This PDF is a selection from an out-of-print volume from the National Bureau
of Economic Research
Volume Title: Immigration and the Workforce: Economic Consequences
for the United States and Source Areas
Volume Author/Editor: George J. Borjas and Richard B. Freeman, editors
Volume Publisher: University of Chicago Press
Volume ISBN: 0-226-06633-9
Volume URL: http://www.nber.org/books/borj92-1
Conference Date: January 14-17, 1990
Publication Date: January 1992
Chapter Title: When the Minimum Wage Really Bites: The Effect of the U.S.-Leve
Minimum on Puerto Rico
Chapter Author: Alida Castillo-Freeman, Richard B. Freeman
Chapter URL: http://www.nber.org/chapters/c6909
Chapter pages in book: (p. 177 - 212)
6 When the Minimum Wage
Really Bites: The Effect of
the U.S.-Level Minimum
on Puerto Rico
Alida J. Castillo-Freeman and Richard B. Freeman
Since the passage of the Fair Labor Standards Act in 1938, Puerto Rico has
been subject to minimum wage regulations. For many years, industry boards
set separate minima by industry and occupation that were markedly below the
U.S. federal minimum wage. In 1974, the U.S. Congress, supported by the
Puerto Rican government, initiated a policy to raise the level and coverage of
federally mandated minimum wages on the island to U.S. standards. By 1983,
the minimum wage in Puerto Rico reached the $3.35 per hour rate then prevailing
in the United States, and coverage matched the U.S. rate of 60 percent
or more of the work force. With hourly earnings on the island just two-thirds
of mainland hourly earnings, the result was an extraordinarily high ratio of
the minimum wage to average pay-producing a minimum wage with genuine
economic bite.
To what extent has the U.S.-level minimum reduced employment in Puerto
Rico? How important has migration been in adjusting to the minimum wageinduced
loss of jobs? Has the minimum contributed to the migration of lesseducated
and less-skilled Puerto Ricans to the United States (Ramos, in this
volume)?
This paper seeks to answer these questions using diverse bodies of data on
employment and earnings in Puerto Rico and on the employment and earnings
of Puerto Rican migrants in the United States. It reports the following findings.
(1) The U.S.-level minimum altered the distribution of earnings in
Alida J. Castillo-Freeman is a research assistant at the National Bureau of Economic Research.
Richard B. Freeman is professor of economics at Harvard University,director of the Labor Studies
Program of the National Bureau of Economic Research, and senior research fellow at the Centre
for Economic Performance, the London School of Economics.
The authors benefited from discussion with Fernando Ramos, who guided them through much
of the Census of Population data on Puerto Rico, and to Roland0 Vele, Antonio Ortiz Miranda,
and Antonio Sotto for helping them with CPS data for the island. They also thank Dan Sweeney
of the U.S. Department of Labor for furnishing industry-level minimum wage data.
177
178 Alida J. Castillo-Freeman and Richard B. Freeman
Puerto Rico to an extraordinary extent, creating marked spikes that dominate
the earnings distribution. (2) Imposing the U.S.-level minimum reduced total
island employment by 8-10 percent compared to the level that would have
prevailed had the minimum been the same proportion of average wages as in
the United States. In addition, it reallocated labor across industries, greatly
reducing jobs in low-wage sectors that had to raise minima substantially to
reach federal levels. (3) Migrants from Puerto Rico to the United States are
drawn largely from persons jobless on the island, with characteristics that
make them liable to have been disemployed by the minimum wage. As the
Puerto Rican minimum rose toward U.S. levels, the education of migrants fell
below that of nonmigrants. (4) Migration was critical in allowing Puerto Rico
to institute U.S.-level minimum wages and played a major role in the longterm
growth of real earnings in Puerto Rico by reducing the labor supply and
raising the average qualifications of workers on the island.
We present the evidence for these claims in four stages. First, we describe
the minimum wage system in Pugrto Rico and show that it altered the island’s
distribution of earnings. Second, we estimate the employment consequences
of the minimum using time-series and cross-industry data. Third, we examine
the volume and characteristics of migrants to the United States as the island
moved toward the federal minimum. Finally, we consider the consequences of
migration for the minimum wage system and outcomes in the Puerto Rican
labor market.
6.1 The Minimum Wage in Puerto Rico
The 1938 Fair Labor Standards Act (FLSA) established minimum wages in
Puerto Rico as it did in the United States proper. At first, the law applied the
mainland minimum rate ($0.35) to Puerto Rico, but Congress soon recognized
that this would devastate the island’s economy and passed a separate
amendment that established committees in some forty industries to set separate
industry and occupational minima “that would not substantially curtail
employment” while also not giving Puerto Rico “an unfair competitive advantage
over mainland competitors” (U.S. Department of Commerce 1979,
2:633-42). From 1940 until 1974, amendments to the FLSA expanded coverage
on the island but maintained the industry-committee mode of setting
minima. With the 1974 and 1977 Amendments to the FLSA, however, Congress
introduced a new policy, increasing coverage and enacting automatic
increases in Puerto Rican minima to bring them to the U.S. level. The 1977
Amendment required industries with minima at U.S. levels to follow the
scheduled mainland increases and those whose minima were below U.S. levels
to increase wages by $0.30 per year until they reached the federal minimum.
By 1983, Puerto Rico had effectively reached the mainland minimum.
Table 6.1 records levels of the minimum wage and coverage on the island
and, for purposes of comparison, in the United States, in each year that Con-
179 Effect of the U.S.-Level Minimum Wage on Puerto Rico
Table 6.1 Minimum Wage, MinimumiHnurly Earnings in Manufacturing, and
Coverage
Puerto Rico United States
(Min./Mfg.) (Min./Mfg.)
Min. ($) Min./Mfg. Cov. X Cov. Min. ($) Min./Mfg. Cov. X Cov.
Year (1) (2) (3) (4) ( 5 ) (6) (7) (8)
1950
1956
1961
1963
1967
1968
1914
1915
1916
1978
1919
1980
1981
1987
.20 .41 .29 .I4 .15
.45 .10 .29 .20 1.oo
.61 .62 .29 .I8 1.15
.12 .64 .29 .I9 1.25
.91 .70 .44 .31 I .40
1.10 .11 .44 .31 1.60
1.68 .12 .41 .34 2.00
1.87 .13 .66 .48 2.10
2.03 .13 .64 .47 2.30
2.5 1 .75 .64 .48 2.65
2.71 .15 .64 .48 2.90
3.00 .15 .64 .48 3.10
3.26 .I4 .64 .41 3.35
3.35 .63 .64 .40 3.35
.52 .36 .19
.51 .38 .19
S O .43 .22
.5 1 .44 .22
.50 .55 .28
.53 .54 .29
.45 .62 .28
.44 .60 .26
.44 .60 .26
.43 .62 .21
.43 .63 .21
.43 .63 .21
.42 .63 .26
.34 .64 .22
Sources: h e r t o Rico: minimum calculated from U.S. Department of Labor’s “Minimum Wage
Industry Studies”; average hourly earnings in manufacturing from the Yearbook of Labour Statistics;
coverage based on unpublished estimates from the U.S. Department of Labor, Employment
Standards Administration. United States: minimum wages from the 1990Staristical Abstract of
the United States. table 675; manufacturing earnings from the 1990 Economic Report of rhe
President; coverage is estimated from Welch (1978) by multiplying by ratio of nonagricultural
private employees to total employment.
Note: Cov = number of covered nonsupervisory employees divided by civilian employment
gress changed the law and in 1987. As there was no single minimum in Puerto
Rico until the 1980s, the pre-1983 “average minimum” for the island in column
1 is the employment-weighted average of forty-four separate industry
minima (based in some cases on averages of occupational minima within industry,
as described in Castillo [19831).Column 2 gives the ratio of the average
minimum to average hourly earnings in manufacturing; it shows that
industry boards set rates on the order of 60-70 percent of average manufacturing
earnings through 1974. Column 3 presents estimates of the ratio of the
number of workers covered by the minimum to civilian employment. Because
agriculture, government, and much of the trade and service sector were not
covered by the law, however, only 29 percent of the island work force was
subject to the minimum from the 1950s though 1967, compared to much
higher proportions of the U.S. work force. As a result, until 1967the ratio of
the coverage-weighted minimum to average earnings-a crude measure of
the overall strength of the minimum wage-was lower in Puerto Rico than in
the United States (col. 4 vs. col. 8). Hence, the effect of the minimum on
the aggregate Puerto Rican labor market was modest. Not until 1975 did the
180 Alida J. Castillo-Freemanand Richard B. Freeman
coverage-weighted minimum on the island markedly exceed the U.S. level,
after which it remained 80 percent or so higher through 1987.
6.1.1
An effective minimum wage should produce spikes in the distribution of
earnings in the area of the minimum, allowing the analyst to infer the level of
the minimum from the shape of the distribution. Our investigation of three
data sets of earnings shows such a pattern for the island, with remarkable
spikes at the relevant minima in each distribution, implying that the minimum
wage law is a major determinant of actual wages paid.
The first data set consists of wage distributions for workers in covered industries
obtained by the Bureau of Labor Statistics U.S. Department of Labor
(“Minimum Wage Industry Studies”) as part of its assessment of separate industry
minima through the early 1970s.We examined distributions for dozens
of industries and found that, in all low-wage industries, an extraordinary proportion
of workers was paid the industry minimum, with modal pay changing
with changes in the minimum. For example, in 1964, the hourly minimum
was $0.83 in shoes and related products, and 41 percent of workers received
exactly $0.83; in 1968, when the industry minimum had increased to $1.17,
84 percent were paid $1.17. Similarly, in 1964, when the hourly minimum in
the women’s and children’s underwear industry was $0.96, 49 percent of
workers received $0.96, whereas in 1972, when the industry minimum was
$1.45,41 percent received that wage. In rubber products, 46 percent were
paid the $0.98 minimum in 1963and 36 percent the $1.32 minimum in 1969.
Comparable distributions for industries in the United States show no such
bunching of wages around the minimum except for such sectors as sawmills
in the south in the early years of the minimum wage law.
