by 11.1% in 1985–1986, and by 17.9% in 1987–1990. The statistics for the cities in the low-skill placebo are 8.6%, 7.7%, and 10.0%, respectively.

WAGE IMPACT OF THE MARIELITOS 1103

of the 1980s played in the market for illicit narcotics, affected Miami’s econ- omy in ways that extend far beyond what our models capture. Put bluntly, the ceteris paribus assumption in the neoclassical model does not apply. As a result, it is difficult to infer much about the dynamics of the wage effect from the evidence provided by the Mariel supply shock.

The coefficients reported in Table 6 suggest that the wage of male high school dropouts in Miami fell by 10 to 30% during the first six years after Mariel (depending on the placebo and data set used). The supply shock increased the number of high school dropouts by approximately 20%, so that the implied wage elasticity (d log w/d log L) is between 20.5 and 21.5.

Either of these elasticity estimates is far higher than the typical wage effect estimated in non-experimental cross-city regressions that link wages to immigration (which sometimes cluster around a negligible number). They are also higher than the wage elasticities estimated by correlating wages and immigration across skill groups in the national labor market (Borjas 2003), an elasticity that clusters around 20.3 to 20.4. The estimates are close, however, to those reported in Monras (2015) and Llull (2015), who use new instruments (including the Peso Crisis in Mexico, natural dis- asters, armed conflicts, and changes in political conditions) to correct for the endogeneity of migration flows. Monras reported a wage elasticity of 20.7 while Llull estimated an elasticity of about 21.2.

Granted, many caveats need to be resolved regarding the specification of the regression model and the external validity of the Mariel experience before we fully buy into an elasticity estimate of between 20.5 and 21.5. Nevertheless, the key implication of the evidence is unambiguous. The wage of high school dropouts in the Miami labor market fell significantly after the Mariel supply shock. Any attempt at rationalizing this fact as caused by something other than the Marielitos will need to specify precisely what those other factors were.

The Choice of a Placebo

One lesson from the evidence presented in the previous sections is that the choice of a placebo matters. To easily document this sensitivity, I estimate the short-run wage impact using alternative specifications of the synthetic control method, for which I use different sets of control variables to create the synthetic city.

Specifically, row 1 of Table 7 again reports the impact estimated with the controls introduced earlier: employment growth, low-skill employment growth, and low-skill wage growth. The wage impact (in the sample of men) is 20.34 in the March CPS and 20.16 in the ORG. The control variables in row 2 are those used by Card (1990) to (manually) define his placebo: employment growth, percentage of the workforce that is Hispanic, and per- centage that is black. Note that the implied wage impact in the March CPS is about the same, but the wage impact in the ORG drops by almost half to 20.09.

1104 ILR REVIEW

Although it may seem sensible to include percentage Hispanic as a con- trol variable, the use of this variable could be problematic. A lot of hetero- geneity occurs within the Hispanic population; the labor market opportunities available to low-skill Cubans probably have little in common with those available to low-skill Mexicans. Rows 2 and 3 show that the esti- mated wage effect can vary by 5 percentage points (in either direction) depending on whether the control variable is the fraction that is Hispanic or the fraction that is Cuban. The implied synthetic city in row 2, which uses percentage Hispanic, is an amalgam of Greensboro-Winston (with a weight of 0.44), Los Angeles (0.53), and San Diego (0.03). Simply replacing per- centage Hispanic with percentage Cuban changes the synthetic city to Greensboro-Winston (0.33), Newark (0.46), and Tampa (0.21). Not surpris- ingly, the implied synthetic city is very sensitive to the choice of control vari- ables, so careful consideration needs to be given to which set of variables should be included in the model.

Row 5 of Table 7 presents the most general specification, which adds vari- ables denoting the fraction of the workforce in each of 11 industries, as well as the fraction of the workforce that is female or low skill.33 The estimated

Table 7. Sensitivity of Short-Run Wage Effects to the Definition of a Synthetic Control

Men Men and women

Set of control variables March CPS CPS-ORG March CPS CPS-ORG

1. D log E, D log EU, D log wU 20.343 20.155 20.175 20.075 (0.081) (0.028) (0.051) (0.024)

2. D log E, percentage Hispanic, percentage black 20.367 20.092 20.264 20.057 (0.108) (0.032) (0.065) (0.023)

3. D log E, percentage Cuban, percentage black 20.317 20.134 20.251 20.070 (0.080) (0.039) (0.064) (0.039)

4. D log E, D log EU, D log wU, percentage Hispanic, percentage Cuban, percentage black, percentage female

20.301 20.072 20.221 20.057 (0.084) (0.037) (0.051) (0.023)

5. Same as row 4, plus percentage low-skill and industry mix

20.325 20.097 20.254 20.048 (0.052) (0.033) (0.044) (0.024)

Notes: Robust standard errors are reported in parentheses. Data consist of annual observations for each city between 1977 and 1986 (1980 excluded). The control variable D log E is the city’s employment growth rate between 1977 and 1979; D log EU is the concurrent growth rate of low-skill employment; and D log wU is the concurrent growth rate of the low-skill wage. The table reports the interaction coefficients between a dummy variable indicating if the metropolitan area is Miami and if the observation is drawn from the post-Mariel period. See the notes to Table 5 for more details on the regression specification.

