The Effect of Low-Skilled Immigration on U.S.
Prices: Evidence from CPI Data
Patricia Cortes
University of Chicago
I exploit the large variation across U.S. cities and through time in the
relative size of the low-skilled immigrant population to estimate the
causal effect of immigration on prices of nontraded goods and services. Using an instrumental variables strategy, I find that, at current
immigration levels, a 10 percent increase in the share of low-skilled
immigrants in the labor force decreases the price of immigrantintensive services, such as housekeeping and gardening, by 2 percent.
Wage equations suggest that lower wages are a likely channel through
which these effects take place. However, wage effects are significantly
larger for low-skilled immigrants than for low-skilled natives, implying
that the two are imperfect substitutes.
I.
Introduction
Most research to date on the impact of low-skilled immigration on the
U.S. economy has focused on native wage levels. The net effect of immigration on natives’ purchasing power, however, depends not only on
wage but also on price effects. If immigration bids down the price of
I am very grateful to Joshua Angrist, David Autor, and Esther Duflo for their support
and valuable comments. Suggestions from an anonymous referee, a second editor, and
Steven Levitt (the editor) substantially improved the paper. I thank Daron Acemoglu,
Abhijit Banerjee, Marianne Bertrand, Jose Tessada, and participants in the field lunches
at Massachusetts Institute of Technology and in seminars at University College London,
University of California at Los Angeles, San Diego, and Irvine, Chicago Graduate School
of Business, Stanford Graduate School of Business, and Northwestern for helpful comments. I also thank the Consumer Price Index unit of the Bureau of Labor Statistics, in
particular William Cook, for generous assistance with the data. This work is supported by
the William Ladany Faculty Research Fund at the Graduate School of Business, the University of Chicago.
[ Journal of Political Economy, 2008, vol. 116, no. 3]
䉷 2008 by The University of Chicago. All rights reserved. 0022-3808/2008/11603-0004$10.00
381
382
journal of political economy
low-skilled labor, this will reduce the price of unskilled-intensive goods
and services, thereby raising the welfare of consumers of these goods.
This paper uses confidential micro data from the consumer price
index (CPI) and exploits the large variation across cities and through
time in the relative size of the low-skilled immigrant population to estimate the causal effect of immigration on prices of nontraded goods
and services. The empirical strategy presents two main challenges. First,
immigrants do not choose their locations randomly. Unobserved economic factors that attract immigrants to a city are likely to also affect
prices, thereby biasing the estimation. Second, the effect of immigration
on prices of nontraded goods will depend not only on the relative
immigrant labor intensity of the nontraded good’s technology but also
on the degree to which natives respond to immigration to a local market
by moving their labor or capital to other cities. Displacement effects
will disperse the price effect across cities, and the empirical exercise
will provide only a lower bound for the full magnitude of the effects.
To ameliorate the bias that arises from endogenous location choices
of immigrants, I use the historical distribution of immigrants from countries for which data are available in 1970 as an instrument for the recent
distribution of the immigrant population. The instrument is based on
the notion that immigrant networks are an important consideration in
the location choices of prospective immigrants and that the determinants of the historical distribution of immigrants are likely to be uncorrelated with recent changes in the relative economic performance
of cities in the same geographic region.
As initial evidence of the impact of low-skilled immigration, I present
reduced-form price equations for nontraded services that use intensively
low-skilled immigrant labor. The explanatory variable of interest is the
share of low-skilled workers in the labor force. The instrumental variable
estimates imply that at current immigration levels, a 10 percent increase
in the share of low-skilled immigrants in the labor force of a city reduces
prices of immigrant-intensive services, such as gardening, housekeeping,
babysitting, and dry cleaning, by approximately 2 percent. The magnitude of the effect suggests that the immigration wave of the 1980–
2000 period decreased the prices of immigrant-intensive services by at
least a city average of 9–11 percent.
I estimate a similar specification for the prices of nontraded services
that use low-skilled immigrant labor moderately and for services that
use it very little. I find that the coefficient for the former is negative
but significantly smaller in magnitude than for immigrant-intensive services, and zero for the latter.
An alternative specification is to interact the share of low-skilled workers in a city with a measure of how low-skilled immigrant labor intensive
a nontraded good or service is. Using Census data, I construct the share
low-skilled immigration and u.s. prices
383
of total workers in the industry who were foreign-born high school
dropouts. The direct effect of low-skilled immigrants on the prices of
nontraded goods is close to zero and not statistically significant. The
interaction term with the technology variable is highly negative and
statistically significant.
The focus on local conditions limits the set of goods and services
whose prices can be analyzed to those considered nontradable at the
city level. Consistent with the hypothesis that U.S. cities could be considered small open economies, I find that the local concentration of
low-skilled immigrants has little impact on the prices of traded goods.
This is true in all the specifications.
I next turn to explore the channels through which the price effects
are likely to take place. Given the nature of the services in which immigrants work, the obvious channel through which the prices go down
after a low-skilled immigration shock is a reduction in wages. Most crosscity studies (including the present), however, have failed to find sizable
effects on low-skilled native wages. A simple trade model that allows lowskilled immigrants and low-skilled natives to be imperfect substitutes
provides a solution to the puzzle of the joint finding of sizable price
effects and no native wage effects. If natives and immigrants are imperfect substitutes, a low-skilled immigration shock should affect mostly
the wages of other low-skilled immigrants and have a smaller effect on
the wages of typical low-skilled natives. Immigrant wage regressions show
significant negative effects for female immigrants but fail to find a similar
result for immigrant males. Simple tabulations of Census data, combined
with inconclusive results from previous studies, suggest that wage data
for low-skilled immigrants might be noisy. To indirectly tackle this issue,
I show that low-skilled immigration has had a large negative effect on
the wages of the low-skilled natives who are most likely to compete
directly with immigrants in the labor market, those of Hispanic origin,
and those with low English proficiency. The magnitude of the wage
effects for these groups is comparable to the magnitude of the price
effects.
No native wage effects at the city level do not imply that immigrants
have no impact on natives’ labor opportunities. If displacement effects
are sizable for natives, the cross-city comparison will not capture the
full impact on natives’ wages. Following Borjas (2006), I estimate wage
regressions in which the labor market is defined at the state level. I find
larger and now statistically significant effects of an immigrant influx on
the wages of low-skilled natives, suggesting that there is some native
migration response to immigration shocks at the city level. Note, however, that the presence of displacement effects does not imply that immigrants and natives are perfect substitutes. If they were, wage effects
384
journal of political economy
estimated at the city level would have to be of a similar magnitude for
natives and immigrants.
Section V of the paper uses data on consumption patterns from the
Consumer Expenditure Survey (CEX) to assess which groups of the
native population benefit the most from a decline in the prices of
immigrant-intensive services and by how much this decline reduces the
cost of their consumption basket. I find that price effects resulting from
the immigration wave of the past two decades were larger for highskilled natives, especially for those with a graduate degree (⫺0.4 percent), rather than for low-skilled natives (⫺0.3 percent), because they
devote a larger share of their budget to immigrant-intensive services.
However, even for very high-skilled natives, the effect is small: expenditure shares of immigrant-intensive services are never higher (on average) than 4.3 percent of the total budget.
Very few papers have looked at immigration’s impact on the prices
of goods and services. A recent exception is the paper by Lach (2007),
who studies how an unexpected arrival of a large number of immigrants
from the former Soviet Union to Israel affected the prices of goods and
services. Exploiting cross-city variation in the concentration on immigrants, he finds that a one-percentage-point increase in the ratio of
immigrants to natives decreased prices by 0.5 percentage point. He
argues that the higher elasticity of demand of immigrants and their
lower search costs gave incentives to retailers to lower their prices to
attract the new customers. His proposed mechanism is unlikely to explain the results obtained here for several reasons. First, immigrants
consume very little of household-related services. Second, U.S. lowskilled immigrants, unlike the immigrants who arrived in Israel, have
very high labor force participation rates and therefore do not have
particularly low search costs compared with natives. Finally, Lach’s estimates are for the short run; as he mentions, the effects are likely to
decline in the long run as immigrants become assimilated.1
However, a vast literature has looked at the wage effects of immigration. My estimates of the local impact of low-skilled immigration on
natives’ wages are in line with what most other cross-city studies have
found (see Card 1990, 2005; Altonji and Card 1991; LaLonde and Topel
1991; Card and Lewis 2007): the effect of immigration on the labor
market outcomes of natives is small. A few hypotheses have been proposed to explain this finding. Borjas, Freeman, and Katz (1996) and
Borjas (2003, 2006) argue that if labor and capital adjust to immigration
by moving across cities, then the relevant unit of analysis is the entire
1
Khananusapkul (2004) is also related to the present study. The author’s finding that
an increase in the proportion of low-skilled female immigrants in a metropolitan area
raises the proportion of private household workers and lowers their wages supports the
hypothesis presented here.
low-skilled immigration and u.s. prices
385
country, and cross-city comparisons will fail to find significant effects.