Second, an examination of earnings from the 1980 Census of Population
for Puerto Rico reveals spikes at pay levels where differing industry minima
covered workers during the period of transition to the U.S. minimum. The
Census asked individuals to report annual earnings, weeks worked, and usual
hours worked per week in the preceding year; we used these data to calculate
hourly pay by dividing annual earnings by weeks worked times hours worked
per week.’ In contrast to the Department of Labor’s surveys, the Census encompasses
the entire island, is obtained from individuals who have no incentive
to report wages at the minimum, and covers the period when federal regulation
rather than industry boards set the minima. Figure 6.la displays the
distribution of 1979 hourly pay for full-time workers in Puerto Rico from the
Census files. Since not all industries had reached the federal minimum in
Effect of the Minimum on Wages
1. One disadvantage of the Census data is that they have not been cleaned for errors due to
miscoding etc., as is typically done with CPS data. We deleted observations when weeks worked
was less than twenty, when hours worked last week was less than ten or equal to ninety-nine, and
when wage was less than $0.50 per hour.
181 Effect of the U.S.-Level Minimum Wage on Puerto Rico
1979, the law cannot be expected to create a single spike in the earnings distribution.
In 1979, approximately 50 percent of covered workers were at the
U.S. minimum of $2.90,. 13percent were covered by a minimum within $0.10
of that value, and 25 percent were covered by the industry minimum in the
$2.50-$2.60 range. An effective minimum would thus produce one spike at
$2.90 and a smaller one around $2.50-$2.60. The figure shows such a pat-
tern.
A third source of data on the Puerto Rican earnings distribution is usual
hourly earnings (= usual weekly earnings/usual hours worked) from the Current
Population Survey (CPS) in Puerto Rico. The CPS provides earnings information
in the 1980s, when the Puerto Rican minimum reached the U.S.
level throughout the island. The CPS usual weekly earnings and hours information
is closer to wage rates than the census figures for annual earnings and
hours/weeks worked and thus should be subject to smaller measurement error
than the Census data. We estimated the distribution of hourly earnings for
Puerto Rico using CPS files for 1983 and 1988. The distributions for both
years show the dominance of the U.S. minimum on the island’s pattern of
earnings. Figure 6.l b displays the distribution of hourly earnings from the
CPS for 1983. It reveals a spike around the $3.35 U.S. minimum: 25 percent
of the workers were paid between $3.30 and $3.40 in that year. The change in
the shape of the earnings distribution from one centered on $2.50-$2.60 and
$2.90 in 1979 to one centered on $3.35 in 1983 indicates that imposition of
the U.S. minimum on Puerto Rico altered the distribution of pay on the island.
Figure 6.lc gives the comparable distribution for 1988,at which time the U.S.
minimum had been in effect for roughly five years. The spike in the $3.30$3.40
range is as pronounced as in 1983, with 28 percent paid in this range,
the vast majority at exactly $3.35. That the $3.35 minimum continued to be
the modal rate of pay despite five years of rising nominal average wages is
striking evidence that the minimum did indeed constrain island wage setting.
As a final test of how the minimum affected earnings in Puerto Rico, we
regressed the log of average hourly earnings economy-wide and in manufacturing
on the log of the minimum wage from 1951 to 1987, controlling for a
general time trend, real GNP in Puerto Rico, and the GNP deflator in Puerto
Rico. We used three related statistical specifications: ordinary least squares, a
first-order autoregressive structure, and least squares with earnings lagged one
year as an additional regressor. The minimum obtained a positive significant
coefficient in each regression (table 6.2) with elasticities in the range of 0.2-
0.4. The below-unit elasticities are consistent with the evidence in figure 6.1
that the minimum affects the shape of the earnings distribution by increasing
wages at the lower end rather than by producing general wage inflation.
In sum, imposition of the U.S.-level minimum wage in Puerto Rico altered
hourly earnings on the island in ways that make Puerto Rico an excellent
institutional setting for assessing the effects of the minimum on the job market.
Does the minimum have the “economic bite” on employment in Puerto
182 Alida J. Castillo-Freemanand Richard B. Freeman
A. 1979
n
1
1 I
I 2.352190 4 10
Wage
B. 1983
.02
1
I3
Wage
Rico that textbook discussions of wage-fixing laws lead one to expect? Has
the minimum wage affected migrant flows as well?
6.2 Employment Effects of the Minimum Wage
To determine the employment effects of the minimum, we performed two
related analyses. First, we used time-series data to estimate the effect of the
minimum on the employment-population rate and unemployment rate on the
183 Effect of the U.S.-Level Minimum Wage on Puerto Rico
C. 1988
c
0.-
wage
Fig. 6.1 Distributionof hourly earningsin Puerto Rico
Source: a, 1980 Puerto Rican Census. b, 1983 Puerto Rican CPS. c, June 1988 Puerto Rican
CPS.
Nore: The Census data are based on annual earnings. We deleted observations when weeks
worked was less than twenty, when hours worked last week was less than ten or equal to ninetynine,
and when the wage was less than $0.50 per hour.
Table 6.2 . The Effect of the Minimum Wage on Average Earnings in Puerto
Rico, 1951-87
Average Earnings Average Manufacturing Earnings
(1) (2) (3) (4) (5) (6)
Constant -4.34 -3.69 -2.10 - .26 - .42 - .21
Log min. .32 .27 .I9 .31 .21 .20
(.@) (.05) (.04) (.05) (.07) (.04)
Trend .01 .03 - .oo .04 .05 - .oo
(.01) (.01) (.01) (.01) (.02)
Log PR def. .39 .45 .I5 .I9 .06 .09
(.12) (. 11) (.W (.2% (.09)
Log PR GNP .51 .02 .30 .01 - .03 .05
(.07) ~ 0 3 ) ~ 0 7 ) (.20) (. 16) (.06)
Lagged earnings .I2
(.09)
R2 ,999 ,999 ,999 ,998 ,999 ,999
SE ,028 .025 .020 ,031 ,028 ,022
A N1) . . . .61 . . . . . . .81 . . .
(. 14)
Source: Estimated by the authors from data in app. A.
Nore: Standard errors are given in parentheses.
184 Alida J. Castillo-Freeman and Richard B. Freeman
island. Our time-series regressions are based on models analogous to those in
U.S. minimum wage studies (see Brown, Gilroy, and Kohen 1982) in which
the key independent variable is an average of the ratio of the minimum to
average earnings times coverage among industries. Second, exploiting the
fact that industries in h e r t o Rico had different minima for many years and
reached the U.S. level at different times, we estimated a cross-sectional timeseries
model linking industry employment to industry minima.
Table 6.3 presents the results of our time-series analysis for the period
1951-87. As employment and population data for Puerto Rico are given in
calendar year and fiscal year terms and are periodically revised on the basis of
the latest Census of Population, we present estimates using two related employment
and population series. In columns 1-2, our dependent variable is
based on calendar year employment and population data for persons aged 16
and over adjusted to the 1980 Census for 1963-87 and fiscal year data for
persons aged 14 and over for earlier years.* In columns 3-4, our dependent
variable uses a consistent fiscal year series for persons aged 14 and over, also
adjusted to the 1980census benchmark. In columns 5-6, the dependent variable
is the In of the unemployment rate. For the minimum, we use two related
measures. The first is a “Kaitz” employment-weighted average of each sector’s
minimudaverage hourly earnings multiplied by its coverage:
where a, is industry i’s share of island employment; miis the minimum in
industry i, w iis the average hourly earnings in industry i, and c,is the coverage
in industry i. The second measure is the ratio of the average minimum (m)to
averagelhourly earnings (w)in the economy multiplied by an economy-wide
coverage figure obtained from a different source than the industry coverage
figures (see app. A).3In all cases, the regressions are in In form and include
Puerto Rican and U.S. GNPs (in In constant dollars) and a linear trend. Because
ordinary least squares regressions showed considerable serial correlation,
the calculations are based on an AR(1) model (OLS estimates yielded
larger coefficients on the minimum, so our results are the more conservative
ones).4
The regressions show that, however specified, the minimum had a significant
effect on the employment-population rate, with estimated short-run elasticities
that range from 0.15 (cols. I and 3) to 0.10 (col. 4). These elasticities
2. There is only a modest effect of changing from the age 14 and over to the age 16 and over
data (see U.S. Department of Commerce 1979,2:600, table 16).
3. Because the minimum wage alters average earnings in herto Rico, however, we also estimated
a model in which the GDP deflator for Puerto Rico replaces average earnings as the deflator
in the minimum variable. Our results are similar but weaker than those in the table.
4. The OLS specification yielded coefficients (standard errors) for the minimum wage variables
in the columns in the table of -.24 (.05), - .I9 (.04), -.I8 (.05), and - .I4 (.04) on the
employment-population regressions and of .29 (.26) and .27 (.20) on the unemployment rate.
185 Effect of the U.S.-Level Minimum Wage on Puerto Rico
Table 6.3 Regression Coefficients (standarderrors) for the Effect of the Minimum
Wage (and other factors) on Employment-Populationand Unemployment
Rates in Puerto Rico, 1951-87
Employment Rate Unemployment Rate
Minimum:
Kaitz
(MiniAvg.)X comp.