33The industries are agriculture or mining, construction, manufacturing, transportation, wholesale or retail trade, finance, business and repair services, personal services, entertainment services, professional services, and public administration. The control variables giving the change in local employment or wages were calculated using the March CPS; all other variables were calculated using the 1980 decennial census.

WAGE IMPACT OF THE MARIELITOS 1105

wage impact is 20.33 in the March CPS and 20.10 in the ORG. Put bluntly, the construction of the placebo is not an innocuous decision—even when the methodology used to construct the placebo is thought to be relatively free of researcher intervention (as the synthetic control method is some- times advertised).

One alternative (and perhaps preferable) way of determining whether the key finding of a negative wage impact is sensitive to the choice of pla- cebo is to estimate the wage impact for every potential placebo, and then examine the resulting distribution of potential wage effects. I illustrate this approach by using the regression model in Equation (3) to estimate the short-run effect in each of the 123,410 possible four-city placebos.

The two panels of Figure 8 illustrate the frequency distribution of esti- mated effects when the dependent variable is the log wage of male high school dropouts, while Table 8 reports summary statistics for the distribu- tions. Consider the density of estimated effects in the March CPS data. The mean effect is 20.283, and almost all of the effects are statistically significant.

Note that if the set of placebos is restricted to those in which the average employment growth in the four placebo cities was roughly similar to that of pre-Mariel Miami, the mean wage effect rises to 20.319. If we look at the smaller subset, with both average employment growth and average low-skill employment growth similar to that of Miami, the mean wage effect rises fur- ther to 20.330. Put differently, the closer we get to a placebo that seems to replicate the pre-existing conditions in Miami, the more likely we are to find that the Marielitos had a larger wage effect on low-skill Miamians. Table 8 shows that the same trend is implied by the frequency distribution of wage effects computed in the ORG data.

Conclusion

Card’s (1990) classic article on the labor market impact of the Mariel boat- lift stands as a landmark study in labor economics. The finding that the sup- ply shock had little effect on the labor market opportunities of native workers influenced what we think we know about the economic conse- quences of immigration. And the elegance of the methodological approach—the exploitation of a fascinating natural experiment to estimate a parameter of economic interest—also influenced the way that many applied economists frame their questions, organize the data, and search for an answer.

This article brings a new perspective to the analysis of the Mariel supply shock. I revisit the question and the data armed with the insights provided by three decades of research on the economic impact of immigration. One key lesson from the vast literature is that the effect of immigration on the wage structure depends crucially on the differences between the skill distri- butions of immigrants and natives. The direct effect of immigration is most

1106 ILR REVIEW

likely to be felt by those workers who had similar capabilities as the Marielitos.

The Mariel supply shock was composed disproportionately of low-skill workers; approximately 60% were high school dropouts. Remarkably, none of the previous examinations of the Mariel experience documented what happened to the pre-existing high school dropouts in Miami, a group that composed over a quarter of the city’s workforce. Given the literature

Figure 8. Distribution of Short-Run Wage Impacts across All Possible Four-City Placebos, 1977–1986

Notes: The figure shows the distribution of the interaction term from the difference-in-differences regres- sion model in Equation (3) resulting from comparing Miami to all possible 123,410 placebos. Regressions use annual observations for each city in the period 1977–1986 (1980 excluded), and the coefficients measure the impact in the short run (i.e., 1981–1986). All regressions were weighted by the number of observations used to calculate the mean wage of high school dropouts in city r at time t.

WAGE IMPACT OF THE MARIELITOS 1107

sparked by Borjas (2003), any analysis of the Mariel supply shock should focus on the labor market outcomes of these low-skill workers.

The examination of wage trends among high school dropouts quickly overturns the stylized fact that the supply shock did not affect Miami’s wage structure. In fact, the absolute wage of high school dropouts dropped dra- matically, as did their wage relative to that of either high school graduates or college graduates. The drop in the average wage of the least-skilled Miamians between 1977–1979 and 1981–1986 was substantial, between 10 and 30%. The examination of wage trends in every other city identified by the CPS shows that the steep post-Mariel wage drop experienced by Miami’s low-skill workforce was a very unusual event.