Lewis (2005), however, claims that local economies are the relevant unit
of observation, but that the technologies of local firms—rather than the
wages that they offer—respond to changes in local skill mix associated
with immigration. The hypothesis that cities adapt to immigration by
shifting industry composition is rejected by Card and Lewis (2007).
My estimations provide evidence for Borjas’s hypothesis, but they also
imply that displacement effects cannot be the only explanation; exploiting cross-city variation, I find large negative wage effects for immigrants and native Hispanics. These results point to a complementary
reason: low-skilled natives and low-skilled immigrants are far from being
perfect substitutes. Therefore, a low-skilled immigration shock should
affect mostly the wages of other low-skilled immigrants and have a
smaller effect on the wages of low-skilled natives.2
The rest of the paper is organized as follows. In Section II, I describe
the data and descriptive statistics and discuss industry variation in the
use of low-skilled immigrant labor. A simple theoretical framework is
presented in Section III. Empirical results are reported and discussed
in Section IV. Purchasing power calculations are reported in Section V,
and Section VI presents conclusions.
II.
Data
A.
Immigration Data
This paper uses the 1980, 1990, and 2000 Public Use Microdata Samples
(PUMS) of the Decennial Census to measure the concentration of lowskilled immigrants among cities. Low-skilled workers are defined as those
who have not completed high school. An immigrant is defined as someone who reports being a naturalized citizen or not being a citizen. The
analysis is restricted to people aged 16–64 who report being in the labor
force.
Table 1 shows the evolution of the share of low-skilled immigrants in
the labor force for the 30 largest cities in the United States. Two facts
should be emphasized. First, there is substantial variation across cities
in the concentration of low-skilled immigrants. Immigrants are heavily
concentrated in large cities, such as Los Angeles, New York, and Miami.
In Los Angeles, for example, one of six workers is a high school dropout
immigrant. In other smaller cities, low-skilled immigrants are a negligible share of the labor force. In Cincinnati, for example, in 2000 there
were fewer than five low-skilled immigrant workers per 1,000 participants
in the labor force. Second, during the 1990s, new waves of low-skilled
2
Most of the literature to date has assumed perfect substitutability between immigrants
and natives. A recent exception is Ottaviano and Peri (2006).
386
journal of political economy
TABLE 1
Share of Low-Skilled Immigrants in the Labor Force (%)
City
1980
1990
2000
Atlanta
Baltimore
Boston
Buffalo
Chicago
Cincinnati
Cleveland
Columbus
Dallas–Fort Worth
Denver-Boulder
Detroit
Honolulu
Houston
Kansas City
Los Angeles
Miami
Milwaukee
Minneapolis
New Orleans
New York
Philadelphia
Phoenix
Pittsburgh
Portland
St. Louis
San Diego
San Francisco
Seattle
Tampa
Washington, DC
.38
.76
3.53
1.48
4.99
.44
1.82
.43
2.13
1.18
1.76
4.71
3.96
.58
11.64
15.13
1.07
.49
1.20
8.91
1.39
2.19
.57
1.03
.49
4.59
4.40
1.22
1.50
1.61
.84
.44
2.71
.72
5.09
.23
.89
.25
5.17
1.42
.93
3.66
7.03
.47
15.90
14.44
.84
.37
1.13
7.82
.91
3.30
.27
1.53
.24
5.92
6.73
1.00
1.69
2.52
3.23
.67
2.62
.47
5.86
.34
.65
.81
8.63
4.13
1.35
3.18
9.21
1.44
15.09
11.36
1.54
1.43
1.08
8.15
1.06
6.41
.21
3.27
.53
6.34
6.19
1.94
2.15
3.76
Source.—U.S. Census.
immigrants chose to locate to new cities. So, despite the large flows of
new immigrants to the country, Los Angeles, New York, and Miami did
not see an increase in the share of high school dropout immigrants in
their labor forces. Cities such as Denver, Dallas, Washington, DC, and
especially Atlanta, however, experienced a significant increase in the
concentration of low-skilled immigrants.
B.
Price Data
Under a confidentiality agreement with the Bureau of Labor Statistics
(BLS), I was granted access to the CPI Research Database (RDB). This
data set contains the store-level data used to construct the CPI. It also
includes estimates of price indexes at lower levels of geography and
product classification that are not available to the public. The RDB
covers the years 1986–2002. To increase the number of observations, I
low-skilled immigration and u.s. prices
387
(linearly) interpolate the immigration variables for 1986, so that I have
price data for 30 cities and three time periods for each good.3
The paper uses price indexes at the city-industry level. The analysis
is restricted to A-sized cities as defined by the BLS (metropolitan areas
with 1980 populations greater than 1.2 million), where sufficient quotes
are collected to produce reliable indexes.4 The number of goods and
services included in the CPI that can be used in the present analysis is
restricted by the ability to match them to the industry classification of
the census. The paper uses the Census data to construct a measure of
the industry’s low-skilled labor factor share. The matching process results in a sample of 74 goods and services, 34 of which are nontradable
(see App. table A1 for a list of goods and services).5
C.
Industries Intensive in Low-Skilled Immigrant Labor
There is much between-industry variation in the use of low-skilled immigrant labor. Apparel is a good example of a low-skilled-immigrantintensive industry: 38 percent of all workers in the industry are lowskilled immigrants, close to eight times the percentage in the overall
labor force. However, there are industries with virtually no low-skilled
immigrants: in the accounting services industry, for example, fewer than
one of every 1,000 employees is a foreign-born high school dropout.
Table 2 shows the industries with the highest share of low-skilled
immigrants, low-skilled female immigrants, and low-skilled male immigrants in the year 2000. With the exception of agriculture and apparel/
textiles, almost all other industries fall into the category of nontraded
services: landscaping, housekeeping, laundry and dry cleaning, car
washes, shoe repair, and services for buildings and dwellings. Once other
inputs are taken into account, it is very likely that some of these services
are significantly more immigrant labor intensive than the top traded
industries in table 2.6 The low-skilled immigrant concentration in these
nontraded services is very large. For example, whereas low-skilled im3
All the results are robust to restricting the empirical analysis to price changes between
1990 and 2000 or to using 1980 immigration variables for 1986 prices. See App. table C1.
4
A revision to the CPI in 1998 changed some of the geographic area samples and
incorporated a new item structure. Buffalo and New Orleans, A-sized units before the
revision, were excluded from this group after the revision, and Washington, DC, and
Baltimore were merged into a single unit. For these cities, in an effort to maximize the
number of observations, the present analysis assigns the index value of December 1997
to January 2000. A similar strategy was followed for items that did not have a perfect
match in the new structure.
5
Because of confidentiality restrictions, I cannot present detailed statistics of the price
data used in this paper.
6
Data from the 2002 Economic Census show that payroll is half the costs of materials
and capital expenditures for the textile products industry and double for the apparel
industry.
TABLE 2
Top Industries Intensive in Low-Skilled Immigrant Labor
Immigrants (%)
Natives (%)
A. All Low-Skilled Workers
Labor force
Apparel/textile industry
Landscaping services
Private households
Crop production
Animal slaughtering
Fruit and vegetable preservation
Car washes
Services to buildings
Shoe repair
Furniture and fixtures
Dry cleaning and laundry services
Sugar products
5.2
37.8
29.7
27.0
25.3
24.7
20.1
19.9
19.6
19.6
19.4
19.2
19.2
6.4
7.7
10.8
9.3
12.9
12.2
8.7
23.2
11.9
10.6
8.4
12.6
9.8
B. Male Low-Skilled Workers
Labor force
Landscaping services
Crop production
Shoe repair
Car washes
Apparel/textile industry
Animal slaughtering
Furniture and fixtures
Recyclable material
Bakeries, except retail
Construction
Taxi and limo service
Coating, engraving activities
3.3
28.9
20.0
19.6
17.4
16.4
16.0
15.4
13.5
12.0
11.9
10.6
10.4
3.6
9.9
10.1
9.5
20.1
1.4
8.1
6.5
11.8
5.8
8.8
4.0
9.6
C. Female Low-Skilled Workers
Labor force
Apparel/textile industry
Private households
Fruit and vegetable preservation
Dry cleaning and laundry services
Sugar products
Services to buildings
Animal slaughtering
Traveler accommodation
Pottery, ceramics
Nail salons
Bakeries, except retail
Seafood products
1.9
28.7
25.4
12.4
12.2
11.3
11.3
8.7
8.0
7.7
7.1
6.4
6.3
2.8
2.6
8.1
4.3
8.8
5.8
5.8
4.0
5.4
2.2
5.1
3.9
2.3
Source.—2000 Census.
