the impact of the mariel boatlift on the miami labor …

THE IMPACT OF THE MARIEL BOATLIFT ON THE MIAMILABOR MARKET

DAVID CARD*

Using data from the Current Population Survey, this paper describesthe effect of the Mariel Boatlift of 1980 on the Miami labor market. TheMariel immigrants increased the Miami labor force by 7%, and thepercentage increase in labor supply to less-skilled occupations andindustries was even greater because most of the immigrants wererelatively unskilled. Nevertheless, the Mariel influx appears to have hadvirtually no effect on the wages or unemployment rates of less-skilledworkers, even among Cubans who had immigrated earlier. The authorsuggests that the ability of Miami's labor market to rapidly absorb theMariel immigrants was largely owing to its adjustment to other largewaves of immigrants in the two decades before the Mariel Boatlift,

ONE of the chief concerns of immigra-tion policy-makers is the extent to

which immigrants depress the labor mar-ket opportunities of less-skilled natives.Despite the presumption that an influx ofimmigrants will substantially reduce nativewages, existing empirical studies suggestthat the effect is small, (See the survey byGreenwood and McDowell [1986] andstudies by Grossman [1982], Borjas [1987],and Lalonde and Topel [1987].) There aretwo leading explanations for this finding.First, immigrants have, on average, onlyslightly lower skills than the native popu-lation. Thus, econometric studies based onthe distribution of the existing stock ofimmigrants probably understate the effectof unskilled immigration on less-skilled

* The author is Professor of Economics, PrincetonUniversity, He thanks George Borjas, Alan Krueger,Bruce Meyer, and seminar participants at PrincetonUniversity for their comments,

A data appendix with copies of the computerprograms used to generate the tables in this paper isavailable from the author at the Industrial RelationsSection, Firestone Library, Princeton University,Princeton, NJ 08544,

natives. Second, the locational choices ofimmigrants and natives presumably de-pend on expected labor market opportu-nities. Immigrants tend to move to citieswhere the growth in demand for labor canaccommodate their supply. Even if newimmigrants cluster in only a few cities (asthey do in the United States), inter-citymigration of natives will tend to offset theadverse effects of immigration.

These considerations illustrate the diffi-culty of using the correlation across citiesbetween wages and immigrant densities tomeasure the effect of immigration on thelabor market opportunities of natives. Theyalso underscore the value of a natural ex-periment that corresponds more closely toan exogenous increase in the supply of im-migrants to a particular labor market.

The experiences of the Miami labormarket in the aftermath of the MarielBoatlift form one such experiment. FromMay to September 1980, some 125,000Guban immigrants arrived in Miami on aflotilla of privately chartered boats. Theirarrival was the consequence of an unlikelysequence of events culminating in Gastro's

Industrial and Labor Relations Review, Vol, 43, No, 2 (January 1990), © by Cornell University,0019-7939/90/4302 $01,00

245

246 INDUSTRIAL AND LABOR RELATIONS REVIEW

declaration on April 20, 1980, that Gubanswishing to emigrate to the United Stateswere free to leave from the port ofMariel.' Fifty percent of the Mariel immi-grants settled permanently in Miami, Theresult was a 7% increase in the labor forceof Miami and a 20% increase in thenumber of Guban workers in Miami,

This paper summarizes the effects ofthe Boatlift on the Miami labor market,focusing on wages and unemploymentrates of less-skilled workers. The analysisis based on individual micro-data for1979-85 from the merged outgoing rota-tion group samples of the Gurrent Popu-lation Survey (GPS).

Three features of the Mariel incidentand the Gensus data greatly facilitate theanalysis. First, the GPS sample of theMiami metropolitan area is relatively large:roughly 1,200 individuals per month.Second, a comprehensive picture of theMiami labor market in the months justbefore the Mariel Boatlift is available fromthe 1980 Gensus, which was conducted onApril 1, 1980, Finally, unlike most otherethnic groups, Gubans are separatelyidentified in the GPS questionnaire. Thus,it is possible to estimate wage rates,unemployment rates, and other economicindicators for both Gubans and non-Gubans in the Miami labor market, and tomeasure the effects of the Mariel immigra-tion on the two groups separately.

Overview of the Miami LaborMarket Before the Boatlift

For at least a decade prior to the MarielBoatlift, Miami was the most immigrant-intensive city in the country. Tabulationsfrom the 1980 Gensus indicate that 35,5%of residents in the Miami Standard Metro-politan Statistical Area (SMSA) were for-eign-born,2 compared to 22.3% in Los

' See Masud-Piloto (1988, chaps, 6-7) for anoverview of the political developments that led to theMariel Boatlift,

^ See U,S, Department of Commerce (1983), TheMiami SMSA consists of Dade County, and includesMiami City as well as a number of smaller towns andcities. Throughout this paper, I use "Miami" to referto this broader geographic region.

Angeles, the city with the next-highestimmigrant fraction, and 6,1% nationwide.At the time of the Gensus, 56% ofimmigrants in Miami were of Gubanorigin. The remaining foreign-born resi-dents, who accounted for 16% of theMiami population, included other His-panic groups and a broad selection ofGaribbean and European nationals,

Miami also has a significant black popu-lation. The fraction of black residents was15.0% in 1970 and had increased to 17,3%by the time ofthe 1980 Gensus, The largeconcentrations of both immigrants andblacks makes Miami ideal for studying theeffect of increased immigration on thelabor market opportunities of black na-tives.

Table 1 describes the four major groupsin the Miami labor force in 1979: whitenon-Hispanics; black non-Hispanics; Gu-bans (foreign-born and native-born); andother Hispanics, For simplicity, I haverestricted attention to individuals age16-61, a group that represents roughly60% of the Miami population. The frac-tions of Gubans and blacks in the 16-61age group are 27,2% and 26,3%, respec-tively, and the fractions of white non-Hispanics and non-Guban Hispanics are34,4% and 11.1%, Overall, 73% of 16-61-year-olds participated in the laborforce, with somewhat higher rates amongwhites and Gubans and lower rates amongblacks and other Hispanics, Educationlevels in Miami are somewhat below thenational average: the mean of completededucation for 16-61-year-olds in 1979 was11,8 years in Miami, compared with 12,2years nationwide.