Puerto Rican GNP
U.S. GNP
Trend
R2
SE
-.15
(.07)
.25
(.11)
.38
(.22)
- ,024
(.006)
-5.51
.65
.92
.027
-.15
~ 0 7 )
-.I1
(.05)
.27 .04
(.Ill (. 10)
.37 .62
(.22) (.25)
(.006) (.006)
-.024 -.024
-5.54 -5.70
.64 .43
(. 16) (. 19)
.92 .91
,027 ,029
.27
(.20)
- .84
- 1.21
(.57)
~ 7 1 )
,081
(.022)
12.56
.85
(. 12)
.89
,083
.21
(. 14)
- .87
(57)
(.70)
,080
(.023)
- 1.17
12.52
.86
(. 11)
.90
,084
Source: Estimated using data in app. A. Employment rate: measure A uses the PREFQP series, measure
B the PREPOPF series.
are of comparable magnitude to those found on the effects of the minimum on
teenage employment in the United States prior to the 1980s (Brown 1988;
Brown, Gilroy, and Kohen 1982)but with lower standard errors, presumably
because in Puerto Rico a much greater proportion of the work force is affected
by the minimum, giving us the equivalent of a larger data set from which to
draw statistical inferences. In addition to the calculations in table 6.3, we did
several other analyses of the time-series data to check on the robustness of our
findings. In one set of calculations, we added additional lagged dependent
variables: the coefficients on the minimum wage variable were similar in magnitude
and significance to those in the table, and the coefficients on the lagged
dependent variable were on the order of 1/2. If we interpret the lagged coefficients
in a partial adjustment context, the responses of employment are
roughly twice those in table 6.3.5In another set of calculations, we estimated
ARIMA models of different orders. The results were comparable to those in
5. Because of the lagged term, we estimated regressions for 1952-87. Our coefficients (standard
errors) on the minimum wage variable and the lagged dependent variable for the models in
table 6.3are - .17 (.06)and .45 (.19); - . I 3 (.04)and .48 (.17); - .12 (.OS) and .49 (.17); - .09
(.04)and .50 (.l6); .30 (.13)and .65 (.16); and .41 (.19) and .41 (.19).
186 Alida J. Castillo-Freeman and Richard B. Freeman
table 6.3, indicating that the finding of a substantial minimum wage effect is
robust to the precise model used to analyze the data. Rather than giving all
our specifications, we present the time-series data in appendix A, which readers
can manipulate as they desire.6
The short-run elasticities of employment to the minimum in table 6.3,
while modest in magnitude, suggest that Puerto Rico experienced massivejob
losses as a result of the application of the U.S. minimum to the island. This is
because the minimum is so high relative to average earnings in Puerto Rico.
In 1987, for example, the coverage-weighted minimudmanufacturing earnings
in table 6.1 were 0.63 In points higher in Puerto Rico than in the United
States, and a Kaitz index of the minimum for Puerto Rico comparable to
the U.S. Kaitz index was 0.64 In points higher in Puerto Rico than in the
United States.’ A 0.64 In difference in the minimum implies that, even with
relatively modest elasticities, island employment would have been 9 percent
higher in 1987 than if the minimum was at the same level relative to pay as in
the United States. For the period 1973-87, our analysis suggests that the increased
minimum in Puerto Rico reduced the employment-population ratio by
roughly 0.02 points, accounting for over one-third of the 0.052 point actual
drop.*
Turning to the unemployment results in table 6.3, economic theory has no
prediction about the effects of the minimum wage on unemployment rates.
Some workers displaced by the minimum may leave the work force (in the
case of Puerto Rico, they may leave the island), while others may be attracted
to the labor force by the higher pay-and our regressions show correspondingly
weaker effects on unemployment rates than on employment-population
rates. Still, given the magnitude of the minimum in Puerto Rico, the point
estimates imply that the minimum raised the unemployment rate substantially.
A 0.63 In point increase in the Kaitz measure of the minimum, for example,
6. In one analysis, we entered the In of the minimum, coverage, and average earnings separately
and obtained the following coefficients (standard errors): - .12 (.09)on In minimum; - .10
(.07)on In coverage; and - .10(.20)on In averageearnings. That the In minimum and In coverage
have similar coefficients supports the restriction of entering them together in the equation. The
insignificant coefficient on average earnings suggests that it plays little role in the results. In another
analysis, we estimated the effects of the minimum for the period prior to imposition of the
U.S.minimum, 1951-73, and the period when island minima began moving toward the federal
level, from 1974 to 1987. The result was a large significant coefficient in the latter period but an
insignificant coefficientin the former.
7. The U.S. Kaitz index relates to nonagricultural private wage and salary employment,
whereas our measure for Puerto Rico was based on total employment. Accordingly, we estimated
a Kaitz index for Puerto Rico based solely on private nonagricultural wage and salary workers in
Puerto Rico.
8. Specifically, we multiply the 0.64 difference in the Kaitz minimum wage variable by the
0.15 coefficient in col. 1 and get a 0.096 In point reduction in the employment-population rate.
With a 1987 employment-population rate in Puerto Rico of 0.369, this produces a 0.406
employment-population rate, for a 10percent increase in employment. Similarly,the -0.11 coefficient
in col. 2 yields an estimated reduction in the employment rate of 0.07 In points, for a 7
percent increase in employment.
187 Effect of the U.S.-Level Minimum Wage on Puerto Rico
would have raised the rate of unemployment by 3 percentage points, according
to the coefficient in column 5.
Note finally that, if workers displaced by the minimum were more likely
than other workers to migrate to the United States, the table 6.3 employment
and unemployment effects understate the full effect of the minimum in displacing
labor. This is because they are based solely on the population in
Puerto Rico rather than on the larger number of persons who would have been
on the island and unemployed absent the migration option.
6.2.1 Cross-Industry/Time-SeriesAnalyses
The existence of separate minima by industry in Puerto Rico from the
1950s through 1983, and the corresponding differential increase of industry
minima to the U.S. rate, offers an alternative to the standard time-series mode
of estimating the effect of minimum wages on employment. It allows us to
assess the effects of the minimum wage by contrasting changes in employment
among industries as their minima moved toward the federal level. For this
analysis, we created a cross-industry time-series data set for Puerto Rico from
1956 to 1987 by matching employment and earnings data for forty-two industries
that cover the entire economy, save agriculture and g~vernment,~with
minimum wages from the industry reports of the U.S. Department of Labor
(“Minimum Wage Industry Studies”) and with industry coverage from the
U.S. Department of Commerce (1979,table 1, 2:634).With forty-two industries
and thirty-one years (we excluded 1982 as there was no Survey of Manufacturing
conducted in Puerto Rico in that year), we have a sample of 1,302
observations that provides a stronger test of the hypothesis that the minimum
affected employment than thirty-one time-series observations.
We use the pooled cross-industrykime-series data set to estimate the effect
of minimum wages on employment in an analysis of covariance framework:
(1) In EMP,, = a + b ln(c,, x m,,/w,,)+ T, + IND, + u,,,
where T,is a vector of year dummy variables to control for cyclical or trend
factors that influence employment, and IND, is an industry dummy variable to
control for the scale of employment in an industry, and uI,is the error term.
The pooled cross-industryhime-series analysis differs from the time-series
analysis in several ways beyond sample size. It allows us to enter separate
year and industry dummies and thus isolate the within-industry within-year
variation in variables that is generally more difficult to explain than pure timeseries
variation. It also permits us to separate our analysis between the period
when minima were set by the industry councils and the period when the Congress
mandated changes in the minimum and thus to test for the potentially
9. We use data for thirty-sevendetailed manufacturing industries from the Puerto Rican Survey
of Manufacturing and for five one-digit nonmanufacturing industries from the Departamento del
Trabajoy Recursos Humanos.
188 Alida J. Castillo-Freeman and Richard B. Freeman
greater employment effects under the latter regime. With output terms excluded
from the equation, moreover, the minimum will capture changes in
employment due to minimum wage-induced changes in industry output as
well as those due to minimum wage-induced changes in employment with
output (level of demand) fixed. And by focusing on employment by industry,
the analysis captures the movement of workers from industries with large increases
in minimum wages to those with small increases as well as from employment
to nonemployment. The latter two considerations suggest that employment
effects from the cross-industry analysis will exceed those from the
aggregate time-series analysis.
Table 6.4 presents the regression coefficients and standard errors on the
minimum wage variables from our industry analysis for the entire period and
for the subperiods 1956-73 and 1974-87. The estimates strongly confirm the
proposition that minimum wage substantially affected employment in Puerto
Rico. In column 1, the minimudaverage times coverage variable obtains a
negative significant coefficient of 0.54, more than four times its standard error.
The estimates in columns 2 and 3 show that the effect of the minimum
occurs entirely in the period after the 1974Amendment imposed increases in
minima toward the U.S. federal level. The elasticity of employment to the
minimum is -0.91 after 1974, compared to an estimated 0.20 elasticity before
1974. This suggests that the industry councils took seriously the mandate
to set minima so that they “would not substantially curtail employment,”
whereas the congressionally mandated changes were, of course, exogenous to
the economic conditions on the island.
Underlying the sizable minimum wage effects in table 6.4 is a substantial
Table 6.4 Regression Coefficients (standard errors) for the Effect of the
Minimum Wage and Other Variables on Ln Employment-Population,
1951-87, and on Ln Employment by Industry, 1956-87
Cross Industry
1956-87 1956-73 1974-87
(1) (2) (3)
Minimum - .54 .20 - .91
Industry dummies 41 41 41
Year dummies 30 17 12
Sample size 1,302 756 546
R2 .87 .95 .95
~ 1 3 ) (.I21 (,241
Source: Estimated by authors from a data set available on disk from the NBER. The detailed
industry data for Puerto Rico are obtained from the U.S. Department of Labor’s“Minimum Wage
Industry Studies” and the Departamento del Trabajo’s “Seria historica del empleo, despempleo y
grupo trabajador in Puerto Rico.”
Note: Minimum is the multiplicand of coverage for industry times minimudhourly earnings in
industry. Figures exclude 1982 owing to the absence of the Puerto Rican Survey of Manufacturing.
The regressions cover forty-two industries, with agriculture and government excluded.
189 Effect of the U.S.-Level Minimum Wage on Puerto Rico
reallocation of Puerto Rican workers from low-wage industries that increased
their minima greatly to reach U.S. standards to high-wage industries that required
only modestly higher minima to reach the federal level. In industries at
the U.S. minimum in 1973, employment grew by 1 percent from 1974 to
1983. In industries whose minimum was within $0.10 of the U.S. minimum
in 1973, employment increased by 2 percent over the same period. But in
industries whose 1973 minimum was more than $0.10 below the U.S. minimum
and that therefore had to raise minima substantially to reach the U.S.
level, employment dropped by 32 percent from 1974 to 1983! These calculations
indicate that the U.S.-level minimum reallocated employment on the
island as well as reducing the total employment-population rate. They show
that a major reason for the higher elasticities of employment to the minimum
in table 6.4 than in table 6.3 is the movement of workers across industries.