Note.—Percentage represents the share of low-skilled immigrants in total employment of the industry. The
30 largest U.S. cities are included.
low-skilled immigration and u.s. prices
389
migrant women represented 1.9 percent of the total labor force in the
year 2000, they represented more than 25 percent of the workers in
private household occupations and 12 percent of the workers in laundry
and dry cleaning services. Similarly, the immigrant men’s share in gardening was nine times larger, and their share in shoe repair six times
larger, than their share in the total labor force.
The concentration of low-skilled immigrants in these industries is not
solely an outcome of their low education level; indeed, native high
school dropouts are much less likely to work in these industries (see
table 2). Gardening and housekeeping are good examples: whereas 60
percent of high school dropouts in the labor force in 2000 were natives,
they make up only close to a quarter of the dropouts working in one
of these two services. Immigrants’ low proficiency in English and, for
some, their legal status might limit their job opportunities compared
with low-skilled natives.
III.
Theoretical Background
In this section I use a simple Heckscher-Ohlin framework to model the
impact of immigration on the prices of nontraded goods. The presence
of a nontraded sector in the model breaks the “factor price insensitivity”
result of a two-traded-sectors model (Leamer 1995), and thus, wages of
low-skilled workers will be determined locally. Wages are going to be
the main mechanism through which the impact on prices occurs. A key
feature of the model is that low-skilled immigrants and low-skilled natives
need not be perfect substitutes. This is important because it will help
reconcile the seemingly contradictory findings of negligible native wage
effects and significant price effects of a low-skilled immigration influx.
The many simplifying assumptions (the model abstracts from the possibility of low-skilled immigration affecting prices through changes in
demand, wages of high-skilled workers, or returns to capital) require
that when comparing the model’s predictions with the results of the
empirical model, one has to be careful in interpreting the econometric
estimates as informative structural parameters. Nevertheless, the model
does provide a simple framework to test the economic plausibility of
the empirical results and introduces some issues that will become relevant in the analysis of the results later on.
A.
Setup
Consider a small open economy that produces two goods, one traded
(T) and one nontraded (NT). There are three factors of production:
high-skilled native labor (H), low-skilled native labor (L), and low-skilled
immigrant labor (I ). The total supply of factors is represented by H, L,
390
journal of political economy
and Ī , respectively. For simplicity, I assume that only high-skilled native
labor participates in the production of the traded good:
T p HT .
(1)
The nontraded good production function is a nested constant elasticity
of substitution:
a
r
NT p HNT
[(LrNT ⫹ INT
)1/r]1⫺a,
(2)
where 0 ! r ≤ 1 and 0 ! a ! 1. This specification implies that the elasticity of substitution between the low-skilled labor aggregate and highskilled labor is equal to one and that the elasticity of substitution between
L and I is j p 1/(1 ⫺ r). This specification allows for perfect substitution
between immigrant and native low-skilled labor (j p ⬁), for a CobbDouglas specification between the two factors (j p 1), and for perfect
complementarity between them (j p 0).
To keep the analysis simple, I have excluded capital from the production functions. Doing so is equivalent to keeping the supply of capital
perfectly elastic, a reasonable assumption for local markets.
The economy admits a representative consumer with a Cobb-Douglas
type utility:
U p T g(NT )1⫺g.
(3)
Note that the assumption of homotheticity of the utility function implies
no income effects and that there is no role for differences in preferences
between natives and immigrants.
I assume that all markets are competitive. The economy takes as given
the price of the tradable good PT , which is normalized to one.
B.
Equilibrium
The maximization of utility leads consumers to spend a fraction g of
their income in the consumption of the traded good and 1 ⫺ g in the
consumption of the nontraded good. This condition plus market clearing in the nontraded market imply that the following equation holds:
a
HNT
[(Lr ⫹ I¯ r)1/r]1⫺a p
(1 ⫺ g)(wH H ⫹ wLL ⫹ wII¯ )
PNT
,
(4)
where the left side of equation (4) represents the total supply of the
nontraded good and the right side the total demand. Note that I have
already incorporated the conditions L NT p L and INT p I¯ .
Because all factors will be paid the value of their marginal product
in competitive markets and because the marginal product of high-skilled
low-skilled immigration and u.s. prices
391
workers should be equal in both sectors, the following equilibrium equations result:
a
wL p (1 ⫺ a)PNTHNT
Lr⫺1(Lr ⫹ I¯ r)[(1⫺a)/r]⫺1,
(5)
a ¯ r⫺1
wI p (1 ⫺ a)PNTHNT
I (Lr ⫹ I¯ r)[(1⫺a)/r]⫺1,
(6)
a⫺1
1 p aPNTHNT
[(Lr ⫹ I¯ r)1/r]1⫺a,
(7)
and
where the right sides of equations (5), (6), and (7) represent the value
of the marginal product of low-skilled natives, low-skilled immigrants,
and high-skilled workers, respectively.
Equations (4)–(7) provide a system of four equations and four unknowns (PNT , HNT , wL , and wI). Solving the system, I obtain that the
equilibrium relative price of nontraded goods is given by
⫺(1⫺a)
(Lr ⫹ I¯ r)1/r
(1 ⫺ g)1⫺a
PNT p a
#
.
a [a ⫹ g(1 ⫺ a)]1⫺a
H
[
]
(8)
Equation (8) shows that the relative price of the nontraded good
depends positively on the consumer’s preference for it, 1 ⫺ g, and negatively on the relative supply of low-skilled versus high-skilled labor. An
increase in H will raise the relative production of the traded good (and
therefore reduce its relative price), the more so the higher 1 ⫺ a is,
that is, the less intensive the nontraded good is in high-skilled labor.
The same logic explains why the relative price will decrease as low-skilled
labor becomes more abundant.
Equilibrium wages for low-skilled natives and low-skilled immigrants
are given by
⫺1
( )
Lr ⫹ I¯ r
1⫺g
r⫺1
wL p (1 ⫺ a)
#L #
a ⫹ g(1 ⫺ a)
H
(9)
and
⫺1
( )
Lr ⫹ I¯ r
1⫺g
wI p (1 ⫺ a)
# I¯ r⫺1 #
a ⫹ g(1 ⫺ a)
H
.
(10)
If immigrants and natives are perfect substitutes, then both groups
will receive the same wage rate, which will depend negatively on the
relative supply of low- versus high-skilled labor.
392
C.
journal of political economy
The Effect of an Immigration Shock
To determine the effect of an immigration shock on prices, I begin by
taking logs of equation (8):
ln (PNT) p c ⫺ (1 ⫺ a) ln
(Lr ⫹ I¯ r)1/r
[
]
H
,
(11)
where
c p ln
{
}
(1 ⫺ g)1⫺a
.
a [a ⫹ g(1 ⫺ a)]1⫺a
a
Differencing (11) with respect to low-skilled immigration, I obtain
the elasticity of the relative price of the nontraded good to low-skilled
immigration:
⭸ ln [(Lr ⫹ I¯ r)1/r/H ]
⭸ ln PNT
p
⫺(1
⫺
a)
! 0.
⭸ ln I¯
⭸ ln I¯
(12)
Equation (12) shows that the elasticity of the relative price of the
nontraded good to a shock to low-skilled immigration depends on two
factors: the low-skilled labor intensity of the nontraded good (1 ⫺ a)
and immigration’s effect on the relative amount of aggregate low-skilled
labor to high-skilled labor. This last effect can be decomposed as follows:
⭸ ln [(Lr ⫹ I¯ r)1/r/H ]
Lr
⭸ lnL
Ī r
p
#
⫹
.
r
r
r
⭸ ln I¯
⭸ ln I¯ L ⫹ I¯ r
L ⫹ I¯
(13)
If immigrants displace natives, ⭸L/⭸I¯ ! 0,7 the local price effects of
immigration will be attenuated. If displacement effects are strong
enough, one to one in the case of perfect substitution, price effects will
disappear. However, if displacement effects are negligible, then
(
)
⭸ ln PNT
I¯ r
p
⫺(1
⫺
a)
! 0,
r
⭸ ln I¯
L ⫹ I¯ r
(14)
and the elasticity of prices with respect to a low-skilled immigration
shock will be proportional to the initial share of immigrants in the lowskilled labor aggregate.
The negative effect of an immigration shock on the relative price of
7
Note that, for simplicity, I have assumed that L and I are perfectly inelastic. Therefore,
allowing ⭸L/⭸I¯ ( 0 refers to the possibility that natives move to another city, and not to
people changing their labor supply decisions.
low-skilled immigration and u.s. prices
393
the nontraded good will hold under more general constant-returns-toscale technologies if two conditions are satisfied: there are more factors
than traded goods and the nontraded sector is more intensive in lowskilled immigrant labor than the traded sector. The magnitude of this
effect will depend on the difference in skill intensities between the two
sectors.8
As shown in equation (15) below, the effect of low-skilled immigration
on the wages of low-skilled natives depends on two key factors: the
elasticity of substitution between the two inputs and the degree to which
immigrants displace natives:
⭸ lnL
Lr
⭸ ln wL
I¯ r
p
⫺
1
⫺
r
⫹
r
⫺
r
.