The occupation distributions in rows7-17 of Table 1 give a crude indication ofthe degree of labor market competitionamong the four groups, Gubans and otherHispanics have very similar occupationdistributions, with a higher representationin craft and operative occupations thaneither whites or blacks. Blacks are morehighly concentrated in laborer and service-related occupations, and are significantlyunder-represented in managerial occupa-tions,

A useful summary measure of the

THE MARIEL BOATLIFT 247

Table 1. Characteristics of 16-61-Year-Olds in Miami, 1979,

Characteristic

Characteristics of Population Age 16-61

I, Estimated Number (lOOO's)2, Mean Education3, Percent in Labor Force

Characteristics of Those in Labor Force

4, Estimated Number (lOOO's)5, Mean Education6, Percent Age 16-24

Occupation Distribution (Percent of Employed)

1. Professional and Technical8, Managers9, Sales

10, Clerical11, Craftsmen12, Operatives13, Transportation Operatives14, Laborers15, Farm Workers16, Less-Skilled Service Workers17, More-Skilled Service Workers

Whites

319,312,875,6

241,313,121,1

19,115,76,2

21,913,34,42,65,11,15,05,7

Blacks

244,111,468,3

166,611,824,1

10,92,84,4

21,09,48,48,1

10,50,1

13,310,9

Cubans

252,411,077,2

194,711,322,0

9,58,67,8

19,115,119,45,44,70,46,14,0

Hispanics .

102,911,668,8

70,811,926,0

10,18,17,6

20,912,716,75,94,00,8

10,23,0

All

928,411,873,1

678,212,122,8

13,29,46,5

20,912,811,15,26,30,68,06,2

Notes: White and black groups exclude bispanics, Hispanic group includes all hispanics other than Cubans,Less-skilled service workers include cleaning and food service workers. More-skilled service workers includehealth service, personal service, and protective service workers.

Source: Based on samples of employed workers in the outgoing rotation groups of the Current PopulationSurvey in 1979,

overlap in the occupation distributions ofthe different groups is the average per-cent increase in labor supply in occupa-tions held by one group that would resultfrom a one percentage point increase inthe overall fraction of workers in a secondgroup,^ This index has the simple formSy5].52, / Sj, where 5,̂ is the fraction ofworkers or̂ group 1 in occupation j , S2j isthe fraction of workers of group 2 inoccupation j , and 5, is the fraction of allworkers in occupation j . Based on thedistributions in Table 1, an inflow ofimmigrants resulting in a one percentagepoint increase in the fraction of Gubans inMiami would lead to a weighted averageincrease of ,95% in the supply of labor tooccupations held by whites. Under thesame conditions the increase would be,99% for blacks, 1,02% for non-GubanHispanics, and 1,06% for Gubans them-selves. These calculations suggest that theoverlap between the occupational distribu-tions of the four groups is relatively high.

^ This index is derived in Altonji and Card(1989:15-16),

The Mariel Immigration

Due to the unauthorized nature of theBoatlift, no exact count of the number ofMariel immigrants is available, and thereis little precise information on the charac-teristics or final destinations of the immi-grants. This section summarizes some ofthe available information, including datafrom the March 1985 Mobility Supple-ment to the Gurrent Population Survey,which allows Mariel immigrants to bedistinguished from other Gubans.

Most sources estimate the number ofMariel immigrants at between 120,000and 125,000, A recent Gensus Bureaureport (U,S, Department of Gommerce1988:9) states that 126,000 refugees en-tered the United States as "Guban En-trants" (the special immigration statusawarded to the Mariel refugees) betweenApril 1980 and June 1981. Based on thesettlement pattern of earlier Gubans, it iswidely assumed that about one-half ofthese refugees settled permanently inMiami, The Gensus Bureau "Experimen-tal Gounty Population Estimates" file


shows an increase of 80,500 in the DadeCounty population between April 1 andJuly 1 of 1980; 59,800 of these newentrants were age 16-6L My own tabula-tions from the CPS indicate that theCuban share of the 16-61 age groupincreased from 27% in 1979 to 33% in1981.'* A similar increase is registered inCPS-based estimates of the Cuban share ofthe 16-61-year-old labor force, whichchanged from 37.2% in 1979 to 44.8% in1981. Assuming that the Cuban share ofthe labor force would have remainedconstant between 1979 and 1981 in theabsence of the Boatlift, these figuressuggest that the Mariel immigration addedapproximately 45,000 to the Miami laborforce—an increase of 7%.

From the first days of the Boatlift, thecharacteristics of the Mariel immigrants(hereafter referred to as Mariels) havebeen a subject of controversy. Amongthose who were permitted to leave Cubawere several hundred inmates of mentalhospitals and jails. Many of these individ-uals were arrested by immigration officialsupon their arrival in the United States,and over 1,000 were sent to a specialprison facility in Atlanta to await deporta-tion back to Cuba.5 A similar number werearrested for crimes committed in theUnited States, and they still await determi-nation of their ultimate immigration sta-tus.6 Contemporary reports indicate thatthe Mariels included a relatively highfraction of less-skilled workers and a highfraction of individuals with low Englishability {Business Week 1980).

Although the regular Current Popula-tion Survey questionnaire does not distin-guish Mariels from other foreign and

•* These tabulations are presented in greater detailin an earlier version of this paper (Card 1989, Table

^ See Masud-Piloto (1988:100-103). Under a 1984agreement a total of 2,700 Mariel immigrants were tobe returned to Cuba.