6.3 The Minimum Wage and Migration
Since Puerto Ricans face no legal restrictions migrating to the United
States, migration depends on the incentives facing individuals, including
those that result from the minimum wage. Changes in the minimum can affect
both the volume and the composition of migration. Economic analysis has no
clear prediction about how the volume of migration might respond to higher
minimum wages. Workers will be more likely to migrate if they are disemployed
by the minimum than if they hold jobs at below-minimum wages, but
workers whose earnings are increased by the minimum will be less likely
to migrate. The net effect of the minimum on the size of the migrant flow
will depend on the number of workers in the two groups and their response
to the minimum wage-induced change in their conditions. lo Economics
does, however, predict how the minimum will affect the composition of migrants.
Since employers are likely to lay off or forgo hiring the least skilled
workers when wages rise, the minimum should induce greater migration of
the less skilled.
6.3.1 Volume of Migration
Puerto Rican migration to the United States has been immense. In 1980,
one-third of 20-64-year-old persons born in Puerto Rico resided in the United
States (Ramos, in this volume, table 2.1). Although there are no administrative
statistics on migration from Puerto Rico, data from the U.S. and Puerto
Rican Censuses of Population and CPSs and passenger travel records show
sizable changes in the volume of migration over time. In the 1950s, massive
10. If, in accord with our estimates in sec. 6.2, the elasticity of demand to the minimum is
below one, expected earnings of all workers will be higher under the minimum. This does not,
however, imply that the minimum reduces migrant flows. for workers who suffer the loss of jobs
but do not gain the benefits of higher wages may respond more to their condition than those who
obtain modest gains in wages. Moreover, the elasticity is based on total employment, including
workers whose skills make the minimum irrelevant to them.
190 Alida J. Castillo-Freeman and Richard B. Freeman
migration from Puerto Rico more than doubled the share of the Puerto Rico
born living in the United States (see table 6.5). For the next two decades,
migration rates fell, seemingly stabilizing the share of the Puerto Rico born
on the mainland, only to surge once more in the 1980s.Figure 6.2 graphs the
only available measure of annual migrant flows-net passengers journeying
from the island-relative to the population on Puerto Rico, revealing further
fluctuation in flows from year to year. From 1951 to 1987,migration averaged
0.9 percent of the island population, with a standard deviation of 0.9. In the
1950s, 1.9 percent of the island population left annually; in the 1960s, 0.5
percent; in the 1970s, 0.3percent; and in the 1980s, 1.0 percent.
To see if the swing in migration is related to the minimum wage, we regressed
the ratio of net migrants to the Puerto Rican population (MIG) on the
Kaitz measure of the minimum, economic conditions in the United States and
Puerto Rico, and the proportion of the Puerto Rico born living in the United
States in the previous year (PRUS), using various measures of economic conditions
and specifications. In nearly all cases, the minimum variable had a
statistically insignificant effect on the volume of migration. For example, controlling
for GNP in Puerto Rico and the United States, we obtained
MIG= -.14+.07 log USGNP- .05 log PRGNP+ ,005 log M I N - . ~ ~PRUS
(.02) (.oa (.01) (.05)
R2 = .57 (standard errors in parentheses).
Here, the coefficient on the minimum wage implies that the 0.64 In point
lower minimum that would make it comparable to the minimum on the mainland
relative to average earnings would reduce migration by 0.3percent of the
population, but the standard error is too high to place any confidence in this
estimate. One interpretation of the high standard error is that the emigrationTable
6.5 Migration from Puerto Rico, 1950-87
Puerto Rico. Puerto Ricans ’3 Puerto Ricans
Total Population Living in U.S. in U.S. (2)/
( I ,000s) ( I ,000s) ( 1 +2)
(1) (2) (3)
1950 2,203 226 ,093
I960 2,339 617 ,209
1970 2.659 810 ,233
198@ 2,889 93I ,244
1987 3.294 1.155 ,260
Sources: “Puerto Ricans in the U.S.” (1950, 1960, 1970); “Characteristics of the Population:
Puerto Rico” (1950, 1960. 1970, 1980): “General Social and Economic Characteristics” (1980,
table 167);Junta de Planificacion, “Informe economico al gobernador,” various eds.; Centro de
Estudios de Puertoriquenos (1979, table 9, p. 187).
Note: The number of Puerto Ricans living in the United States in 1987 IS estimated by summing
annual net passenger flows from 1981 to 1987 and adding half the 1980 flow. The flows are from
Falcon (1990).
191 Effect of the U.S.-Level Minimum Wage on Puerto Rico
3.54
years
Fig. 6.2 Percentage of Puerto Ricans who migrated to the United States
inducing effect of the minimum through disemploying workers and the
emigration-reducing effect of the minimum through raising wages roughly
balance out. Another interpretation is that the minimum has a nonnegligible
effect on migration that cannot be reliably detected in the fluctuating timeseries
data. While econometric manipulations that smooth the data might
yield better estimates of the effect of the minimum,Il we believe that there is
no convincing story about the volume of migration in the time series and turn
instead to micro survey data on how the minimum may have influenced the
characteristics of migrants.
6.3.2 Premigration Employment Experience of Migrants
Economic analysis predicts that workers displaced by the minimum should
be more likely to leave the island than others and, given the higher capital/
labor ratios in the United States, should have greater success finding jobs in
the United States than in Puerto Rico. Consistent with this, Puerto Rican male
and female migrants have much higher employment-population rates than
Puerto Rican residents with the same years of schooling (cols. 1-2 of table
6.6). However, because these data do not tell us how migrants fared when
they were in Puerto Rico, they leave open the possibility that migrants actually
came largely from the ranks of the employed. To determine the employment
experience of migrants prior to emigration, we turned to the 1982-83
“Encuesta de migracion” of the Junta de Planificacion de Puerto Rico (1984).
The encuesta asked migrants at the international airport their employment status
during the previous three months. Because migrants are younger than typ-
11. Santiago (1990) estimates a complex model with monthly flowsthat suggests that the minimum
raised migration, but he has no other control variables, and his raw correlations give the
opposite result. We interpret this as indicating the difficultyof making inferences from these data.
192 Alida J. Castillo-Freeman and Richard B. Freeman
Table 6.6 PercentageEmployed of Puerto Rico-born Men and Women, by Year
of Schoolingand Migrant Status, 1980 and 1983
Males
1980 1983
Migrants to U.S.
Years of In Puerto Rico 3 Months before
Schooling In Puerto Rico In U.S. (adjusted for age) Migration
5 6 39 64 32
7-9 49 68 40
10-1 I 55 72 52
12 67 81 60
13-15 71 84 71
16+ 87 85 79
All 57 71 55
30
33
36
38
48
50
38
Females
5 6 10 22 7 5
7-9 17 26 14 11
1c-I 1 21 31 13 5
12 38 46 34 21
13-15 55 56 46 32
16f 72 61 67 55
All 28 33 29 20
Sources: 1980,tabulated from the Public Use Files of the U.S. and Puerto Rico Censuses. 1983,
tabulated from the Puerto Rican CPS. 1983 migrants, tabulated from the Junta de Planificacion
de Puerto Rico (October 1984),tables 12, 14, and A7.
Note: The'1983 rates are weighted by the age distribution of migrants from the Junta de Planificacion
de Puerto Rico (October 1984), using table A-5 for weights. The same age weights were
used for both sexes.
ical island residents, we contrasted migrant employment rates with a weighted
average of age-specific employment rates of residents, using the age distribution
of migrants as weights. Further, to avoid problems due to changes in
island employment over time or between surveys and Censuses,12 we contrasted
these rates with 1983employment rates from the Puerto Rican CPS.
The results of these comparisons show that the percentage of migrants employed
is markedly below the percentage of island residents employed in the
same education-sex group (cols. 3 and 4 of table 6.6).In the aggregate, 38
percent of male migrants were employed, compared to 55 percent of male
residents. Given the numbers of residents and migrants, this implies that approximately
4.4 percent of employed males and 7.5 percent of jobless males
12. The U.S. Department of Commerce (1979, 2:601) noted substantial differences between
Census and CPS estimates of unemployment rates. The revisions in annual employment and population
estimates after the 1980 Census benchmark was introduced also imply inconsistencies
between the two sources.
193 Effect of the U.S.-Level Minimum Wage on Puerto Rico
migrated. Among women, the pattern is weaker, presumably because women
are “tied movers” (Mincer 1976): 20 percent of migrants worked, compared
to 29 percent of residents, giving rates of migration of 2.6 percent for the
employed and 3.7 percent for the not employed.I3The encuesra also asked
why individuals were migrating. Consistent with the notion that lack of work
induced considerable migration, the vast majority of migrants exclusive of
military personnel and students said that they were migrating to work (33
percent) or to search for work (47 percent) (Junta de Planificacion 1984, table
A-18). Less-educated and younger persons were especially likely to be moving
to search for work.
6.3.3
Ramos (in this volume) has shown that male Puerto Rican migrants to the
United States had fewer years of schooling than similarly aged nonmigrants
in 1980, which he attributes to the higher payoff to education in Puerto Rico.
Since the minimum will disproportionately reduce the employment of lesseducated
workers, this pattern is also consistent with the minimum wage influencing
the characteristics of migrants. To try to identify a separate minimum
wage effect, we compared the education of migrants to that of residents
in Puerto Rico in earlier Censuses. If migrants had less schooling than nonmigrants
in periods when the minimum was weak, we would be loathe to
ascribe much of the 1980 census pattern to the minimum wage. By contrast,
evidence that the migrant-nonmigrant schooling gap developed when the minimum
increased would lend support to the hypothesis that the minimum influenced
the educational composition of migrants.
Published Census data on the median years of schooling of Puerto Rican
migrants and of residents on the island aged 25 years and older in 1950, 1960,
and 197014showed the opposite pattern of relative attainment to that for 1980.
In 1950, when relatively few Puerto Ricans had migrated to the United States,
migrants had 7.7 years of school completed, compared to 3.7 years for persons
on the island. In 1960, migrants completed 7.9 years of schooling, nonmigrants
4.6 years. In 1970, migrants had 8.4 years and nonmigrants 6.9
years completed-a smaller but still substantial gap that implies that as late
as 1970 the bulk of Puerto Rican migrants were more educated than nonmigrants.