⭸ ln I¯
⭸ ln I¯
Lr ⫹ I¯ r
Lr ⫹ I¯ r
(
)
(15)
This equation makes clear that a very small response of wages of lowskilled natives to an influx of low-skilled immigrants can be the result
of two very different forces. On the one hand, wages will not change
significantly if natives leave the city or the labor force, an argument
made by the critics of cross-city studies (Borjas 2003). On the other
hand, even if ⭸ lnL/⭸ ln I¯ p 0, wages of natives will not react much if
native labor and immigrant labor are not close substitutes (r r 0).
A way to disentangle the two alternatives is to look at the wages of
low-skilled immigrants:
⭸ lnL Lr
⭸ ln wI
I¯ r
p
⫺(1
⫺
r)
⫺
r
⫺
r
.
r
r
r
⭸ ln I¯
⭸ ln I¯ L ⫹ I¯
L ⫹ I¯ r
(16)
If the presence of displacement effects is the major explanation behind the low response of native wages, immigrant wages should not
react much either. If imperfect substitution plays an important role,
wages of low-skilled immigrants should decrease significantly after an
immigration flow, even if displacement effects are nonzero.
8
Specifically, if the traded good’s production function is
T p HTb[(LTr ⫹ ITr)1/r]1⫺b,
it is easy to show that
r
⭸ ln [(L ⫹ I¯ r)1/r/H]
⭸ ln PNT
p
(a
⫺
b)
#
.
⭸ ln I¯
⭸ln I¯
394
IV.
journal of political economy
Empirical Strategy and Estimations
The paper exploits the large variation across cities and through time
in the relative size of the low-skilled immigrant population to identify
the impact of immigration on prices and wages. There are two major
concerns with the validity of this identification strategy. The first one is
that immigrants do not choose their location randomly, and thus location decisions might be correlated with unobserved determinants of
prices and wages. Subsection A discusses the instrument I use to solve
for the endogeneity of the distribution of immigrants.
The second concern is that natives might respond to the influx of
low-skilled immigrants to a city by changing their location decisions. As
the model from the previous section shows, if there are displacement
effects (⭸L/⭸I¯ ! 0), cross-city comparisons will not capture the full effects
of low-skilled immigration on prices and wages. If displacement effects
are not very large, however, estimates using variation at the city level
will provide informative lower bounds on the full magnitude of the
effects and will also help explain differences across cities in prices and
in the wages of low-skilled natives and immigrants.
Most of the papers that have empirically tested natives’ migration
response to immigration have not found evidence of large displacement
effects. Card and DiNardo (2000) and Card (2001, 2005), using different
samples and specifications, have found that native mobility has virtually
no offsetting effect on the relative supply shocks created by immigration.
Evidence that immigrants do displace natives, but at low rates, is provided by Federman, Harrington, and Krynski (2006), who found that
for every five Vietnamese manicurists who entered the market in California, two non-Vietnamese were displaced, and that the displacement
appears to have been primarily due to a reduction in the number of
non-Vietnamese entering the occupation rather than to an increase in
the number of current manicurists leaving it. Saiz (2007) argues that
his result that immigration has changed the local rents and house prices
is consistent with the idea that immigrants do not displace natives, at
least not one for one.
Larger but still not perfectly offsetting displacement effects are found
by Borjas (2006). He estimates that 6.1 fewer native workers choose to
reside in a city for every 10 new immigrants that arrive to the city.
After discussing the instrument, I will provide evidence that, even
though natives do respond to the entry of immigrants to a city, displacement effects are not nearly large enough to equalize prices and
wages across cities.
low-skilled immigration and u.s. prices
A.
395
Instrument
The instrument exploits the tendency of immigrants to settle in a city
with a large enclave of immigrants from the same country. Immigrant
networks are an important consideration in the location choices of prospective immigrants because these networks facilitate the job search
process and assimilation to the new culture (Munshi 2003). The instrument uses the 1970 distribution of immigrants from a given country
across U.S. cities to allocate the new waves of immigrants from that
country. For example, if a third of Mexican immigrants in 1970 were
living in Los Angeles, the instrument allocates 33 percent of all Mexicans
in the 1990s to Los Angeles.
Formally, the instrument for the number of low-skilled immigrants in
city i and decade t can be written as
冘
j
Immigrantsji1970
# LS Immigrantsjt ,
Immigrantsj1970
(17)
where j are all countries of origin included in the 1970 Census,9
Immigrantsji1970 /Immigrantsj1970 represents the percentage of all immigrants from country j included in the 1970 Census who were living
in city i, and LS Immigrantsjt stands for the total number of low-skilled
emigrants from country j to the United States in decade t. As can be
seen in table 3, the instrument is a good predictor of low-skilled workers
and immigrant shares.10 The magnitudes of the coefficients suggest that,
at current U.S. immigration levels, an increase of 10 percent in the
predicted number of low-skilled immigrants increases the share of lowskilled immigrants in the labor force between 6.2 percent and 8.1 percent and the share of low-skilled workers between 2.7 percent and 3.5
percent.
Most of the econometric specifications in the paper include city and
region#decade fixed effects. Therefore, the instrument will help in
identifying the causal effect of immigration concentration on prices and
wages as long as three conditions hold:
9
A list of the countries included in the construction of the instrument is presented in
App. table B3.
10
Appendix table B1 presents the first-stage regressions that use the 1986 interpolated
values instead of the 1980 Census data. The results are quantitatively and qualitatively very
similar. I use a logarithmic functional form in table 3 because the price equation derived
from the theoretical model (Sec. II) is expressed in logs. Appendix table B2 presents an
alternative functional form for the first stage that will be used in Sec. IV.B and serves as
a check on the robustness of the instrument.
冘
.797
(.173)***
No
Yes
.950
90
.623
(.198)***
Yes
Yes
.960
90
(2)
.809
(.206)***
Yes
No
.958
81
(3)
.350
(.061)***
No
Yes
.973
90
(4)
.284
(.064)***
Yes
Yes
.976
90
(5)
.270
(.065)***
Yes
No
.981
81
(6)
Note.—OLS estimates. City and decade fixed effects are included in all the regressions. Robust standard errors are reported in parentheses. Regressions are weighted by the size of the labor
force. Results from unweighted regressions are almost identical.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
Region#decade fixed effects
Includes California
R2
Observations
ln [ j (Immigrantsji 70/Immigrantsj 70 ) # LS Immigrationjt]
(1)
ln[(LS Immigrants ⫹ LS Natives)/
Labor Force]
Dependent Variable
ln(LS Immigrants/Labor Force)
TABLE 3
First Stage
low-skilled immigration and u.s. prices
397
1.
The unobserved factors determining that more immigrants decided
to locate in city i versus city i (both cities in the same region) in
1970 are not correlated with changes in the relative economic opportunities offered by the two cities in the following decades.11 The
identification assumption is not violated, for example, by Sun Belt
cities growing faster than cities in other regions (for several decades)
and at the same time being important immigrant cities.
2. The total (national) flow of low-skilled immigrants in a given decade
(second term in the interaction) is exogenous to differential shocks
to cities within a given region.
3. The only channel through which immigrant distribution in 1970
affects recent changes in prices is its effect on the actual distribution
of low-skilled immigrants across cities (exclusion restriction).
B.
Displacement Effects
I follow Card (2005) and use the following specification to test for
displacement effects:
LS Nativesirt ⫹ LS Immigrantsit
LS Immigrantsit
pd
⫹ fi ⫹ wrt ⫹ ␧ijt ,
Labor Forceit
Labor Forceit
(18)
where i is city, r region, and t decade, and fi and wrt represent city and
region#decade fixed effects, respectively.
If there are no displacement effects, d p 1, an increase of one percentage point in the share of low-skilled immigrants should lead to an
increase of the same magnitude in the share of low-skilled workers. If
11
A more formal discussion is presented here. Consider the following simplified version
of the model:
yirt p a ⫹ bIirt ⫹ prt ⫹ vi ⫹ eirt ,
where i represents a city, r a region, and t a decade; yirt is the dependent variable of interest
(e.g., prices); Iirt is a variable related to low-skilled immigration; prt are region#decade
fixed effects; and vi are city fixed effects. The equation above can be reexpressed as
(yirt ⫺ yirt⫺1) ⫺ (yi rt ⫺ yi rt⫺1) p b[(Iirt ⫺ Iirt⫺1) ⫺ (Ii rt ⫺ Ii rt⫺1)]
⫹ [(␧irt ⫺ ␧irt⫺1) ⫺ (␧i rt ⫺ ␧i rt⫺1)],
where i and i are located in the same region. If the instrument for Iirt is given by
li,1970 # TIt (li,1970 is the share of immigrants who were in city i in 1970 and TIt the national
flow of immigrants in year t), then the identification condition becomes
(␧irt ⫺ ␧irt⫺1) ⫺ (␧i rt ⫺ ␧i rt⫺1) is uncorrelated with (li,1970 ⫺ li ,1970)(TIt ⫺ TIt⫺1).