® Mariel immigrants were blamed for, and indeedseem to have committed, many crimes in the first fewmonths after the Boatlift. Wilbanks (1984) reportedthat 38 of the 574 homicides in Miami in 1980 werecommitted by Mariel immigrants. Disaffected Marielswere involved in 6 airline highjacking attempts inAugust 1980. See Masud-Piloto (1988:95-96).

native-born Cubans, the March 1985 Mo-bility Supplement survey asked each re-spondent where he or she lived in March1980 (one month before the start of theBoatlift). Table 2 presents a descriptivesummary of the Cuban population inter-viewed in this survey, classified by whetherthe respondent claimed to be living abroador in the United States five years earlier.The sample sizes, particularly of post-1980entrants, are small.'' Nevertheless, thesedata confirm the general impression thatMariels, on average, have less education,are somewhat younger, and are morelikely to be male than other Cubans. Thefigures in Table 2 also suggest that theMariels have lower labor force attachmentand lower occupational attainment thanother Cubans. Mariels are more heavilyconcentrated in laborer and service occu-pations, and are less likely to hold sales,clerical, and craft jobs.

The unadjusted wage gap betweenMariels and other Cubans is 34%. Part ofthis differential is clearly attributable tothe lower education levels and youngerages of the Mariels. A simple linearregression for the logarithm of averagehourly earnings fitted to the sample ofCubans with earnings in 1984 suggeststhat the Mariels earned 18% lower wagesthan other Cubans, controlling for educa-tion, potential experience, and gender(the standard error of this estimate is .08).This gap presumably reflects the combina-tion of lower language ability and ashorter assimilation time in the UnitedStates among the Mariels, as well as anydifferences in ability or motivation be-tween the earlier and later Cuban immi-grants.

The Effect of theMariel Immigration on the

Miami Labor Market

Observers in Miami at the time of theBoatiift noted the strain caused by theMariel immigration. The homicide rate

' The weighted count of all Cubans in the March1985 CPS who entered the United States after 1980is 85,800, which is only 69% of the estimated 125,000Mariel refugees.


Table 2. Characteristics of Mariel Immigrantsand Other Cubans: Tabulations from March

1985 CPS.

Characteristic

Educational Attainment(Percent of Population in Each

Category):No High SchoolSome High SchoolCompleted High SchoolSome CollegeCompleted College

Percent MalePercent Under 30 in 1980Mean Age in 1980 (Years)Percent in Miami in 1985Percent Worked in 1984Mean Log Hourly Earnings

MarielImmigrants

56.59.19.56.8

18.155.638.734.953.960.6

1.37Occupation Distribution (Percent

Employed in Each Category):Professional/ManagersTechnicalSalesClericalCraftsmenOperativesTransportation Ops.LaborersFarm WorkersLess-Skilled ServiceMore-Skilled Service

Sample SizeWeighted Count

19.30.04.52.59.5

19.13.8

10.80.0

26.04.6

5042,300 '

All otherCubans

25.413.333.412.015.850.729.638.052.4-73.4

1.71

21.01.5

11.213.519.913.84.33.31.87.42.3

528176,900

Note: The sample consists of all Cubans in theMarch 1985 Current Population Survey age 21-66(i.e., age 16—61 in 1980). Mariel immigrants areidentified as those Cubans who stated that they livedoutside the United States 5 years previously.

increased nearly 50% between 1979 and1980 (see Wilbanks 1984:142). On theweekend of May 17, 1980, a three-day riotoccurred in several black neighborhoods,killing 13. A government-sponsored com-mittee that was set up to investigate theriot identified other long-standing griev-ances in the black community as its cause,but cited the labor market competition ofCuban refugees as an important back-ground factor (Governor of Florida 1980:14-15).

Another widely cited indicator of thelabor market pressure created by theMariel influx is the Miami unemploymentrate, which rose from 5.0% in April 1980

to 7.1% in July. Over the same periodstate and national unemployment ratesfollowed a similar pattern, suggestingthat the changes in Miami were notsolely a response to the Mariel influx.Nevertheless, widespread joblessness ofrefugees throughout the summer of 1980contributed to a perception that labormarket opportunities for less-skilled na-tives were threatened by the Mariel immi-

Tables 3 and 4 present simple averagesof wage rates and unemployment rates forwhites, blacks, Cubans, and other Hispan-ics in the Miami labor market between1979 and 1985. For comparative pur-poses, I have assembled similar data forwhites, blacks, and Hispanics in four othercities: Atlanta, Los Angeles, Houston, andTampa-St. Petersburg. These four citieswere selected both because they hadrelatively large populations of blacks andHispanics and because they exhibited apattern of economic growth similar to thatin Miami over the late 1970s and early1980s. A comparison of employmentgrowth rates (based on establishment-leveldata) suggests that economic conditionswere very similar in Miami and theaverage of the four comparison citiesbetween 1976 and 1984.

The wage data in Table 3 reveal severalfeatures of the Miami labor market.Perhaps most obvious is that earnings arelower in Miami than in the comparisoncities. The differentials in 1979 rangedfrom 8% for whites to 15% for blacks.More surprising is that real earnings levelsof whites in both Miami and the compari-son cities were fairly constant between1979 and 1985. This pattern contrastswith the general decline in real wages inthe U.S. economy over this period (seeBound and Johnson 1989:5-6) and under-scores the relatively close correspondencebetween economic conditions in Miamiand the comparison cities.

^ For example, a Florida State Employment Ser-vice official and a Department of Labor Wage andHours Division official noted downward pressure onwages and working conditions in the unskilledsegment of the Miami labor market (Business Week1980).


Table 3. Logarithms of Real Hourly Earnings of Workers Age 16-61 in Miami and FourComparison Cities, 1979-85.