Not until the 1980 Census do migrants have less education than nonYears
of Schooling of Migrants
13. The encuesfaestimated that there were 59,000 male and 37,000 female migrantsaged 16-
64 in 1983.The PuertoRican CPS suggeststhat there were approximately501,000employedmen
and479,000 not employedmen on the island aged 16-64. There were 294,000employedwomen
and 806,000 not employed women. We used these data to estimate the rate of migration of employedand
not employedmen and women from PuertoRico, unadjustedfor the age of migrants.
14. The figures for the Puerto Rico born living in the United States are taken from the special
Census volume on “Puerto Ricans Living in the U.S.” in 1950, 1960, and 1970. The figures for
residents in Puerto Rico come from the PuertoRico volumesof the Census. The Census canceled
the 1980 volume on PuertoRicans in the United States.
194 Alida J. Castillo-Freeman and Richard B. Freeman
migrants, consistent with the contention that movement toward the U.S. in the
1970saltered the selectivity of migration.
To explore the 1970-80 change in migrant-nonmigrant education differences
further, we estimated equations for years of schooling from the Public
Use Files of the 1970 and 198015U.S. and Puerto Rican Censuses and the
June 1988U.S. and Puerto Rican CPSs. Building on Ramos’s analysis of the
1980 Census, we pooled the records of 16-64-year-old persons born in
Puerto Rico and residing in the United States with those of Puerto Ricans
living in Puerto Rico. We regressed years of school completed on dummy
variables for age (to control for the upward trend in schooling), migrant status,
and (in 1970and 1980) a “recent migrant” dummy that takes the value of one
if the person migrated to the United States in the preceding five years.
The estimated coefficients and standard errors on the migrant dummy variables
are given in table 6.7. The coefficient in the 1980 regression for males
corroborates Ramos’s finding that migrant men had fewer years of schooling
than resident of Puerto Rico. The coefficientin the 1988regression is smaller
but still negative. By contrast, the coefficientin the 1970regressions for males
show migrants having fewer years of schooling than island residents. The regressions
for females tell a similar story: migrants had less education than
nonmigrants in 1980 and 1988 but more education than nonmigrants in 1970.
In 1980 and 1988, moreover, migrant-nonmigrant educational differentials
were greater for women than for men, possibly because the minimum affects
the employment of female workers more than that of male workers owing to
the lower wages of women.
The regressions that distinguish between recent and earlier migrants tell a
more complex story. The estimated coefficients on recent migrants are negative
in the 1970Census as well as in the 1980Census and are absolutelylarger
in the 1970 Census as well. This implies that 1965-70 migrants were disproportionately
drawn from less-educated Puerto Ricans, whereas those who
came earlier were drawn from the more educated. If the 1965-70 migration of
the less educated was due to the minimum wage, we would expect a sharp rise
in the minimum in 1965-70. Table 6.1 above shows such a rise in the
coverage-weighted minimum in Puerto Rico due to the 1965 Amendment to
the Fair Labor Standards Act. While this does not prove that the increased
migration of the less educated in 1965-70 is due to the minimum, it is consistent
with the minimum wage contributingto the change.
15. For Puerto Ricans in the United States in 1970, we used the 1/100 sample of the U.S.
Census; for Puerto Ricans in Puerto Rico in 1970,the l/lOO sample of the Puerto Rican Census
was used; for PuertoRicans in the United States in 1980,we used a sample derived by combining
the lil00 sample with the 5/100 sample from selected states with many Puerto Ricans. The sample
represents 90 percent of the total Puerto Rico born population in the United States. For details,
see Ramos (in this volume). The sample of Puerto Ricans residing in Puerto Rico is extracted
from the 5/1OO Census for Puerto Rico. We included all out-of-school persons not in the military
in the age bracket 16-64.
195 Effect of the U.S.-Level Minimum Wage on Puerto Rico
Table 6.7 Estimated Coefficients (standarderrors) for Differences in Years of
Schooling Completed by Puerto Rican Migrants to the United States
and Puerto Ricans on the Island, 1970-88
Males
1988186 1980 1970 1980 1970
Migrants - .35 - .53 .18
(.25) (45) (.10)
Recent migrants - .34 - .66
(.14) (.24)
Earlier migrants - .55 .32
(.W (.10)
Dummies for age 9 9 9 9 9
R2 .06 .07 .08 .07 .08
N 5,442 56,809 18,850 55,809 18,850
Females
Migrants - .76 - .70 .23
(,201 (.W ( . W
Recent migrants - .65 .57
.05 (.22)
Earlier migrants - .70 .37
(.09)
Dummies for age 9 9 9 9 9
R‘ .13 .15 .12 .I5 .I2
N 6,451 63,561 20,921 63,561 20,921
Sources: 1970 and 1980tabulated from the Public Use Files of the U.S. and Puerto Rican Censuses.
1988tabulated from the June 1988 Puerto Rican CPS, with migrants from June 1988and
1986from the U S . CPS.
We examined the changed selectivityof migrants by education between the
1970and the 1980Censuses in one additional way. We used the two Censuses
to calculate the average years of schooling of the Puerto Rico born in Puerto
Rico and in the United States in specified “pseudocohorts”-persons aged y
years in 1970 and y + 10 years in 1980. Figure 6.3 displays the results of
this analysis in terms of the change in years of schooling completed by pseudocohorts
on the island and in the United States. It shows greater increases in
years completed for cohorts in Puerto Rico than for cohorts in the United
States, implying a substantial sorting of the Puerto Rico born in a given cohort,
with the less educated moving to the United States and the more educated
returning to Puerto Rico. This is what one would expect if the movement
toward the U.S.-level minimum in the 1970s made it more difficult for
the less educated to find work on the island.
An alternative explanation of why migrants had less schooling than nonmigrants
in the 1980s but not in the 1970s is that the rewards to education
196 Alida J. Castillo-Freeman and Richard B. Freeman
26-30 3-35 36-40 41-45 46-50
1970 age cohorts
Migrants
0 Living in Puerto Rico
Fig. 6.3 Change in mean years of schooling completed by cohort and
residence, Puerto Rico-born men, 1970-80
increased from 1970 to 1980 in Puerto Rico relative to the United States. We
examined this possibility for male workers by estimating log earnings equations
for men in Puerto Rico and for migrants to the United States in 1970and
1980 for both annual and hourly eamings.I6Our model included years of potential
work experience dummy variables and a linear years of schooling term
(all the regressions are in app. B). For Puerto Rico-born men in the United
States, the regressions for weekly earnings gave a coefficient on schooling of
0.048 in 1970 and 0.036 in 1980, for a 0.012 drop in the effect of schooling
on earnings. By contrast, the regressions for Puerto Ricans on the island gave
a coefficienton schooling of 0.071 in 1970and 0.077 in 1980, for a 0.006 rise
in the effect of schooling on earnings. These results suggest that changes in
the returns to schooling contributed to the changed educational selectivity of
migrants in the decade and thus buttress Ramos’s story. The regressions for
yearly earnings showed a 0.010 fall in the effect of schooling on the earnings
of migrants, compared to a 0.018 rise in the effect of schooling on the island.
Since yearly and weekly earnings differ by weeks worked, the implication is
that the effect of schooling on weeks worked increased by just 0.002 among
migrants but by 0.012 among island residents. This is consistent with the
notion that the minimum wage altered the selectivity of migrants by reducing
the employment of the less educated more than of the educated. Thus, there is
evidence that both joblessness due to the increased minimum wage and edu-
16. The 1970Censuses did not ask for usual hours worked, so our analysis is based on information
on hours worked last week. In addition, the Public Use tapes present time-worked data in
categories rather than in actual values; we used category midoointg in our analysis.
197 Effect of the U.S.-Level Minimum Wage on Puerto Rico
cational pay differentials contributed to the greater immigration of the less
educated to the United States in the 1970s.
6.3.4 Migration, Education, and Language
Another way to examine how the minimum wage might affect migration is
to contrast the likelihood of migration among workers more or less likely to
be affected by the minimum. We have done this using education as an indicator
of being affected by the minimum. For six education groups, we used 1980
Puerto Rican Census data to estimate the proportion of workers likely to be
affected by the minimum had they resided in Puerto Rico; we also used pooled
U.S. and Puerto Rican 1980 Census data to estimate the proportion of the
Puerto Rico born residing in the United States.
There is no easy way to assess the proportion of workers affected by the
minimum (PMIN). The distribution of earnings from the 1980 Puerto Rican
Census shows the proportion earning the minimum or less, but not the number
of persons who lost their jobs or who migrated as a result of the minimum,
and thus understates the proportion affected by the minimum. The understatement
will, moreover, be greater for groups whose employment is most reduced
and/or who migrate in large numbers, biasing the estimates against
finding a minimum wage effect on those outcomes. Still, as the Census earnings
distribution offers the only direct indicator of PMIN, we used it to compute
the proportions of Puerto Ricans in different education groups with earnings
at or below the $2.90 U.S. minimum. Column 1 of table 6.8 shows that these
proportions decline with education steeply after high school.
To estimate the proportion of the Puerto Rico born who were migrants in
our six educatibn groups, we regressed a zero-one variable for U.S. residence
on dummy variables for years of schooling groups and for ten age groups in
our pooled U.S. and Puerto Rican census file. The coefficients on the schooling
dummies, given in columns 2 and 3 of table 6.8, reveal a nonlinear relation
between years of schooling and migration that was hidden in the comparison
of mean education of migrants and nonmigrants: the probability of
migration is smallest for the least educated and for the most educated and
highest for those who graduate grade school but do not go beyond high
school. The limited migration of the least educated-who are the most affected
by the minimum, who have the lowest employment rates, and who have
the smallest rewards from schooling-appears to be due to their lack of spoken
English (and/or lack of information related to the possession of that language
skill). Of men with zero to six years of schooling, only 13 percent of
those in Puerto Rico could speak English, compared to 31 percent of those
17. As an alternative estimate of PMIN, we assumed that the 1969 earnings distribution given
in the 1970 Census would have held in 1979absent the minimum wage, inflated 1969earnings by
the rate of growth of average hourly earnings in manufacturing, and estimated the proportion of
workers by education in the inflated distribution likely to be paid the 1979U.S. minimum or less.