If, e.g., li,1970 ⫺ li ,1970 was influenced by a shock to the demand for low-skilled services in
city i in 1970 and the shock was permanent, it will cancel out by first differentiating the
error terms.
398
journal of political economy
TABLE 4
Displacement Effects of Low-Skilled Immigration
Dependent Variable: (LS Natives ⫹ LS Immigrants)/Labor Force
OLS
LS Immigrants/
Labor Force
Region#decade
fixed effects
Includes
California
Observations
IV
(1)
(2)
(3)
(4)
(5)
.992
(.197)***
.778
(.147)***
1.274
(.279)***
.712
(.246)***
.796
(.280)***
No
Yes
No
Yes
Yes
Yes
90
Yes
90
Yes
90
Yes
90
No
81
Note.—All regressions include city and decade fixed effects. Standard errors clustered at the city#decade level are
reported in parentheses. The instrument is j (Immigrantsji1970/Immigrantsj1970 ) # LS Immigrantsjt and its square. See
App. table B2 for the first stage. Regressions are weighted by the size of the labor force. Results from unweighted
regressions are almost identical.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
冘
every immigrant that comes to city i triggers a low-skilled native to move
to another city, then d p 0.
Given that the unobserved factors that determine the location of
immigrants are likely to influence also the location decisions of natives,
I use instrumental variables to estimate equation (18). Table 4 presents
the results. The magnitude of the estimated d (d̂ p 0.71) suggests that
there is some displacement taking effect but that it is far from one to
one: for every 10 immigrants that arrive to a city, approximately three
natives move out. Note, however, that one cannot reject the hypothesis
that the coefficient is equal to one.
The comparison of wage regressions estimated at the city level versus
the state level, presented in Section IV.D, provides stronger evidence
on the existence of displacement effects.
C.
Price Effects
A first approach to the study of the effect of low-skilled immigrants on
prices of nontraded goods is to look only at the prices of services that
use their labor intensively. The services included in the empirical analysis
of this section are those for which I have data both on prices and on
the composition of employment, and whose intensity in the use of lowskilled immigrant labor was at least 9 percent in 1980.12 These are laundry and dry cleaning, shoe repair, babysitting, housekeeping, barber
shops, and other household services (including gardening). Ideally, I
12
I use the 1980 data to avoid endogeneity of technology choices. I chose 9 percent as
the threshold because it is a standard deviation above the average across industries.
low-skilled immigration and u.s. prices
399
would have run a separate regression for each service and estimated a
separate effect of the immigration shock and separate city fixed effects.
Because I have so few observations (30 cities and three decades), I pool
all indexes in the same regression and restrict the city fixed effects and
the effect of immigration to be equal across all services.13 I then control
for industry fixed effects. The general estimating equation is
ln Piht p v ln
LS Workersit
⫹ fi ⫹ zh ⫹ wrt ⫹ ␧iht ,
Labor Forceit
(19)
where i is city, h industry, r region, and t decade. The terms fi, zh, and
wrt represent city, industry, and region#decade fixed effects, respectively.
The linear fixed effects in equation (19) control for time-invariant differences in price levels across cities and across industries and for regionspecific shocks.
Note that I use ln (LS Workersit/Labor Forceit) as the explanatory
variable. There are a couple of reasons for this choice. First, it is a
simplified version of the price equations’ main explanatory variable
derived from the model.14 Second, the magnitude of the elasticity of
prices with respect to immigration will depend on the share of immigrants in the low-skilled labor supply, which, relative to the share of lowskilled workers, varies significantly by city.15
Table 5 reports the ordinary least squares (OLS) and instrumental
variable (IV) estimation of equation (19). The OLS coefficient for the
immigration variable is negative and statistically significant. However, as
has been emphasized in the immigration literature, the magnitude of
the cross-sectional correlation between immigrant inflows and economic
outcomes is likely to be biased upward; immigrants choose their loca13
Results are robust to excluding one service at a time. See App. table C1 for this and
other robustness checks.
14
The term ln (LS Workersit/Labor Forceit ) assumes that r p 1 in the theoretical model.
In tables 5, 8, and 10, I replace r p 0.8 (j p 5 ) and r p 0.5 (j p 2). Results change very
little.
15
To derive the elasticity of variable y to low-skilled immigration, the chain rule is
used:
LS Workersit
d
d ln y
d ln y
Labor Forceit
p
#
d(ln LS Immigrants)
LS Workersit
d(ln LS Immigrants)
d
Labor Forceit
(
)(
)
dLS Natives
LS Immigrants
.
dLS Immigrants LS Immigrants ⫹ LS Natives
Note that the share of immigrants in the low-skilled labor supply varies significantly by
city. I use its value for each city from the census (and set dLS Natives/dLS
Immigrants p 0) to calculate the city-specific immigration effect of the low-skilled immigration flow of the period 1980–2000. I report the weighted average across cities of
these effects unless explicitly noted.
⯝v# 1⫹
/
.5 1/.5
(3)
(4)
(5)
No
No
Yes
.639
540
Yes
No
Yes
.641
540
Yes
Yes
Yes
.645
540
No
No
Yes
.627
540
Yes
No
Yes
.620
540
⫺.167
⫺.126 ⫺.247
⫺.479
⫺.638
(.072)** (.091) (.118)** (.135)*** (.207)***
(2)
OLS
Yes
No
Yes
.580
540
⫺.249
(.118)**
(6)
(7)
Yes
No
Yes
.617
540
⫺.615
(.205)***
IV
Yes
No
Yes
.606
540
⫺.578
(.210)***
(8)
Yes
No
No
.637
486
⫺.533
(.199)***
(9)
Note.—Services included in the regression are babysitting, housekeeping, gardening, dry cleaning, shoe repair, and barber shops. All regressions include city, industry, and decade fixed effects.
Standard errors clustered at the city#decade level are reported in parentheses. The number of observations is 540 (30 cities, six industries, and three time periods).
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
Other controls:
Region#decade fixed effects
ln(Employment)
Includes California
R2
Observations
.5
/
.8 1/.8
ln[(LS Natives ⫹ LS Immigrants )
Labor Force]
.8
ln[(LS Natives ⫹ LS Immigrants )
Labor Force]
ln(LS Immigrants/Labor Force)
Main explanatory variable:
ln[(LS Natives ⫹ LS Immigrants)/Labor
Force]
(1)
TABLE 5
The Effects of LS Immigration on Prices of Immigrant-Intensive Industries, 1986–2000
Dependent Variable: ln(Price Index)
low-skilled immigration and u.s. prices
401
tion, at least in part, on the basis of the economic opportunities that
cities offer.
The results are generally invariant to the inclusion of controls for
economic trends that potentially attract immigrant flows. For example,
the coefficient of column 2 of table 5, which includes region-specific
time trends, is very similar to the coefficient of column 1. The inclusion
of the log of the level of employment (col. 3) increases the magnitude
of the immigration concentration coefficient, supporting the hypothesis
that OLS coefficients are biased upward because immigrant flows are
correlated with unobserved economic conditions. This regression is,
however, potentially flawed since employment is likely to be endogenous.
Columns 4–9 of table 5 present the IV estimates of equation (19).
The results suggest that an increase of 10 percent in the share of lowskilled workers in the labor force of a city reduces the prices of services
that intensively use immigrant labor by 4.8–6.3 percent. Given that today,
on average, low-skilled immigrants represent 40 percent of the lowskilled labor supply of a city, a 10 percent increase in the number of
low-skilled immigrants will reduce prices of immigrant-intensive services
by close to 2 percent. For example, the low-skilled immigration shock
experienced by the United States in the period 1980–2000 should have
reduced the prices of these services, on average, by approximately 9
percent. When all the cities in California are excluded from the regression, the coefficient becomes smaller but is still highly statistically
significant.
I can use my model from the previous section to compare my econometric estimates with the theoretical predictions. Equation (11) from
the model suggests that v̂ ⯝ ⫺(1 ⫺ a), the Cobb-Douglas coefficient for
the low-skilled labor aggregate in the production function of the nontraded good. Given that table 2 suggests that around 30–40 percent of
all workers in the immigrant-intensive services are low skilled and that
capital is not heavily used in the production of these services, v̂ p
⫺0.64 from table 5 is reasonably consistent with the theoretical
predictions.