The reappraisal presented in this article also illustrates that the research- er’s choice of a placebo is an important element of any such empirical exer- cise, and that picking a different placebo can easily lead to a weaker or stronger measured impact of immigration. The synthetic control method, for example, can generate disparate wage effects depending on the set of variables the researcher uses to construct the synthetic city. Similarly, the magnitude of the wage effect differs substantially across all potential four- city placebos that can be constructed in the CPS data. Note, however, that despite the variation in the magnitude of the wage effects across placebos, the evidence consistently indicates that the Marielitos had a sizable negative effect on the wage of competing workers.

The evidence has potentially important implications for the literature that purports to measure the wage impact of immigration. Many studies measure the effect by estimating spatial correlations between wages and the

Table 8. Distribution of Estimated Short-Run Wage Effects across All Four-City Placebos

Characteristics of distribution March CPS CPS-ORG

Mean 20.283 20.087 Standard deviation 0.043 0.022 Statistical significance: Fraction of t-statistics above |2.0| 0.998 0.845 Average employment growth of placebo cities within 0.1 standard

deviations of Miami (N = 1,148) Mean 20.319 20.108 Fraction of t-statistics above |2.0| 1.000 0.863

Average low-skill employment growth of placebo cities within 0.1 standard deviations of Miami (N = 1,096) Mean 20.297 20.105 Fraction of t-statistics above |2.0| 1.000 0.973

Average total and low-skill employment growth for placebo cities within 0.1 standard deviations of Miami (N = 74) Mean 20.330 20.120 Fraction of t-statistics above |2.0| 1.000 0.986

Notes: The table reports the distribution of the interaction coefficient between a dummy variable indicating if the metropolitan area is Miami and if the observation is drawn from the post-Mariel period. Regressions were estimated separately in all possible 123,410 four-city placebos. See the notes to Table 5 for more details on the regression specification.

1108 ILR REVIEW

number of immigrants in a particular locality. These spatial correlations are plagued both by endogeneity problems (i.e., immigrants settle in high-wage regions) and by native adjustments (i.e., firms and workers may respond to the supply shock by relocating to other cities). The fact that the spatial cor- relation implied by the Mariel supply shock is strongly negative indicates that the existing non-experimental literature may not have successfully over- come those statistical difficulties. Researchers still have some way to go before non-experimental spatial correlations can be presumed to estimate a parameter of economic interest.

The evidence also has implications for estimates of the economic benefits from immigration. The benefit that accrues to the native population, or the immigration surplus, is the flip side of the wage impact of immigration. The greater the wage impact, the greater the immigration surplus. Borjas (2014: 151) estimated the current surplus to be approximately 0.24% of GDP, or around $43 billion annually. Because there was a much larger reduction in the earnings of the workers most likely to be affected by the Marielitos than was previously believed, we may need to reassess existing estimates of the immigration surplus. That surplus could easily be two or three times as large if the Mariel context correctly measures the wage impact.

It has been a quarter-century since the publication of Card’s (1990) Mariel study. The reappraisal of the evidence presented in this article suggests that much can be gained by revisiting some of those persistent old questions with a new perspective, a perspective that uses the insights accumulated over the years. If nothing else, the reappraisal of the Mariel evidence shows that even the most cursory reexamination of some earlier data with some new ideas can reveal trends that radically change what we think we know.

References Abadie, Alberto, and Javier Gardeazabal. 2003. The economic costs of conflict: A case study

of the Basque Country. American Economic Review 93: 112–32. Abadie, Alberto, Alexis Diamond, and Jens Hainmueller. 2010. Synthetic control methods

for comparative case studies: Estimating the effect of California’s Tobacco Control Pro- gram. Journal of the American Statistical Association 105: 493–505.

Altonji, Joseph G., and David Card. 1991. The effects of immigration on the labor market outcomes of less-skilled natives. In John M. Abowd and Richard B. Freeman (Eds.), Immi- gration, Trade, and the Labor Market, pp. 201–34. Chicago: University of Chicago Press.

Angrist, Joshua D., and Alan B. Krueger. 1999. Empirical strategies in labor economics. In Orley Ashenfelter and David Card (Eds.), Handbook of Labor Economics, Vol. 3, pp. 1277–1366. Amsterdam: Elsevier.

Angrist, Joshua D., and Adriana D. Kugler. 2003. Protective or counter-productive? European labor market institutions and the effect of immigrants on EU natives. Economic Journal 113: F302–F331.

Autor, David H., Lawrence F. Katz, and Melissa S. Kearney. 2008. Trends in U.S. wage inequality: Revising the revisionists. Review of Economics and Statistics 90: 300–23.

Borjas, George J. 2003. The labor demand curve is downward sloping: Reexamining the impact of immigration on the labor market. Quarterly Journal of Economics 118: 1335–74.