The results were sufficiently similar to those in the text that we forgo presenting them.
198 AIida J. Castillo-Freeman and Richard B. Freeman
Table 6.8 Estimated Proportions of Workers Subject to the Minimum in Puerto
Rico and Migrant Status
Proportion Paid Minimum
or Below or
Proportion of Migrants, 1980
Not Employed Adjusted for
Education Not Adjusted Language
0-6
7-8
9-1 I
12
13-15
16+
0-6
7- 8
9-1 1
12
13-1 5
16f
~~
.67
,523
.54
.40
.25
.09
.09 .00
.I6 .oo
.I7 - .03
.09 - .14
.08 - .23
.00 - .36
Females
.73
.71
.64
.55
.38
.I4
. I 1 .04
.I9 .02
.I9 - .03
. I 1 -.I4
.07 - .25
.01 - .37
Source: Column 1 estimated from regressions in appendix C, using 1980 Puerto Rican Census.
Column 2 estimated from regressions in appendix C, using 1980 Puerto Rican and U.S.Censuses.
Column 3 estimated from regressions in appendix C, using 1980 Puerto Rican and U.S.
Censuses, with inclusion of dummy variable for English-speaking proficiency.
with seven to eight years of schooling, 59 percent of high school graduates,
and 92 p'ercent of college graduates. Even among migrants there was a marked
difference in the ability to speak English: 47 percent of migrants with zero to
six years of schooling reported speaking English well, compared to 58 percent
of migrants with seven to eight years of schooling, 96 percent of high school
graduates, and 99 percent of college graduates.I8 Since persons with the least
English-speaking ability are unlikely to migrate, language skill is an omitted
variable associated with education that depresses the migration of the less educated.
Adding dummy variables for ability to speak English to our migrant
regression turns the inverse U-shaped education-migration relation into a
monotonic relation in column 3 of table 6.8. As migrants improve their English,
the estimated effect of English speaking on migration to the United
States will be biased upward, which will in turn bias the estimated coefficients
on education groups in the migration regression, but this bias is unlikely to
18. These calculations are based on questions on English-speaking ability asked in both the
U.S. and the Puerto Rican Censuses. The U.S. Census had a fourfold categorization: speaking
English very well, well, not well, and not at all. The Puerto Rican Census had a threefold categorization:
speaking English easily, with difficulty, and not at all. We collapsed the speaking very
well and the speaking well categorizations in the U.S. Census so that we had comparable threeway
groupings.
199 Effect of the U.S.-Level Minimum Wage on Puerto Rico
account for the elimination of the inverse U-shaped education-migration relation
on addition of the English-speaking variable to our regression.l9
The correlations between PMIN and the adjusted or unadjusted proportion
migrating are high. For men, the correlations are 0.99 with the adjusted proportion
migrating and 0.77 with the unadjusted proportion. For women, the
correlations are .99 with the proportion of migrants corrected for language
and .84 with the proportion of migrants absent the language adjustment. Still,
since those most affected by the minimum have lower wages, it is possible
that the correlation between PMIN and the proportion of migrants by education
group is due not to the minimum but to differences in earnings by education
group. To assess this possibility, we calculated employment rates and earnings
by schooling group for workers on the island and for migrants to the United
States. Figure 6.4 shows that in differences in weekly earnings between migrants
and nonmigrants fell sharply with education, consistent with the earnings
differential explanation of the greater migration of the less educated. It
also shows that In differences in employment rates between migrants and nonmigrants
fell with education, consistent with the minimum wage/joblessness
interpretation of the greater migration of the less educated. With both factors
working in the same direction, it is difficult to assess their relative importance.
In any case, the result is that persons with the skills likely to make them affected
by the minimum dominated the 1970smigrant
6.4 Effects of Migration on the Puerto Rican Job Market
We argue next that, regardless of the causal effect of the minimum wage on
Puerto Rican migration, migration has been a key “safety valve” in the Puerto
Rican job market without which it would have been virtually impossible to
impose the U.S.-level minimum on the island. Migration reduced joblessness,
raised the average skills of workers and the marginal productivity of labor,
and contributed to the growth of real earnings on the island.
19. In the underlying data, the zero to six years of schooling group has a lower proportion in
the United States than the seven to eight years of schooling group, but it has a higher proportion
in the United States for those who speak English well, who speak English somewhat, and for
those who do not speak English. To see if this pattern might be due to those in the United States
learning English, we estimated migration status equations comparable to those in app. C for recent
migrants, who have less time to improve their English proficiency. In our regressions (which
eliminated pre-1975 migrants from the sample), we obtained results similar to those for all migrants:
additional of English-speaking dummies explained the low migration rate among the zero
to six years of schooling group.
20. If the minimum wage induced the less qualified to migrate, we would expect migrants to
do especially poorly in the U.S. labor market conditional on education. Ramos has shown that, in
1980, migrant males, particularly the most recent migrants, had lower earnings than otherwise
comparable to U.S.-born F’uerto Ricans. We estimated log earnings equations for apooled sample
of U.S.-born Puerto Ricans and migrants in 1970 and found a smaller effect for all migrants in
1970 than in 1980, but we found equally large gaps between the earnings of recent migrants and
those of U.S.-born F’uerto Ricans in 1970as in 1980.
200 Alida J. Castillo-Freeman and Richard B. Freeman
- I
0-6 7-13
~ 9-11
n
~ 12 13-15 16+
grade
log dif mig/pr epop
log dif mig/pr earnings
Men
- I
0-6 7-8
19-11
gr
n
~ 12 13-15 16+
de
women
rn log dif mig/pr epop
0log dif mig/pr earnings
Fig. 6.4 Log difference of employment to population and earningsby
education group, migrants to Puerto Ricans
6.4.1
What might the employment-population rate and unemployment rate have
been in Puerto Rico with a U.S.-level minimum wage but no outlet for migra-
tion?
Consider first the potential labor market effects of imposing the U.S. minimum
in the 1970s and 1980sif no Puerto Rican migrated to the United States
after 1974. In this case, there would have been approximately 232,000 additional
persons of working age on the island,21increasing the working-age popMigration
and Joblessness in the Presence of the Minimum
21. This estimate is the sum of approximately 90,000 net migrants from 1975 through early
1980 (based on the 1980Census of Population) and 200.000 net migrants from passenger traffic
201 Effect of the U.S.-Level Minimum Wage on Puerto Rico
ulation by about 10percent. If the number of jobs for the less skilled was fixed
because the minimum disallowed employment-creating movements down the
demand curve, all migrants whose characteristics would have earned them less
than the minimum as well as those with characteristics that would have made
them unemployed would have been jobless on the island. To find out how
many migrants would fit in this set, we estimated the effect of age and schooling
on the probability that someone in Puerto Rico would have been jobless
or paid the minimum or less and applied the resultant equation to the characteristics
of recent migrants. The calculation indicates that 77 percent of migrants
would have been jobless on the island and just 23 percent employed.22
Hence, 199,000 of the 232,000 “return migrants” would have lacked jobs,
and just 53,000would have found work. Assuming that the “return migrants”
participated in the labor force at the same rate as they did in the United States
(56 percent), the number of additional unemployed workers would be 69,000.
Because the island employment-population rate was already low (.359), however,
the addition of these workers would have reduced the island
employment-population rate by only .013, or 3 percent. But, because there
would have been more unemployed than employed return migrants, the unemployment
rate would have risen by 4 percentage points, from 17to 21 percent.
Similar calculations assuming that return migrants would have found
employment at the 31 percent rate of employment of migrants in the three
months preceding migration given in the “Encuesta de migracion” suggest a
substantial 3 percentage point rise in the unemployment rate, although only a
0.5 percentage point fall in the employment-population rate on the island.
6.4.2 Long-Run Migration
How might the Puerto Rican labor market have fared in the entire postWorld
War I1 era if Puerto Ricans had never had the option of migrating to the
United States? For simplicity, we answer this question assuming that all
Puerto Rican migrants to the United States had remained residents of the island
and that other factors that determine economic well-being were unchanged
on the island. As other factors would also have changed, our exercise
should be viewed not as a prediction of what might actually have happened
but rather as a way to demonstrate the magnitude of the economic effects of
immigration on the island. A full analysis of what might have happened absent
migration requires a complete model of the Puerto Rican economy that
lies beyond the scope of this study.
Absent migration from Puerto Rico to the United States, the one-third of
data. This gives 290,000 migrants. On the basis of 1980 Census data, approximately 80 percent
will be between age 16 and age 64, giving the 232,000in the text.
22. We base this estimate on a two-stage analysis. First, we estimated separately for sex the
relation between being employed at above-minimum wages in Puerto Rico in the 1980 Census
and a set of education and age dummy variables. Then we predicted the proportion of recent
migrants who would have been employed on the island, given that equation. The equations are
given in app. C.
202 Alida J. Castillo-Freeman and Richard B. Freeman
the Puerto Rico born of working age living in the United States would be on
the island, raising the working-age population by about 50 percent, or 0.40 In
points. Assuming, as in our earlier calculation for the effects of migration
after 1974, that 70-80 percent of these return migrants were jobless in the
presence of a U.S .-level minimum wage even if island productivity remained
unaffected (i.e., if capital also increased by one-third or so) and that 56 percent
would have been in the work force, the rate of unemployment would have
risen to 30-35 percent. This effect is so large as to call into question the assumption
that the U.S.-level minimum could have been applied to the island
absent migration. More likely than not, the result would have been similar to
that in 1935, when Congress quickly rescinded the U.S.-level minimum and
chose instead the industry-council mode of setting minima. Put differently,
massive migration to the United States was a prerequisite for applying the
U.S. minimum to the island in the 1970s.