As a specification check, I also estimate equation (19) for five other
groups of goods and services: nontraded goods with higher than average
(but not highest) use of low-skilled immigrant labor, nontraded goods
with lower than average use of this input, and traded goods divided into
the three categories of “immigrant labor intensity.”16 Assuming no demand effects of low-skilled immigration, the effect on other nontraded
goods not included in table 5 should have a significantly smaller magnitude than on goods that use immigrant labor intensively. Also, by
16
See App. table A1 for a classification of traded and nontraded goods into these
categories.
402
journal of political economy
TABLE 6
Price Effects of Low-Skilled Immigration on Various Groups of
Goods and Services
Dependent Variable: ln(Price Index)
Coefficient: ln[(LS Natives ⫹ LS Immigrants)/
Labor Force]
Group of Goods and
Services
1. Nontraded, intensive in
low-skilled foreign labor
2. Nontraded, moderate use
of low-skilled foreign
labor
3. Nontraded, low use of
low-skilled foreign labor
4. Traded, intensive in lowskilled foreign labor
5. Traded, moderate use of
low-skilled foreign labor
6. Traded, low use of lowskilled foreign labor
OLS
⫺.126
(.091)
IV
Number of
Industries
⫺.638
(.207)***
6
⫺.170
(.171)
6
.128
(.055)**
.066
(.107)
22
.020
(.076)
.211
(.149)
16
.045
(.050)
.139
(.098)
14
⫺.052
(.096)
.068
(.153)
10
.051
(.094)
Note.—All regressions include city, industry, and region#decade fixed effects. Standard errors clustered at the
city#decade level are reported in parentheses.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
definition, low-skilled immigration at the local level should not have
any effect on the prices of traded goods. In reality, most goods, especially
when considered from the point of view of the consumer, are not purely
tradable; there is always a part of the price that reflects the retailer’s
handling costs, and these costs are likely to be affected by local relative
endowments. However, for most goods, these handling costs represent
only a small share of the final price. For example, Barsky et al. (2003)
estimate an upper bound for the retailing costs of grocery goods, such
as cookies and soft drinks, of approximately 15 percent. Also, assuming
that the effect of retailing costs on prices of goods is not systematically
related to the percentage of low-skilled employees producing the good,
it appears that even if the traded goods are affected by local endowments, that effect should not be especially large for the group of goods
that use immigrant labor intensively in their production process.
The results presented in table 6 are in line with the theoretical predictions and suggest that in table 5 I am actually capturing the causal
effect of immigration on prices through changes in labor supplies. The
low-skilled immigration and u.s. prices
403
coefficient on low-skilled immigrant concentration on the prices of nontraded goods that use immigrant labor moderately is negative but has
a smaller magnitude and is not statistically significant. The coefficients
in the equations for the prices of traded goods are positive but not
statistically distinguishable from zero. The estimates for these groups of
goods also suggest that other mechanisms through which low-skilled
immigrants might affect the prices of goods and services (e.g., changes
in demand) are not very large, or at least are not captured by the
empirical analysis of the present paper.
An alternative specification for nontraded goods that should yield
results similar to those of tables 5 and 6 and directly tests for the labor
supply mechanism is the following:
ln Piht p v ln
LS Workersit
LS Immigrantsh
⫹b#
Labor Forceit
Workersh
# ln
LS Workersit
⫹ fit ⫹ lht ⫹ ␧iht ,
Labor Forceit
(20)
in which I add an interaction of the immigrant concentration in the
city with the share of total workers in the industry that were low-skilled
immigrants (which proxies for the production technology of the good).
The intuition is simple: a low-skilled immigrant shock to a given city
should mostly affect the prices of goods and services that use their labor
intensively (b ! 0). An advantage of this specification is that city#decade
fixed effects can be included, making the identification assumption
much harder to violate.
I estimate equation (20) for the sample of all nontraded goods and
services. The direction of the results (presented in table 7) is in line
with tables 5 and 6. The direct effect of low-skilled immigrant concentration on the prices of nontraded goods is slightly negative but not
statistically significant, and the interaction term is negative and marginally significant. The magnitudes, however, suggest smaller effects.
Discrepancies might come from the additional restrictions imposed in
equation (20), for example, that the city fixed effects do not vary by
industry group.
When equation (20) is estimated for traded goods, both v and b are
small and not significant statistically (see cols. 4 and 5 in table 7).
D.
Wage Effects
The estimated coefficients from the previous subsection demonstrate
that low-skilled immigration has an effect on the prices of low-skilled
immigrant-intensive services; this subsection explores the channels
through which the effect is likely to take place. Given the nature of the
404
journal of political economy
TABLE 7
Production Technology and the Effects of Low-Skilled Immigration on Prices
Dependent Variable: ln(Price Index)
All Nontraded
Main Explanatory
Variables
OLS
(1)
IV
(2)
ln(LS Workers/Labor Force)
.075
(.048)
⫺.063
(.108)
⫺.144
(.180)
No
3,058
⫺.836
(.495)*
No
3,058
% LS Immigrants in Production#Ln(LS Workers/
Labor Force)
City#decade fixed effects
Observations
All Traded
IV
(3)
IV
(4)
IV
(5)
.173
(.088)*
⫺.836
(.500)*
Yes
3,058
⫺.395
(.414)
No
3,597
⫺.394
(.417)
Yes
3,597
Note.—All regressions include city, industry#decade, and region#decade fixed effects. Standard errors clustered
at the city#decade level are reported in parentheses.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
services included in the regressions of table 5 (very labor-intensive production processes, particularly of the unskilled type), the obvious channel through which the prices go down after a low-skilled immigration
shock is a reduction in wages. However, as mentioned in the introduction, all the studies that have used cross-city variation for the identification of wage effects have found negligible effects of immigrant groups
on the wages of competing natives, including the very low skilled. The
existing evidence in the literature on the wage effects of new waves of
immigrants on older immigrants does not help in clarifying the matter;
it is scarce and inconclusive.17
The model from Section III predicts that the strong evidence of no
wage effects on natives invalidates the wage mechanism for the reduction
in prices if low-skilled natives and low-skilled immigrants are perfect
substitutes in production. There are good reasons, however, to think
that low-skilled immigrants and natives are far from being perfect substitutes. For example, Census data for 2000 show that more than 80
percent of the low-skilled immigrants who arrive each year in the United
States speak very little English and that, even after 10 years in the country, still 60 percent report not speaking English or not speaking it well.
English skills matter for occupational choice: Cortes (2006) finds a
strong negative correlation between the language skills intensity of a
17
The three papers that study the impact of immigration on earlier immigrants are
Card (1990), LaLonde and Topel (1991), and Borjas (2003). Card finds no significant
negative effect of an immigration shock on the labor outcomes of earlier immigrants.
LaLonde and Topel find that higher immigration modestly lowers the wages of more
recent immigrant cohorts, and Borjas finds a sizable, but not statistically significant, negative effect.
low-skilled immigration and u.s. prices
405
low-wage occupation and the ratio of immigrants to natives who work
in that occupation. Also, conditional on not having a high school diploma, the average years of education of immigrants is significantly lower
than that of natives, and the quality of each year of education is potentially higher in the United States. The legal status of most low-skilled
foreigners working in the United States is the final reason for which
natives and immigrants might not be perfect substitutes.
To empirically test my hypothesis that lower wages drive the reduction
in prices, but that the wages of competing immigrants are affected significantly more than those of natives because natives and immigrants
are far from perfect substitutes, I run wage regressions for different
groups of the population.
I use the following econometric specification:
ln wnit p v e ln
LS Workersit
⫹ X n Le ⫹ Wit Se ⫹ fie ⫹ wrte ⫹ ␧eijt,
Labor Forceit
(21)
where e is the group of the population I am analyzing, n a low-skilled
worker, i a city, r a region, and t a decade. The term X n represents
individual-level characteristics, namely, age, age squared, sex, a black
dummy, and a Hispanic dummy. The term Wit represents city time-varying variables, specifically the percentage of males in the low-skilled labor
force. The linear fixed effects in equation (21) control for time-invariant
differences in labor outcomes across cities and for region-specific shocks.
Wage data for the estimation of equation (21) come from the 1980,
1990, and 2000 Censuses. The sample is restricted to workers who reported a positive annual labor income, a positive number of total weeks
worked last year, and a positive number of “usual hours worked per
week.” Top-coded incomes were multiplied by 1.5, and hourly wages
were adjusted for inflation. Workers with a wage per hour of less than
$2.00 were excluded from the sample.
The estimated v e ’s are reported in table 8. Panel A in the table estimates equation (21) for all native high school dropouts. In line with
the previous cross-city literature, under most specifications, I find a
negative but small and insignificant effect of low-skilled immigration on
wages. I find a larger and marginally significant effect on the wages of
female native workers (col. 8).