———. 2014. Immigration Economics. Cambridge, MA: Harvard University Press. ———. 2015. The wage impact of the Marielitos: A reappraisal. NBER Working Paper No.

21588. Cambridge, MA: National Bureau of Economic Research.

WAGE IMPACT OF THE MARIELITOS 1109

———. 2016. The wage impact of the Marielitos: Additional evidence. NBER Working Paper No. 21850. Cambridge, MA: National Bureau of Economic Research.

Borjas, George J., and Kirk B. Doran. 2012. The collapse of the Soviet Union and the produc- tivity of American mathematicians. Quarterly Journal of Economics 127: 1143–1203.

Borjas, George J., Jeffrey Grogger, and Gordon H. Hanson. 2012. On estimating elasticities of substitution. Journal of the European Economic Association 10: 198–210.

Cameron, Colin, and Douglas L. Miller. 2015. A practitioner’s guide to cluster-robust infer- ence. Journal of Human Resources 50(2): 317–73.

Card, David. 1990. The impact of the Mariel Boatlift on the Miami labor market. Industrial and Labor Relations Review 43: 245–57.

Carrington, William J., and Pedro de Lima. 1996. The impact of 1970s repatriates from Africa on the Portuguese labor market. Industrial and Labor Relations Review 49: 33–47.

Dustmann, Christian, Tommaso Frattini, and Ian P. Preston. 2013. The effect of immigration along the distribution of wages. Review of Economic Studies 80(1): 145–73.

Dustmann, Christian, Uta Schönberg, and Jan Stuhler. 2017. Labor supply shocks and the dynamics of local wages and employment. Quarterly Journal of Economics, forthcoming.

Friedberg, Rachel. 2001. The impact of mass migration on the Israeli labor market. Quarterly Journal of Economics 116: 1373–1408.

Garthwaite, Craig, Tad Gross, and Matthew J. Notowidigdo. 2014. Public health insurance, labor supply, and employment lock. Quarterly Journal of Economics 129: 653–96.

Glitz, Albrecht. 2012. The labor market impact of immigration: A quasi-experiment exploit- ing immigrant location rules in Germany. Journal of Labor Economics 30: 175–213.

Grossman, Jean Baldwin. 1982. The substitutability of natives and immigrants in production. Review of Economics and Statistics 54: 596–603.

Hamilton, James D. 1983. Oil and the macroeconomy since World War II. Journal of Political Economy 91: 228–48.

Hunt, Jennifer. 1992. The impact of the 1962 repatriates from Algeria on the French labor market. Industrial and Labor Relations Review 45: 556–72.

Jaeger, David, Joakim Ruist, and Jan Stuhler. 2016. Shift-share instruments and the impact of immigration. Working Paper. New York: Graduate Center, City University of New York.

Lemieux, Thomas. 2006. Increased residual wage inequality: Composition effects, noisy data, or rising demand for skill. American Economic Review 96(June): 461–98.

Llull, Joan. 2015. The effect of immigration on wages: Exploiting exogenous variation at the national level. Barcelona GSE Working Paper 783. Barcelona, Spain: Graduate School of Economics.

Monras, Joan. 2014. Appendix to Immigration and wage dynamics: Evidence from the Mexi- can peso crisis. Sciences Po, January 8.

———. 2015. Immigration and wage dynamics: Evidence from the Mexican peso crisis. Sciences Po Economics Discussion Papers 2015-4. Paris, France: Department of Economics.

Ottaviano, Gianmarco I. P., and Giovanni Peri. 2012. Rethinking the effect of immigration on wages. Journal of the European Economic Association 10: 152–97.

Peri, Giovanni, and Vasil Yasenov. 2015. The labor market effects of a refugee wave: Applying the synthetic control method to the Mariel Boatlift. NBER Working Paper No. 21801. Cambridge, MA: National Bureau of Economic Research.

Pinotti, Paolo, Ludovica Gazzè, Francesco Fasani, and Marco Tonello. 2013. Immigration pol- icy and crime. Report prepared for the XV European Conference of the Fondazione Rodolfo Debenedetti. June 22, 2013; Caserta, Italy.

Saiz, Albert. 2003. Room in the kitchen for the melting pot: Immigration and rental prices. Review of Economics and Statistics 85: 502–21.

Stabile, Benedict L., and Robert L. Scheina. 2015. U.S. Coast Guard operations during the 1980 Cuban exodus. United States Coast Guard, Coast Guard History. Accessed at http:// www.uscg.mil/history/articles/uscg_mariel_history_1980.asp (September 30, 2015).

Vogel, Ronald K., and Genie N.L. Stowers. 1991. Miami: Minority empowerment and regime change. In H. V. Savitch and John Clayton Thomas (Eds.), Big City Politics in Transition, pp. 115–31. New York: Sage Publications.

1110 ILR REVIEW