Croup

Miami:

Whites

Blacks

Cubans

Hispanics

Comparison Cities:

Whites

Blacks

Hispanics

1979

1.85(.03)1.59(.03)1.58(.02)1.52(.04)

1.93(.01)1.74(.01)1.65(.01)

1980

1.83(.03)1.55(.02)

1.54(-02)

1.54(.04)

1.90(.01)1.70(.02)1.63(.01)

1981

1.85(.03)1.61(.03)1.51(.02)

1.54(.05)

1.91(.01)1.72(.02)1.61(.01)

1982

1.82(.03)1.48(.03)1.49(.02)

1.53(.05)

1.91(.01)1.71(.01)1.61(.01)

1983

1.82(.03)1.48(.03)1.49(.02)1.48(.04)

1.90(.01)1.69(.02)1.58(.01)

1984

1.82(.03)1.57(.03)1.53(.03)1.59(.04)

1.91(.01)1.67(.02)1.60(.01)

1985

1.82(.05)1.60(.04)1.49(.04)1.54(.06)

1.92(.01)1.65(.03)1.58(.02)

Note: Entries represent means of log hourly earnings (deflated by the Consumer Price lndex-1980= 100)for workers age 16-61 in Miami and four comparison cities: Atlanta. Houston, Los Angeles, and Tampa-St.Petersburg. See note to Table 1 for definitions of groups.

Source: Based on samples of employed workers in the outgoing rotation groups of the Current PopulationSurvey in 1979-85. Due to a change in SMSA coding procedures in 1985, the 1985 sample is based onindividuals in outgoing rotation groups for January-June of 1985 only.

In contrast to the pattern for whites, thetrends in earnings for nonwhites andHispanics differ somewhat between Miamiand the comparison cities. Black wages inMiami were roughly constant from 1979to 1981, fell in 1982 and 1983, and rose totheir previous level in 1984. Black earn-ings in the comparison cities, on the otherhand, show a steady downward trendbetween 1979 and 1985. These dataprovide no evidence of a negative impactof the Mariel immigration on black wagesin Miami. The data do suggest a relativedownturn in black wages in Miami during1982—83. It seems likely, however, thatthis downturn reflects an unusually severecyclical effect associated with the 1982-83recession. (I return to this issue in Table 6,below.)

Wage rates for non-Cuban Hispanics inMiami were fairly stable between 1979 and1985, with only'a slight dip in 1983. Incontrast, Hispanic wage rates in thecomparison cities fell about 6 percentagepoints over this period. Again, there is noevidence of a negative effect in Miami,

either in the immediate post-Mariel periodor over the longer run.

Table 3 does indicate a decline in Cubanwage rates relative to the wage rates ofother groups in Miami. Relative to thewages of whites, for example, Cubanwages fell by 6-7 percentage points be-tween 1979 and 1981. Assuming that thewages of earlier Cuban immigrants wereconstant, this decline is consistent with theaddition of 45,000 Mariel workers to thepool of Cubans in the Miami labor force,and with the 34% wage differential be-tween Mariels and other Cubans noted inTable 3. A more thorough analysis ofCuban wages is presented in Table 7,below.

The unemployment rates in Table 4lead to the same general conclusions as thewage data in Table 3. There is no evidencethat the Mariel influx adversely affectedthe unemployment rate of either whites orblacks. The unemployment rates suggest asevere cyclical downturn in the black labormarket in Miami in 1982-83. Black unem-ployment rates in Miami, which had been


Table 4. Unemployment Rates of Individuals Age 16—61 in Miami andFour Comparison Cities, 1979—85.(Standard Errors in Parentheses)

Croup

Miami:

Whites

Blacks

Cubans

Hispanics

Comparison Cities:

Whites

Blacks

Hispanics

1979

5.1(1.1)8.3

(1.7)5.3

(1.2)

6.5(2.3)

4.4(0.3)10.3(0.8)6.3

(0.6)

1980

2.5(0.8)5.6

(1.3)7.2

(1.3)7.7

(2.2)

4.4(0.3)12.6(0.9)8.7

(0.6)

1981

3.9(0.9)9.6

(1.8)10.1(1.5)11.8(3.0)

4.3(0.3)12.6(0.9)8.3

(0.6)

1982

5.2(1.1)16.0(2.3)10.8(1.5)9.1

(2.5)

6.8(0.3)12.7(0.9)12.1(0.7)

1983

6.7(1.1)18.4(2.5)13.1(1.6)7.5

(2.1)

6.9(0.3)18.4(1.1)11.8(0.7)

1984

3.6(0.9)14.2(2.3)7.7

(1.4)12.1(2.4)

5.4(0.3)12.1(0.9)9.8

(0.6)

1985

4.9(1.4)7.8

(2.3)

5.5(1.7)3.7

(1.9)

4.9(0.4)13.3(1.3)9.3

(0.8)Note: Entries represent means of unemployment indicator variable for individuals age 16-61 in Miami and

four comparison cities: Atlanta, Houston, Los Angeles, and Tampa-St. Petersburg. Samples are based onindividuals in the labor force. See notes to Table 3 for definitions of groups and data sources.

2-4 pointscomparison cities

lower than those in thefrom 1979 to 1981,

equalled or exceeded those in the compar-ison cities from 1982 to 1984. The 1985data indicate a return to the pre-1982pattern, although the sampling errors arelarge enough to prevent precise infer-ences.

Unlike the situation for whites andblacks, there was a sizable increase inCuban unemployment rates in Miamifollowing the Mariel immigration. Cubanunemployment rates were roughly 3 per-centage points higher during 1980—81than would have been expected on thebasis of earlier (and later) patterns. Assum-ing that the unemployment rates of earlierCuban immigrants were unaffected by theMariel influx, this effect is consistent withunemployment rates of around 20%among the Mariels themselves. Althoughfar from conclusive, this simple calculationsuggests that the increase in Cuban unem-ployment rates could easily be explainedas a result of the addition of the Marielrefugees to the Cuban population, withlittle or no effect on earlier immigrants.