What might have been the level of wages on the island absent migration and
absent the minimum wage? If nothing else changed, labor supply would have
been 0.40 In points higher in 1980 than it was. Given an elasticity of substitution
of u, a fixed capital stock, labor’s share of output of a, and market
clearing, a 0.40 In point increase in the work force would reduce pay by
[u/(1 - a)].40 In points.23At any plausible levels of a and u, the change in labor
supply would have a devastating effect on real wages. For instance, with a
Cobb-Douglas production function and labor’s share of two thirds of GNP,24a
0.40 In point increase in labor supply would reduce average earnings by 1.2
In points, or to 30 percent of their current level. Even with a relatively small
elasticity of substitution of, say, 0.15, comparable to our estimated elasticity
of employment to the minimum wage, wages on the island would have been
cut by 0.18 In points, to 84 percent of their current level. Of course, the economy
would have made other adjustments: the return to capital would have
risen, inducing greater investment that would have partially restored the capital/labor
ratio; fertility might have fallen; investments in human capital might
have risen; and so on. But the first-order effect of massive return migration
would clearly fall on real wages.
Given the lower education of migrants, moreover, there would be an additional
reduction in earnings and productivity due to the reduced qualifications
of the work force. If all the Puerto Rico born had remained on the island, the
average education of the Puerto Rican work force would be roughly 0.3 years
23. With market clearing, the elasticity of substitution (u)is (K‘ - L’) / (w’ - r‘). so that,
with fixed capital stock, L’ = -u(w’ - r’). The factor price frontier equation is
aw’ + (1 - a)r’ = 0, where the price of output is the numeraire. Substituting for r’, we obtain
L = -a{w’ + [a/(1 - a)]w’}, which simplifies toL = uw‘/(l - a), as in the text.
24. In 1980, compensation of employees was $7,202 million, national income was $9,722
million, and GNP was $11,031 million. Thus, labor’s share of national income was 0.74, or
approximately three-quarters, while its share of GNP was .65, or approximately two-thirds (see
Junta de Planificacion, “lnforme economico al gobernador,” 1981).In the 1950s. labor’s share of
output was smaller owing to the greater importance of agriculture.
203 Effect of the U.S.-Level Minimum Wage on Puerto Rico
lower, implying that wages would have been some 2 percent less owing to
educational qualifications. In addition, if we follow Ramos and interpret the
within-education differential in earnings between migrants and U.S.-born
Puerto Ricans as indicating the lower qualifications of migrants, we estimate
that the earnings power of the return migrants would be 13 percent less than
that of nonmigrants, reducing the productivity and earnings of the average
Puerto Rican by 4 percent. All told, even with a 0.15 elasticity of substitution,
we estimate that, absent migration (and further capital investments), real earnings
in Puerto Rico would conservatively be on the order of 25 percent lower
in 1988 than in fact they were. The implication is that, at the minimum, onequarter
of the long-term trend in real earnings on the island (an increase of
174 percent from 1951 to 1988) can be attributed to migration to the United
States.
6.5 Conclusion
This paper has shown that the imposition of the U.S.-level minimum wage
to Puerto Rico distorted the Puerto Rican earnings distribution, substantially
reduced employment on the island, reallocated labor across industries, and
affected the characteristics of migrants to the United States. In addition, we
argued that, absent migration of the less skilled, imposition of a U.S.-level
minimum on the island would have raised unemployment so much as to call
into question the viability of such a policy. Thus, migration was a prerequisite
for the high minimum wage. Our estimates indicate further that migration was
a major contributor to the growth of real earnings on the island. All told, the
massive migration from Puerto Rico demonstrates both the interplay between
domestic labor market policy-in this case, imposition of a minimum wage
with a bite-and migration and the potential contribution of migration to the
growth of real wages in a source economy.
Appendix A
Documentationfor the Puerto Rican MinimumWage
Time-SeriesData Set
Minimum Wage-Related Variables
Average minimumwages (AVEMIN) is a weighted average of forty-four industry
minimums (thirty-seven three-digit manufacturing and seven one-digit
industries). The data were gathered from the individual U.S. Department of
Labor reports (the “Minimum Wage Industry Studies”) that record the industry
minimums in the years when industry committees set minima. The reports
usually give minima for very detailed occupations. To arrive at a single minimum
wage for each industry, the data had to be amalgamated. Because em-
204 Alida J. Castillo-Freeman and Richard B. Freeman
ployment by occupation was unavailable, we took a simple average of the
occupational minimum.
Average coverage (AVECOV)is a weighted average of coverage for the eight
one-digit industries, based on table 1 of U.S. Department of Commerce
(1979, 2:634). All three-digit manufacturing industries are covered by the
same figure. The Department of Commerce table gives the number of wage
and salary workers covered by the changes in the minimum wage law (in 1966
and 1974). This number was divided by total employment in each industry to
determine the effect of the minimum on the entire economy. Since the law
changed in the middle of 1974, the coverage figure for that year is the average
between the 1973 and the 1975 numbers. We also created average coverage
excluding agriculture and government (COVAG),for wage and salary workers
(AVENCOV), and for wage and salary workers excluding agriculture and government
(NCOVAG).
Economy-wide coverage (COVT) is based on coverage figures for 1962,
1964, 1965, 1966, 1969, 1970, 1971, 1972, 1975, and 1976 from U.S. Department
of Labor, Employment Standards Administration (1977). For 1976,
the figures are the same as in table 1 of U.S. Department of Commerce (1979,
2:634). We divided the figures by total employment to obtain the coverage
number.
Average wage (AVEWAG)is the weighted average of the forty-four industry
average hourly earnings. The thirty-seven detailed three-digit manufacturing
earnings are from the Departamento del Trabajo’s “Census of Manufacturing
Industries” (1956-87). The Census was collected every October through 1981
and then not again until March 1983, so there are no figures for 1982. In the
time-series analysis, the 1982 figure is the average between 1981 and 1983.
For the years 1950-55, the Census was not conducted, so we applied the
change in one-digit manufacturing hourly earnings from each year to 1956 to
the 1956three-digit earnings, on the assumption that earnings in each detailed
sector changed at the same rate as the average in manufacturing. The one-digit
industry data were obtained from the “Salario semanal mediano de 10s empleados
asalariados por grupo industrial principal” (Departamentodel Trabajo
y Recursos Humanos). This source gives weekly earnings by month. To make
hourly earnings, we divided the July weekly earnings by thirty-two hours.
Average manufacturing wage (MFGWAG)comes from the Yearbook of Labour
Statistics (1950-87).
Kaitz minimum wage index (KAITZ) is the employment-weighted average of
coverage times minimudhourly earnings:
where a, is the share of employment in industry i, m,is the minimum in industry
i, w,is average hourly earnings in industry i, and c, is the coverage in that
industry. The index used the coverage, minimum, and hourly earnings figures
described above. We also created a Kaitz index for wage and salary workers
(NKAITZ).
205 Effect of the U.S.-Level Minimum Wage on Puerto Rico
The employment by industry numbers used in the weighting come from two
sources. The individual manufacturing industry numbers are from the Departamento
del Trabajo’s “Census of ManufacturingIndustries”for the three-digit
manufacturing industries (1956-87). To get the 1950-55 numbers, we took
the ratio of employment in all manufacturingin each of these years to employment
in all manufacturing in 1956 and multiplied this by the 1956 employment
in the detailed industry. For the remaining seven one-digit industries, the
employment numbers are from the Departamento del Trabajo’s “Seria historica
del empleo, desempleo y grupo trabajador en Puerto Rico.”
We used thirty-seven three-digit manufacturing industries:footwear, leather
gloves, electrical, women’s and children’sclothing, children’souterwear, corsets
and brassieres, men’s and boy’s clothing, leather handbags, women’s outerwear,
miscellaneous apparel, miscellaneous fabricated textiles, toys and
athletic goods, jewels and jewelry, costume jewelry, office and art supplies,
alcoholic beverages, cigars, tobacco, drugs, petroleum, chemicals, food,
household furniture, other furniture, sawmills, paper and allied products, cement,
cut stone and asbestos, portland cement and pottery, glass, sugar, textile
mill products, plastics, rubber, footwear, professional instruments, and machinery
and transportation equipment.
We used seven one-digit industries: transportation, construction, services,
trades, finance, agriculture, and public administration.
Macroeconomic Variables
Puerto Rican deflator (PRDEF)is from the Junta de Planificacion’s“Informe
economico a1gobernador,” 1954base year.
Puerto Rican GNP (PRGNP)is from the Junta de Planificacion’s “Informe
economica al gobernador,” 1954constant dollars.
These series Puerto Rican employment to population ratio (PREFQP and
PREPOPF) come from the monthly household surveysdone in Puerto Rico. The
PREFQP series combines two series: 1950-63 uses the fiscal year fourteen
years old and over, reported in the Departamentodel Trabajo’s “Seria historica
del empleo, desempleo y grupo trabajador en Puerto Rico” (table IV); 1964-
87 uses the calendar year sixteen years old and over numbers from the Departmento
del Trabajo’s “Empleo y disemployeo en Puerto Rico” (table 17), adjusted
to the 1980Census benchmark. The PREPOPF series uses the fiscal year
fourteen years old and over numbers for the entire time period, also adjusted
to a 1980Census benchmark.
Puerto Rican unemployment rate (PRUNEMP)is from the Departamento del
Trabajo’s “Seria.historica del empleo, desempleoy grupo trabajador en Puerto
Rico.” This series comes from the monthly household surveys done in Puerto
Rico similar to our Current Population Survey.
US.GNP (USGNP)is from the Economic Report of the President.
Table 6A. 1 gives figures for both the minimum wage-related variables and
the macroeconomic variables.