I then estimate the wage effect of recent flows of immigrants on the
wages of low-skilled immigrant workers who have been in the United
States for at least 10 years. I find significant negative effects when individuals of both sexes are included in the specification (table 8, panel
B). Columns 7 and 8 of panel B show that the result is driven by the
female sample.
I present two pieces of evidence that suggest that the male coefficient’s
⫺.085
(.041)***
No
Yes
.134
117,049
.023
(.054)
No
Yes
.238
509,975
⫺.116
(.050)**
No
Yes
.134
117,049
⫺.044
(.073)
No
Yes
.238
509,975
IV
rp1
(2)
IV
r p .8
(5)
⫺.114
(.069)
Yes
No
.242
456,739
⫺.048
(.069)
Yes
Yes
.239
509,975
⫺.047
(.068)
Yes
Yes
.239
509,975
IV
r p .5
(6)
⫺.123
(.059)**
Yes
Yes
.135
117,049
⫺.043
(.077)
Yes
No
.141
71,913
⫺.133
(.064)**
Yes
Yes
.135
117,049
⫺.160
(.076)**
Yes
Yes
.135
117,049
B. Low-Skilled Previous Immigrants
⫺.050
(.071)
Yes
Yes
.239
509,975
IV
rp1
(4)
A. Low-Skilled Natives
All
IV
rp1
(3)
.042
(.069)
Yes
Yes
.124
71,477
⫺.045
(.082)
Yes
Yes
.280
306,569
IV
rp1
(7)
Male
⫺.292
(.059)***
Yes
Yes
.055
45,572
⫺.099
(.058)*
Yes
Yes
.097
203,406
IV
rp1
(8)
Female
Source.—1980, 1990, and 2000 Census.
Note.—City and decade fixed effects are included in all specifications. Controls at the individual level are age, age squared, gender, black dummy, and Hispanic dummy. Other controls are %
males in the native group studied. Regressions for immigrants include dummies for Hispanic group (Mexican, Cuban, etc.) and the percentage of each Hispanic group in the immigration population
of the city. Standard errors clustered at the city#decade level are reported in parentheses. All regressions are weighted using the census person weight.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
Region#decade fixed effects
Includes California
R2
Observations
ln(LS Nativesr ⫹ LS Immigrantsr)1/r
Region#decade fixed effects
Includes California
R2
Observations
r 1/r
ln(LS Natives ⫹ LS Immigrants )
r
OLS
rp1
(1)
TABLE 8
The Effects of Low-Skilled Immigration on the Hourly Wages of Low-Skilled Workers
Dependent Variable: ln(Hourly Wage)
low-skilled immigration and u.s. prices
407
TABLE 9
Labor Force Participation Statistics from the Census and Other Sources
1980
1990
2000
Current
Population
Survey
1999–2001
.815
(.388)
.814
(.389)
.675
(.468)
.830
(.375)
.827
(.378)
.492
(.500)
.506
(.500)
.431
(.495)
.482
(.500)
.483
(.500)
.678
(.467)
.604
(.489)
.517
(.500)
.552
(.497)
.484
(.500)
.683
(.465)
.618
(.486)
.506
(.500)
.574
(.495)
.513
(.500)
.862
(.345)
.858
(.349)
.741
(.438)
.899
(.301)
.862
(.345)
.827
(.378)
.866
(.340)
.785
(.411)
.833
(.301)
.838
(.368)
.908
(.289)
.907
(.290)
.864
(.343)
.900
(.299)
.895
(.307)
Census
Low-skilled male
immigrants
Low-skilled female
immigrants
Low-skilled male
natives
Low-skilled male Hispanic natives
High-skilled male
immigrants
Some college male
immigrants
College graduate male
immigrants
American
Community
Survey
2005–6
magnitude and significance are due to data issues and do not represent
a real effect. First, I estimate equation (21) for the sample of low-skilled
male immigrants separately for the two decades included in the analysis
and find that the problematic result comes from the 1990–2000 regression. The wage effect for this period is positive and highly significant.18 Simple descriptive statistics of labor force participation reported
by decade, shown in table 9, suggest that there is a large degree of
misreporting in the 2000 Census, especially from low-skilled male immigrants.19 Second, I estimate equation (21) for the native groups who
are similar to low-skilled immigrants in terms of race and English proficiency and therefore are more likely to compete directly with immigrants in the labor market. My assumption is that the effect of lowskilled immigration on the wages of these groups should provide a lower
18
The coefficient for 1980–90 is ⫺0.146 (0.07) and for 1990–2000 is 0.714 (0.149). The
coefficient for the sample of low-skilled native Hispanic males for the period 1990–2000
is ⫺0.474 (0.280).
19
A likely explanation for the underreporting of labor force participation is that the
2000 Census included many more illegal immigrants than the previous censuses, and
undocumented immigrants might lie about their labor force status. The price effects
estimates are robust to using as the explanatory variable the total number of low-skilled
immigrants, and not just those who reported being in the labor force. See App. table C1.
408
journal of political economy
bound for the wage effects on other low-skilled immigrants. As observed
in table 10, panel A, the wage effects for native Hispanics, both female
and male, are negative and statistically significant and have a magnitude
similar to those for the impact on female immigrants. Moreover, the
coefficient is significantly larger, though not very precisely estimated
(the number of observations drops significantly), for low-skilled Hispanic natives who reported that they do not speak English very well.
Taking the estimated coefficients for these two groups to proxy for the
effect on immigrants already in the country implies that, at current
immigration levels, a 10 percent increase in the share of low-skilled
immigrants in the labor force of a city will reduce the wages of lowskilled immigrants already present in the United States by between 2
and 4 percent. Notice that this number is reasonably consistent with
the price effect of immigration on immigrant-intensive services.
As mentioned in the introduction, the estimation of small and insignificant effects of low-skilled immigration on native wages using crosscity variation does not imply that immigrants have no effect on the labor
outcomes of natives. If natives respond to immigration by moving to
other cities, the full effect of immigration will not be captured by the
local influx coefficient, which will provide only a lower bound (Borjas
2006). To check whether the wage effects at the local level are being
attenuated by the native migration response, I estimate equation (21)
at the state level and summarize the results in table 11. Columns 1 and
2 report the estimation when the sample is restricted to individuals living
in the 30 cities included in the price regressions; columns 4 and 5
include all cities. Two observations are worth mentioning. First, wage
effects at the state level are always larger than at the city level for all
groups of low-skilled workers, particularly for the native population. This
result suggests that workers do move as a response to immigration shocks
and that immigrants do have a negative effect on the wage of the typical
low-skilled native. It also suggests that the median native’s migration
decision is more responsive to city wage differences than that of native
Hispanics and immigrants. Second, even at the state level, the wage
effects of immigrants are much larger for previous immigrants and native Hispanics than for the average native, suggesting again that the two
labor inputs are not perfect substitutes.
The state-level estimates will still not capture the full wage effects of
immigration if workers move across states. Ideally, wage effects should
be estimated at the national level; unfortunately, given my empirical
strategy, this will imply not including year fixed effects and not using
an instrument to solve for the omitted variable bias. Exploiting variation
in immigrant shocks within education#experience cells, Borjas (2006)
⫺.035
(.089)
No
Yes
.121
17,647
IV
rp1
(4)
IV
r p .8
(5)
IV
r p .5
(6)
⫺.301
(.143)**
Yes
Yes
.147
58,124
⫺.328
(.156)**
Yes
No
.137
42,348
⫺.298
(.147)**
Yes
Yes
.147
58,124
⫺.295
(.174)*
Yes
Yes
.147
58,124
A. Low-Skilled Native Hispanics
IV
rp1
(3)
⫺.271
(.151)*
Yes
Yes
.170
35,978
IV
rp1
(7)
⫺.433
(.177)**
No
Yes
.119
17,647
⫺.637
(.370)*
Yes
Yes
.121
17,647
⫺.728
(.469)
Yes
No
.116
14,237
⫺.545
(.304)*
Yes
Yes
.121
17,647
⫺.435
(.247)*
Yes
Yes
.122
17,647
⫺.578
(.336)*
Yes
Yes
.115
11,439
B. Low-Skilled Native Hispanics with Low English Proficiency
⫺.297
(.123)**
No
Yes
.145
58,124
IV
rp1
(2)
Male
⫺.739
(.559)
Yes
Yes
.043
6,208
⫺.331
(.178)*
Yes
Yes
.058
22,146
IV
rp1
(8)
Female
Source.—1980, 1990, and 2000 Census.