The simple averages of wages and

unemployment rates in Tables 3 and 4,which combine workers of all ages andeducation levels, do not directly addressthe question of whether the Mariel immi-gration reduced the earnings of less-skilled natives in Miami. A more directanswer is provided by the data in Table 5.In order to identify "less-skilled" workers,I fit a linear regression equation for thelogarithm of hourly earnings to workers inthe comparison cities. The explanatoryvariables in this regression included edu-cation, potential experience, squared po-tential experience, indicator variables foreach gender and race group, and interac-tions of the gender-race indicators withpotential experience and squared poten-tial experience. I then used the estimatedcoefficients from this equation to form apredicted wage for each non-Cuban workerin Miami, and sorted the sample fromeach year into quartiles on the basis ofpredicted wage rates.

This procedure provides a simple wayto identify more- and less-skilled workersin the Miami labor market. Means ofactual log wages for each quartile and yearare presented in the first four columns of


Table 5. Means of Log Wages of Non-Cubans in Miami by Quartileof Predicted Wages, 1979-85.

(Standard Errors in Parentheses)

Year

1979

1980

1981

1982

1983

1984

1985

1st Quart.

1.31(.03)

1.31(.03)

1.40(.03)

1.24(.03)

1.27(.03)

1.33(.03)

1.27(.04)

Mean of Log Wage by

2nd Quart.

L61(.03)

1.52(.03)

1.57(.03)

1.57(.03)

1.53(.04)

1.59(.04)

1.57(.04)

Quartile of Predicted Wage

3rd Quart.

1.71(.03)

1.74(.03)

1.79(.03)

1.77(.03)

1.76(.03)

1.80(.04)

1.81(.04)

4th Quart.

2.15(.04)

2.09(.04)

2.06(.04)

2.04(.04)

2.11(.05)

2.12(.04)

2.14(.05)

Difference ofMeans: 4th - 1st

.84(.05)

.77(.05)

.66(.05)

.80(.05)

.84(.06)

.79(.05)

.87(.06)

Note: Predicted wage is based on a linear prediction equation for the log wage fitted to individuals in fourcomparison cities; see text. The sample consists of non-Cubans (male and female, white, black, and Hispanic)between the ages of 16 and 61 with valid wage data in the earnings supplement of the Current PopulationSurvey. Wages are deflated by the Consumer Price Index (1980= lOO).

Table 5. The difference in mean wagesbetween the first and fourth quartiles,which provides an index of the spread inthe wage distribution, is presented in thefifth column of the table.

If the Mariel immigration reduced thewages of less-skilled natives, one wouldexpect to observe a decline in the wage ofworkers in the lowest skill quartile, at leastrelative to workers in the upper quartile.The actual averages show no evidence ofthis effect. Apart from the temporaryincrease in relative wages of workers in thelowest quartile between 1979 and 1981,the distribution of non-Cubans' wages inthe Miami labor market was remarkablystable between 1979 and 1985. Takentogether with the data in Table 3, thesedata provide little evidence of a negativeeffect of the Mariel influx on the earningsof natives.

A final check is provided in Table 6,which contains more detailed informationon wages, employment rates, and unem-ployment rates for blacks in Miami be-tween 1979 and 1985. I separately ana-lyzed the set of all blacks and the set ofblacks with less than 12 years of educationin order to isolate any stronger effect on

the less-skilled segment of the blackpopulation. For both groups I calculatedthe differential in wages between Miamiand the comparison cities (both the unad-justed difference in mean log ways and aregression-adjusted differential that con-trols for education, gender, marital status,part-time status, private/public employ-ment, and potential experience) and thedifferentials in the employment-popula-tion rate and the unemployment ratebetween Miami and the comparison cities.

As indicated in Table 3, the wagedifferential for blacks in Miami relative tothose in the comparison cities decreasedslightly between 1979 and 1981. Thedifferential increased substantially in 1982,but then began a steady downward trendafter 1983. By 1985, the wage gap was lessthan 5% for all black workers, and wasactually positive for less-educated blacks.The magnitudes of the regression-adjusted wage differentials are not signifi-cantly different from the unadjusted wagedifferentials, and show no evidence of anyeffect of the Mariel immigration on blackwages.

A similar conclusion emerges from thepattern of differentials in employment-


Table 6. Comparison of Wages, Unemployment Rates, and Employment Rates for Blacks inMiami and Comparison Cities.


Year

1979

1980

1981

1982

1983

1984

1985

All

Difference inLog

Miami -

Actual

- . 1 5(.03)

- . 1 6(.03)

- . 1 1(03)

- . 2 4(.03)

- . 2 1(.03)

- . 1 0(.03)

- . 0 5(.04)

• Wages,

• Comparison

Adjusted

- . 1 2(.03)

- . 1 2(.03)

- . 1 0(.03)

- . 2 0(.03)

- . 1 5(.03)

- . 0 5(.03)

- . 0 1(.04)

Backs

Difference inEmp./Unemp.,

Miami -

Emp. -Pop. Rate

.00(.03)

.05(.03)

.02(.03)

- . 0 6(.03)

- . 0 2(.03)

- . 0 4(.03)

- . 0 6(.04)

• Comparison

Unemp. Rate

- 2 . 0(1.9)

- 7 . 1(1.6)

- 3 . 0(2.0)

3.3(2.4)

.1(2.7)

2.1(2.4)

- 5 . 5(2.6)

Low-Education Blacks

Difference inLog

Miami —

Actual

- . 1 3(.05)

- . 0 7(.05)

- . 0 5(.05)

- . 1 7(.05)

- . 1 3(.06)

- . 0 4(.06)

.18(.07)

Wages,

Comparison

Adjusted

- . 1 5(.05)

- . 0 7(.05)

- . 1 1(.05)

- . 2 0(.05)

- . 1 1(.05)

- . 0 3(.05)

.09(.07)

Difference inEmp./Unemp.,

Miami —

Emp. —Pop. Rate

.03(.04)

.03(.04)

.04(.04)

- . 0 4(.04)