Table 6A.l Data in the Time-Series Analysis
Minimum Wage-related Variables Macroeconomic Variables
Year AVEMIN AVEWAGE KAITZ AVECOV COVT MFGWAGE PRDEF PREPOP PREPOPF PRGNP PRUNEMP USGNP Year Dummy 1974
1950 ,198
1951 ,209
1952 ,225
1953 .31I
1954 ,313
I955 ,369
I956 .447
1957 .488
1958 ,555
1959 .588
I960 .616
1961 .608
I962 .707
1963 .723
1964 .809
1965 ,834
,398 ,155 .201
,410 . I 6 4 .207
,421 ,180 .226
.480 ,229 ,231
,508 ,211 ,224
.547 ,231 ,236
.601 ,257 ,245
,685 ,251 .244
,716 .258 ,238
.789 ,266 .260
.840 ,268 ,270
.875 ,251 ,269
.933 .270 ,279
1.036 ,255 ,279
1.097 ,274 ,294
1.176 ,271 ,302
.29 .43
.29 .45
.29 .48
.29 S O
.29 .52
.29 .57
.29 .64
.29 .76
.29 .83
.29 .87
.29 .92
.29 .99
.29 1.06
.29 1.13
.31 1.18
.31 1.23
.859
.881
.953
,970
1.Ooo
I .003
1.011
1.035
1.089
1.110
1.138
I . 173
1.216
1.247
I .298
1.327
.470
,449
.434
,428
,415
,419
,412
.412
,397
.394
,403
,397
,385
,395
,396
,401
.470
.449
,434
,428
,415
.419
,412
,412
,397
,394
,403
,397
,385
,395
,396
,401
878.7
925.0
1,015.9
1,081.3
1,104.4
1,138.5
1,185.1
1,221.8
1,258.4
1,363.6
1,473.2
1,562.8
1,683.9
1,820.7
1,916.8
2,083.0
,154
,160
.I48
.I45
.I53
,132
.I33
,128
,142
.133
.I18
.127
.I28
,110
.I 12
,117
1,203.7
1,328.2
1,380.0
1,435.3
1,416.2
1,494.9
1,525.6
1,551.I
1,539.2
1,629.I
1,665.3
1,708.7
1,799.4
1,873.3
1,973.3
2,087.6
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
0
1966 .854
1967 ,971
1968 1.104
1969 1.149
1970 1.209
1971 1.224
1972 1.257
1973 1.262
I974 1.681
1975 1.871
1976 2.034
1977 2.198
1978 2.509
1979 2.768
1980 2.997
1981 3.264
1982 3.305
1983 3.350
1984 3.350
1985 3.350
1986 3.350
1987 3.350
1.288 .325
1.371 .365
1.512 ,381
1.667 ,360
1.856 .347
1.990 .331
2.144 .316
2.281 ,304
2.452 ,381
2.562 ,434
2.681 ,442
3.023 ,435
3.323 ,456
3.589 ,468
3.883 .461
4.181 ,467
4.318 ,461
4.456 ,454
4.498 .449
4.565 ,440
4.725 .426
4.879 .409
.444
.448
.455
,455
,458
,457
,449
.452
.544
,594
338
S90
.596
,594
,589
.587
.585
.583
,586
,579
,581
,582
.44 1.29
.44 1.39
.44 1.55
.49 1.65
.47 1.76
.47 1.87
.47 2.00
.47 2.13
.60 2.32
.66 2.56
.64 2.78
.64 3.02
.64 3.36
.64 3.69
.64 4.02
.64 4.39
.64 4.64
.64 4.83
.64 5.02
.64 5.19
.64 5.31
.64 5.33
1.358 .399
1.421 ,399
1.500 .403
1.552 .399
1.616 ,428
1.708 .423
1.780 ,423
1.817 ,421
1.946 ,405
2.082 .368
2.174 ,364
2.240 ,358
2.340 ,362
2.483 ,360
2.716 ,359
2.954 ,343
3.175 ,318
3.321 .321
3.461 ,334
3.548 ,331
3.697 ,351
3.787 ,369
,399
.399
,403
,399
.395
,398
,393
,389
,358
.336
,334
,364
,360
.360
,355
,328
.314
.329
,332
,338
,362
,377
2,223.2
2,328.4
2,455.3
2,684.0
2,901.4
3,075.6
3,215.9
3,450.3
3,493.6
3,424.7
3,461.6
3,623.5
3,817.4
4,025.O
4,076.8
4,127.0
3,976.5
3,894.8
4,048.4
4,172.8
4,281.6
4,496.7
,123
,116
,103
,103
.I07
,116
,119
,116
.I32
,181
.I95
.I99
,181
.170
,171
,199
,228
,234
,207
,218
,189
,168
2,208.3
2,271.4
2,365.6
2,423.3
2,416.2
2,484.8
2,608.5
2,744.1
2,729.3
2,695.0
2,826.7
2,958.6
3,115.2
3,192.4
3,187.1
3,248.8
3,166.0
3,279.1
3,501.4
3,607.5
3,713.3
3,819.6
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
0
0
0
0
0
0
0
0
1
2
3
4
5
6
7
8
9
10
11
12
13
14
208 Alida J. Castillo-Freeman and Richard B. Freeman
Appendix B
Log Earnings Equations for Men in Puerto Rico and for
Migrants to the United States
Log Wage Log Annual Earnings
1980 1970 1980 1970
Mig. PR Mig. PR Mig. PR Mig. PR
Grade com-
Experience:
plete
&5
6-10
11-15
16-20
21-25
26 +
Constant
R’
N
.036
(.003)
- .51
(.03
(W
- .40
- .15
~ 0 3 )
- .09
(.W
(.W
1.43
.04
6,247
- .05
. . .
.077
(.001)
- .47
(.02)
- .31
(.02)
- .21
(.02)
-.lo
(.02)
- .07
(.02)
. . .
.51
.12
26,193
.07I ,048
(.002) (.003)
-.39 -.96
(.04) (.05)
(.03) (.04)
(.02) (.04)
( . 0 3 (.W
-.07 -.04
(.03) (.04)
-.29 -.60
-.16 -.24
-.06 -.I1
-.07 8.76
.16 .08
8,694 6,215
,127 ,048
(.001) (.004)
-.96 -.52
(.02) (.07)
(.02) (.05)
(.02) (.05)
(.02) (.05)
(.02) (.05)
-.57 -.26
-.37 -.08
- .20 - .04
-.13 -.04
7.36 .69
.23 .07
26,743 1,912
,071
(.002)
- .40
~ 0 4 )
- .29
(.W
- .I6
(.02)
- .06
(.02)
- .08
(.W
. . .
- .07
.16
8,694
Source: 1980and 1970Puerto Rican and U.S. Censuses.
Nore: “Mig.* is Puerto Rican men who have migrated to the United States; “PR” is men in Puerto Rico.
Standard errors are given in parentheses.
Appendix C
The Effects of Age, Grade, and English Abilityon
Employment in Puerto Rico and Emigration
to the United States
Probability Not
Employed or Paid Probability Migrant
5$2.95 in Puerto Rico Probability Migrant Adjusted for English
Controls Male Female Male Female Male Female
Age:
1620
21-25
2630
31-35
3640
41-45
4650
51-55
56-60
61-64
Grade:
0-6
7-8
9-1 1
12
13-15
16+
English ability:
Well
Not well
Not at all
Constant
R2
N
Source: 1980Puerto Rican and U.S. Censuses.
Note: Standard errors are given in parentheses.
210 Alida J. Castillo-Freeman and Richard B. Freeman
References
Brown, Charles. 1988. Minimum Wage Laws: Are They Overrated? Journal of Economic
Perspectives (Summer), pp. 133-43.
Brown, Charles, Curtis Gilroy, and Andrew Kohen. 1982. Effect of the Minimum
Wage on Employment and Unemployment: A Survey. National Bureau of Economic
Research Working Paper no. 846.
Castillo, Alida Josefina. 1983. Jobless in the Sun: A Study of the Effect of The Federal
Minimum Wage on Employment in Puerto Rico. Undergraduate honors thesis, Harvard
University.
Centro de Estudios Puertoriquenos. 1979. Labor Migration under Capitalism: The
Puerto Rican Experience. London: Monthly Review Press.
Characteristics of the Population: Puerto Rico. 1950, 1960, 1970, 1980. U.S. Census
of Population. Washington, D.C.: US. Bureau of the Census.
Departamento del Trabajo y Recursos Humanos. Various years. Census of Manufacturing
Industries of Puerto Rico. San Juan, P.R.
.Various years. Empleo y desempleo en Puerto Rico. San Juan, P.R.
.Various years. Empleo y salario promedio por hora en las industias manufactureras
de Puerto Rico. San Juan, P.R.
. Various years. Salario semanal mediano de 10s empleados asalariados por
grupo industrial principal. San Juan, P.R.
. Various years. Seria historica del empleo, desempleo y grupo trabajador en
Puerto Rico. San Juan, P.R.
Economic Report of the President. Various years. Washington, D.C.: U.S. Government
Printing Office.
Falcon, Luis. 1990. Migration and Development: The Case of Puerto Rico. Working
Paper no. 18. Washington, D.C.: Commission for the Study of International Migration
and Cooperative Economic Development, February.
General Social and Economic Characteristics. 1980. Washington, D.C.: U.S. Bureau
of the Census.
Junta de Planificacion de Puerto Rico. 1984. Encuesta de migration, 1982-83. San
Juan, P.R.
. Various years. Informe economico a1gobernador. San Juan, P.R.
Mincer, J. 1976. Unemployment Effects of Minimum Wages. Journal of Political
Economy 84:87-104.
Puerto Ricans in the U.S. 1950, 1960, 1970. Special report, U.S. Census of Population.
Washington, D.C.: U.S. Bureau of the Census.
Santiago, C. 1990. Minimum Wages and Migration in Puerto Rico. Paper presented at
the Inter-University Consortium for Latino Research and the Social Science Research
Council, Miami, 11March.
Statistical Abstract of the United States. Various years. Washington, D.C.: US. Department
of Commerce, Bureau of the Census.
U.S. Department of Commerce. 1979. Economic Study ofPuerto Rico. 2 vols. Washington,
D.C.: U.S. Government Printing Office.
U.S. Department of Labor. 1977. Estimated Number of Nonsupervisory Employees
Subject to the Minimum Wage Provisions of the Fair Labor Standards Act in Puerto
Rico, 1962-1976. Washington, D.C.: Employment Standards Administration, 16
May.
. 1978. Handy Reference Guide to the Fair Labor Standards Act. WH Publication
no. 1282. Washington, D.C.
. Various years. Minimum Wage Industry Studies. Washington, D.C.: U.S.
Department of Labor Library.
211 Effect of the U.S.-Level Minimum Wage on Puerto Rico
Welch, Finis. 1978. Minimum Wages:Issues and Evidence. Washington, D.C.:AmerYearbook
of Labour Statistics. Various years. Geneva:InternationalLabour OrganisaicanEnterprisefor
Public Policy Research.
tion.
This Page Intentionally Left Blank