.04(.04)

.05(.04)

.00(.06)

Comparison

Unemp. Rate

- . 8(3.8)

- 8 . 2(3.5)

- 7 . 7(4.2)

.6(4.7)

- 3 . 3(4.7)

.1(4.7)

- 4 . 7(5.6)

Notes: Low-education blacks are those with less than 12 years of completed education. Adjusted differences inlog wages between blacks in Miami and comparison cities are obtained from a linear regression model thatIncludes education, potential experience, and other control variables; see text. Wages are deflated by theConsumer Price Index (1980=100). "Emp.-Pop. Rate" refers to the employment:population ratio. "Unemp.Rate" refers to the unemployment rate among those in the labor force.

population ratios and unemployment rates.^Among all blacks, there is some evidenceof a relative decline in the employment-to-population ratio in Miami between 1979and 1985.'° This effect seems to havestarted in 1982, and is less pronouncedamong low-education blacks than amongthose with more education. As noted in Ta-

^ 1 also computed regression-adjusted employment-population and unemployment gaps using simplelinear probability models. The explanatory power ofthe statistical models is so low, however, that theadjusted differentials are almost identical to theunadjusted differentials.

'" Although they are not reported in Table 6, Ihave also constructed differentials in the labor forceparticipation rate between Miami and the compari-son cities. For blacks as a whole these show a declinein relative participation rates in Miami starting in1982, although the decline is only temporary for thelow-education group. The differential in labor forceparticipation rates is approximately equal to thedifferential in the employment-population rate plusthe differential in the unemployment rate multipliedby the average labor force participation rate (.7 forthe overall group, .55 for the low-education group).

ble 4, the series of unemployment rate dif-ferentials indicates a sharp downturn in la-bor market opportunities for blacks in 1982.Given the lag between the arrival of theMariels and the emergence of this unem-ployment gap, however, the gap seemsmore likely to have resulted from the 1982recession than from the influx of less-skilled immigrants.

The effects of the Mariel immigrationon Cuban labor market outcomes areexamined in detail in Table 7. The firstcolumn of the table reproduces the meansof log wages in each year from the thirdrow of Table 3. The second column givespredicted log wages of Cubans in Miami,using estimated coefficients from a regres-sion equation fit to Hispanics in the fourcomparison cities. The gap between actualand predicted wages is presented in thethird column of the table. These seriesshow that the 9 percentage point declinein Cuban real wage rates in Miamibetween 1979 and 1985 was a result of two


Table 7. Means of Log Wages of Cubans in Miami: Actual and Predicted,and by Quartile of Predicted Wages.


Year

1979

1980

1981

1982

1983

1984

1985

Mean of Log WagesLog in Miami

Actual

1.58(.02)1.54(.02)1.51(.02)1.49(.02)1.48(.03)1.53(.03)1.49(.04)

Pre-

dicted

1.73(.02)1.68(.02)1.68(.02)1.68(.02)1.65(.02)1.69(.02)1.67(.03)

Actual-Pre-

dicted

- .15(.03)

- .14(.03)

- .17(.03)

- .19(.03)

- .17(.03)

- .17(.03)

- .18(.05)

1st

1.31(.02)1.25(.02)1.23(.03)1.27(.03)1.16(.02)1.20(.03)1.19(.06)

Mean ofl^gWages by Quarttteof Predicted Wages

2nd

1.44(.03)1.49(.05)1,43(.03)1.43(.04)1.41(.04)1.40(.04)1.43(.06)

3rd

1.64(.04)1,59(.04)1.55(.04)1,50(.04)1,56(.04)1,65(.05)1,53(.08)

4th

1,90(.05)1,81(.05)1,80(.05)1,77(.06)1,80(.06)1,88(.06)1,80(.09)

Mean Log Wageof Cubans

Outside Miami

1,71(.04)

1,66(.03)1,63(.03)1,71(.03)1,62(.03)1,63(.03)1,77(.06)

Difference in

Miami - Rest-of-U.S.

Actual

- ,13(.04)

- ,12(.04)

- , 13(.04)

- ,22(.04)

- ,14(.04)

- ,10(.04)

- ,27(.07)

Ad-justed

- ,10(.04)

- ,06(.03)

- .09(.03)

- ,12(.03)

- ,08(.03)

- ,08(.03)

- ,19(.05)

Notes: Predicted wage is based on a linear prediction equation for the log wage fitted to individuals in fourcomparison cities; see text. Predicted wages for Cubans in Miami are based on coefficients for Hispanics incomparison cities. The adjusted wage gap between Cubans in Miami and Cubans in the rest of the U.S.obtained from a linear regression model that includes education, potential experience, and other corvariables; see text. Wages are deflated by the Consumer Price Index (1980= 100).

arecontrol

complementary factors: a 6% relativedecline in the "quality" of the Cuban laborforce in Miami, as measured by the declinein their predicted wages, and a 3 percent-age point increase in the quality-adjustedwage gap between Cuban workers inMiami and Hispanic workers in the com-parison cities. Two-thirds of the wagedecline is therefore attributed to tbechanging productivity cbaracteristics ofthe Cuban labor force, and one-third to adecrease in the return to skills for Cubansin the Miami labor market.

The next four columns of Table 7 givethe means of log wages for Cuban workersin each quartile of tbe distribution ofpredicted wages (using the same predic-tion equation as was used to form tbemeans in column 2). Tbese means suggestthat real wage rates of Cubans in tbelowest quartile of the wage distributiondeclined by 11-12 percentage points be-tween 1979 and 1985. Tbe decline issmaller for workers in tbe bigber quar-tiles, but there is some variation between

1984 and 1985, and in ligbt of tbesampling errors it is difficult to drawprecise inferences. Tbe difference be-tween tbe means of tbe first and fourtbquartiles is 9 percentage points higber in1984 than 1979, but tbe gap narrows toonly 2 points in 1985. These figures areconsistent with a larger decline in earningsat tbe low end of tbe Cuban wagedistribution after the Mariel immigration,as might be expected from the addition ofa large group of relatively unskilledworkers to the pool of Cubans. Tbe extentof tbe decline, however, is not preciselymeasured.

An alternative metbod to assess theeffect of tbe Mariel immigration on tbeearnings of Cubans in tbe Miami labormarket is to compare Cuban wages inMiami to the wages of Cubans elsewherein the United States. Since tbe fraction ofMariels in tbe Cuban labor force isroughly the same inside and outsideMiami, tbis comparison controls for anyunobservable differences in skill between


the Mariels and other Cubans (due tolanguage ability, for example)." Theninth and tenth columns of Table 7contain estimates of the wage differentialfor Cubans in Miami relative to thoseelsewhere in the country, both unadjustedand adjusted for education, gender,part-time status, private sector/publicsector employment, marital status (inter-acted with gender), and potential experi-ence.

The earnings differentials computed inthis way are roughly constant between1979 and 1984. The 1982 unadjustedwage differential is 10 percentage pointslarger than earlier or later ones, but theregression-adjusted differential is not sig-nificantly different from any of the otherdifferentials. The 1985 data also indicate aslightly higher Cuban wage rate outsideMiami. In any case, a comparison ofCuban wages inside and outside the Miamilabor market shows no evidence of awidening gap in the years immediatelyfollowing the Mariel immigration. On theassumption that the Mariel influx had noeffect on the wage rates of other Cubansoutside Miami, this finding suggests thatthe observed downturn in Cuban wages inMiami can be attributed solely to the"dilution" of the Cuban labor force withless-skilled Mariel workers.

Interpretation of the Findings

The data in Table 3-7 point to twoconclusions. First, the Mariel immigrationhad essentially no effect on the wages oremployment outcomes of non-Cuban work-ers in the Miami labor market. Second,and perhaps even more surprising, theMariel immigration had no strong effecton the wages of other Cubans. Theobserved decline in average Cuban wagerates in Miami after 1980 is no larger thanwould be expected by simply adding theMariel immigrants to the pool of Cubanworkers, assuming that the Mariels earned

" This proposition is strictly true only if theiiiiobseivable differences have a constant propor-tional effect on all Mariels, independent of the levelof observed skills or location choice.

about one-third less than other compara-ble Cubans (as the March 1985 datasuggest). This conclusion is confirmed bya comparison of Cuban wage rates insideand outside Miami, which shows norelative change over the period.

These conclusions lead naturally to thequestion of how the Miami labor marketwas able to absorb a 7% increase in thelabor force with no adverse effects. Onepossible answer is that the Mariels dis-placed other immigrants and natives whowould have moved to Miami in the early1980s had the Boatlift not occurred. Someevidence on this hypothesis is provided bycomparing population growth rates inMiami to those in other Florida cities.From 1970 to 1980, the Miami populationgrew at an annual rate of 2.5% per yearw hile the population of the rest of Floridagrew at a rate of 3.9%. After April 1,1980, the growth rate in Miami slowed to1.4% per year while that in the rest of thestate decreased to 3.4%.'2 The greaterslowdown in Miami suggests that theBoatlift may have actually held backlong-run population growth in Miami. Infact, the population of Dade county in1986 was about equal to the pre-Boatliftprojection of the University of FloridaBureau of Economic and Business Re-search under their "low population growth"scenario (see Florida Statistical Abstract1981, Table 1.24).

Nevertheless, data from the March 1985Current Population Survey suggest thatMiami continued to attract new foreign-born immigrants after 1980. A total of2.7% of all non-Cuban immigrants whoarrived in the United States after March1980 were living in Miami in March 1985.At the time of the 1980 Census, however,only 1.8% of all non-Cuban immigrants inthe United States lived in Miami. Thus,Miami attracted "more than its share" ofnew non-Cuban immigrants to the countryin the five-year period after the Marielimmigration. The implication is that theslowdown in the rate of growth of the

'"These figures are obtained from U.S. Depart-ment of Commerce (1971, Table 32, and 1988, Table


Miami SMSA after June 1980 occurredbecause of a change in the net migrationrate of natives and older cohorts ofimmigrants, rather than because of achange in the inflow rate of new immi-grants. This finding is consistent with thepattern of domestic migration between1970 and 1980 identified by Filer (1988),who found a strong negative correlationacross SMSAs between the net in-migration rate of natives and the in-migration rate of immigrants.

A second explanation for the rapidabsorption of the Mariel immigrants is thegrowth of industries that utilize relativelyunskilled labor. Altonji and Card's (1989)tabulations from the 1970 and 1980Censuses indicate that a small set ofindustries employ a large fraction ofimmigrants, and that these industriesexpanded more rapidly between 1970 and1980 in cities with large immigrant popu-lations than in other cities. The immigrant-intensive industries identified in theiranalysis are apparel and textiles, agricul-ture, furniture, private household ser-vices, hotels and motels, eating and drink-ing establishments, and business services.These are relatively low-wage industriesthat employ large numbers of semi-skilledoperatives and laborers.

A comparison of the industry distribu-tions of employment in Miami and theentire country before and after the MarielBoatlift shows little change in the relativeimportance of immigrant-intensive indus-tries in Miami. 13 Nevertheless, these tabu-lations suggest that the industry distribu-tion in Miami in the late 1970s was wellsuited to handle an infiux of unskilledimmigrants. Textile and apparel indus-tries were particularly prominent in Mi-ami, with 5.5% of total employment inthese industries as compared to only 2.3%nationwide. Seventy-five percent of work-ers in textiles and apparel and 45% ofworkers in other manufacturing industries

'̂ These tabulations are reported in Card (1989,Table 9). Pre-Boatlift data are based on the 1979 and1980 March Current Population Surveys. Post-Boatlift data are based on the March 1984 andMarch 1985 CPS.

were Cubans. Although employment inimmigrant-intensive industries did notexpand after the Boatlift, and the Cubanshare of employment in these industrieswas relatively stable, the Mariels may havesimply replaced earlier cohorts of Cubanimmigrants as the latter moved to moredesirable jobs.

Conclusions

The experiences of the Miami labormarket in the aftermath of the MarielBoatlift provide a natural experiment withwhich to evaluate the effect of unskilledimmigration on the labor market opportu-nities of native workers. The Marielimmigrants increased the labor force ofthe Miami metropolitan area by 7%.Because most of these immigrants wererelatively unskilled, the proportional in-crease in labor supply to less-skilledoccupations and industries was muchgreater.

Yet, this study shows that the infiux ofMariel immigrants had virtually no effecton the wage rates of less-skilled non-Cuban workers. Similarly, there is noevidence of an increase in unemploymentamong less-skilled blacks or other non-Cuban workers. Rather, the data analysissuggests a remarkably rapid absorption ofthe Mariel immigrants into the Miamilabor force, with negligible effects onother groups. Even among the Cubanpopulation there is no indication thatwages or unemployment rates of earlierimmigrants were substantially affected bythe arrival of the Mariels.

Despite the clear-cut nature of thesefindings, some caution is required in theirinterpretation, since the Miami labor mar-ket is far from typical of other local labormarkets in the United States. Althoughthe arrival of some 60,000 refugees in onlya six-month period occasioned problemsfor the Mariel immigrants, in many re-spects Miami was better prepared toreceive them than any other city. In thetwo decades before the Mariel BoadiftMiami had absorbed a continuing fiow ofCubans, and in the years since the Boatliftit has continued to receive large numbers


of Nicaraguans and other Central Ameri-cans. Thus, the Mariel immigration can beseen as part of a long-run pattern thatdistinguishes Miami from most otherAmerican cities.

Two factors that may have been espe-cially important in facilitating the absorp-tion of the Mariel immigrants are relatedto the distinctive character of the Miamilabor market. First, Miami's industry struc-ture was well suited to make use of aninflux of unskilled labor. This structure,and particularly the high concentration oftextile and apparel industries, evolvedover the previous two decades in responseto earlier waves of immigrants, and mayhave allowed the Mariel immigrants totake up unskilled jobs as earlier Cuban

immigrants moved to better ones. Second,because of the high concentration ofHispanics in Miami, the lack of English-speaking ability among the Mariels mayhave had smaller effects than could beexpected for other immigrants in othercities.

A final factor in the Mariel immigrationis the response of domestic migration. Acomparison of Miami growth rates tothose in the rest of Florida suggests thatthe net migration rate of natives andearlier immigrants into the Miami areaslowed considerably after the Boatlift. Tosome extent the Mariels may have dis-placed other migrants from within theUnited States who could have been ex-pected to move to Miami.

REFERENCES

Altonji, Joseph, and David Card. 1989. "The Effectsof Immigration on the Labor Market Outcomes ofNatives." Princeton University Industrial RelationsSection Working Paper Number 256, August.

Borjas, George. 1987. "Immigrants, Minorities, andLabor Market Competition." Indtistrial and LaborRelations Review, Vol. 40 (April), pp. 382-92.

Bound, John, and George Johnson. 1989. "Changesin the Structure of Wages During the 198O's: AnEvaluation of Alternative Explanations." NationalBureau of Economic Research Working PaperNumber 2983, May.

Business Week. 1980. "The New Wave of Cubans IsSwamping Miami." No. 2651 (August 25), pp.86-88.

Card, David. 1989. "The Impact of the MarielBoatlift on the Miami Labor Market." NationalBureau of Economic Research Working PaperNumber 3069, August.

Filer, Randall. 1988. "The Impact of ImmigrantArrivals on Migratory Patterns of U.S. Workers."Unpublished manuscript. Hunter College and theGraduate Center, City University of New York,October.

Governor of Florida. 1980. Report of the Governor'sDade County Citizen's Committee. Miami.

Greenwood, Michael, and John McDowell. 1984."The Factor Market Consequences of U.S. Immi-gration." Journal of Economic Literature, Vol. 34(December), pp. 1738-72.

Grossman, Jean. 1982. "The Substitutability ofNatives and Immigrants in Production." Review of

Economics and Statistics, Vol. 64 (November), pp.596-603.

Lalonde, Robert, and Robert Topel. 1987. "LaborMarket Adjustments to Increased Immigration."Unpublished manuscript. University of ChicagoGraduate School of Business, September.

Masud-Piloto, Felix Roberto. 1988. With Open Arms:Cuban Migration to the United States. Totowa, N.J.:Rowman & Littlefield.

United States Department of Commerce, Bureau ofthe Census. 1971. 1970 Census of Population-Number of Inhabitants: United States Summary (PCl-Al). Washington, D.C: GPO.

1983. 7950 Census of Population —Characteris-tics of the Population: Detailed Population Characteris-tics. Vol. 1, Chap. D, Part 2, Florida: PC80-1-D11.Washington, D.C: GPO.

-. 1988. Current Population Reports—PopulationEstimates and Projections: United States PopulationEstimates by Age, Sex, and Race: 1980 to 1987. SeriesP-25, No. 1022. Washington, D.C: GPO, March.

_. 1988. Current Population Reports—Local Popu-lation Estimates: South 1986 Population and 1985 PerCapita Income Estimates for Counties and IncorporatedPlaces. Series P-26, No. 86-S-SC. Washington, D.C:GPO, March.

University of Florida Bureau of Economics andBusiness Research. 1981. Florida Statistical Abstract.Gainesville: Uhiversity Presses of Florida.

Wilbanks, William. 1984. Murder in Miami: AnAnalysis of Homicide Patterns and Trends in DadeCounty, Florida, 1917-1983. Lanham, Md.: Univer-sity Press of America.

the impact of the mariel boatlift on the miami labor …

Documents