Note.—City and decade fixed effects are included in all specifications. Controls at the individual level are age, age squared, gender, black dummy, and Hispanic dummy. Other controls are
% males in the native group studied. Regressions for Hispanics include dummies for Hispanic group (Mexican, Cuban, etc.) and the percentage of each Hispanic group in the native Hispanic
population of the city. Standard errors clustered at the city#decade level are reported in parentheses. All regressions are weighted using the census person weight.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
Region#decade fixed effects
Includes California
R2
Observations
r 1/r
ln(LS Natives ⫹ LS Immigrants )
r
Region#decade fixed effects
Includes California
R2
Observations
⫺.013
(.090)
No
Yes
.146
58,124
ln(LS Nativesr ⫹ LS Immigrantsr)1/r
OLS
rp1
(1)
All
TABLE 10
Effects of Low-Skilled Immigration on the Hourly Wages of Low-Skilled Native Hispanics
Dependent Variable: ln(Hourly Wage)
⫺.165
(.073)**
⫺.338
(.150)**
⫺.206
(.101)*
⫺.469
(.094)***
20
⫺.050
(.071)
⫺.301
(.143)**
⫺.123
(.059)**
⫺.292
(.059)***
30
45,572
117,049
58,124
509,975
Observations
⫺.013
(.058)
⫺.221
(.093)**
⫺.209
(.057)***
⫺.292
(.068)***
123
City
Level
⫺.240
(.088)***
⫺.376
(.141)***
⫺.359
(.097)***
⫺.469
(.083)***
37
State
Level
65,728
169,646
101,726
906,780
Observations
All Cities Sample
Source.—1980, 1990, and 2000 Census.
Note.—City/state and region#decade fixed effects are included in all specifications. Controls at the individual level are age, age squared, gender, black dummy, and Hispanic dummy. Other
controls are % males in the native/immigrant group studied. Regressions for Hispanics/immigrants include dummies for Hispanic group (Mexican, Cuban, etc.) and the percentage of each
Hispanic group in the Hispanic/immigrant population of the city/state. Standard errors clustered at the city/state#decade level are reported in parentheses. All regressions are weighted using
the census person weight.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
Number of cities/states
Low-skilled previous immigrants (females)
Low-skilled previous immigrants
Low-skilled native Hispanics
Low-skilled natives
State
Level
City
Level
30-City Sample
TABLE 11
Effects of Low-Skilled Immigration on Wages of Low-Skilled Workers: City vs. State Levels
Dependent Variable: ln(Hourly Wage)
low-skilled immigration and u.s. prices
411
estimates that native migration attenuates the measured impact of immigration on wages in the state labor market by 40 percent.20
It is important to emphasize, however, that the primary purpose of
estimating the wage equations at the city level was to test if indeed wages
are the main channel through which prices of nontraded goods are
affected by local shocks to immigration. The fact that I find wage effects
at the city level for previous immigrants and native Hispanics and that
their magnitudes are reasonably consistent with the price effects is key
for my story. The wage estimates at the city level, even if not capturing
the full effects, also imply that immigrants and natives cannot be perfect
substitutes. If they were, wages should be affected equally at the local
level, regardless of different propensities to migrate.
Back-of-the-envelope calculations using equations (15) and (16) and
my estimates for the local wage effects for low-skilled immigrants and
natives suggest a lower bound for the elasticity of substitution between
the two labor inputs of approximately 4. My estimate is consistent with
the findings of Ottaviano and Peri (2006), who estimate an elasticity of
substitution between 5 and 10.
V.
Purchasing Power Calculations
The previous section demonstrates how low-skilled immigration affects
prices of nontraded goods and wages at the city level. This evidence
alone is not sufficient to determine how natives’ purchasing power is
affected; data on native preferences are needed. This section combines
data on consumption patterns from the CEX with the price and wage
effects obtained in Section IV to estimate how natives’ purchasing power
was changed by the immigration wave of the years 1980–2000.
Natives of all skill levels benefit from low-skilled immigration through
the reduction in the nontraded goods component of the cost of living.
I use the expenditure shares from the 2000 CEX to calculate changes
in the Paasche index caused by the immigration influx of the past two
decades and interpret these changes as the price benefits from immigration. Column 1 of table 12 shows the share of total expenditures
devoted to household services by different education groups. As expected, the share increases with education levels and is almost a third
higher for very educated natives compared with high school dropouts.
Even for the very high skilled, however, the share is very small (4.3
percent), such that the overall effect on purchasing power is not large:
20
I cannot use the variation in Borjas (2006) to estimate national-level effects because
for low-skilled workers, people of different ages are very substitutable. When I control for
decade#city fixed effects, the coefficient of the immigrant shock defined at the education#city#age level in eq. (21) is small and not statistically significant. This is true for
all low-skilled groups. Results are available on request.
3.2
3.2
3.4
3.7
3.8
4.3
Education of Household Head
Less than high school:
Median low-skilled native
Low-skilled native Hispanic
High school graduate
Some college
College graduate
Graduate education
Note.—Coefficients from tables 5, 8, and 10 were used for these calculations.
Expenditure
Share of
ImmigrantIntensive
Services
(1)
⫺9.2
⫺9.2
⫺9.2
⫺9.2
⫺9.2
⫺9.2
Simulated
Change in
Price of
ImmigrantIntensive
Services
(2)
⫺.29
⫺.29
⫺.31
⫺.33
⫺.35
⫺.40
Simulated
Change in
Price Index
1980–2000
(3)
⫺.72, ⫺1.43
⫺4.32
.00
.00
.00
.00
Simulated
Change
in Wages
1980–2000
(4)
⫺.43, ⫺1.14
⫺4.03
.31
.33
.35
.40
Simulated
Change in
Purchasing
Power
1980–2000
(5)
TABLE 12
Simulation Exercise: Price and Wage Effects of the Low-Skilled Immigration Influx of the Period 1980–2000 (%)
low-skilled immigration and u.s. prices
413
the immigration shock of the 1980s and 1990s decreased the cost of
the consumption basket between 0.29 percent for native high school
dropouts and 0.40 percent for natives with graduate education.
Lower prices come at a cost. As discussed in the previous section, the
wages of low-skilled natives of Hispanic origin are reduced by the inflows
of foreign-born high school dropouts to a city. For the average lowskilled native, however, the local effect is small. If one takes the largest
estimate (⫺0.099 for the female sample), the magnitude suggests that
the low-skilled immigration wave of the past two decades decreased lowskilled native wages by a maximum of 1.43 percent. This number increases to 4.2 percent for low-skilled natives of Hispanic origin and with
low English proficiency.
I combine the cost-of-living effects and the wage effects to calculate
the net impact of low-skilled immigration on natives’ purchasing power
by skill group. As observed in table 12, low-skilled natives’ purchasing
power was reduced by between 0.43 percent and 1.14 percent for the
typical native and close to 4 percent for natives of Hispanic origin.
However, high school graduates, workers with some college, and college
graduates benefited: their purchasing power increased between 0.31
percent and 0.4 percent. Given that low-skilled natives represent a small
fraction of all native workers, the average net benefit for the native
population was positive.21
VI.
Conclusion
A large body of literature analyzes the impact of immigration on the
employment opportunities of native workers and the costs it imposes
on taxpayers. With the exception of Borjas (1995), this literature has
not addressed the gains that immigration brings to the native population. The study of the benefits from immigration is important because
the contrast between benefits and costs (not only economic, of course)
informs decisions about immigration policy. This paper contributes to
the immigration literature by estimating the impact of low-skilled immigration on prices, wages, and the purchasing power of natives.
I find that low-skilled immigration benefits the native population by
decreasing the nontraded goods component of the cost of living. At
current U.S. immigration levels, a 10 percent increase in the average
city’s share of low-skilled immigrants in the labor force decreases the
price of immigrant-intensive services such as housekeeping and gar21
These numbers are likely to underestimate the benefits from low-skilled immigration
because they do not take into account the complementarities between low-skilled labor
and capital and between low- and high-skilled labor. However, the estimates abstract from
any effects from the demand side. See Saiz (2003, 2007) for evidence on the effects of
immigration on rents.
414
journal of political economy
dening by 2 percent. A simple theoretical model and my wage estimates
suggest that wages are a likely channel through which these effects take
place. However, wage effects are significantly larger for low-skilled immigrants than for the median low-skilled native, implying that the two
are imperfect substitutes. My results imply that the low-skilled immigration wave of the period 1980–2000 increased the purchasing power
of high-skilled workers living in the 30 largest cities by an average of
0.32 percent and decreased the purchasing power of the typical native
high school dropouts by a maximum of 1 percent and of Hispanic lowskilled natives by 4.2 percent. I conclude that, through lower prices,
low-skilled immigration brings positive net benefits to the U.S. economy
as a whole but generates a redistribution of wealth.
Because of the focus on city-level outcomes, this paper has looked at
prices of only nontraded goods and services. Low-skilled immigration
is also likely to have effects on the prices of traded goods, but these will
occur at an aggregate, national level. A theoretical and empirical exploration of this issue is needed in order to have a complete assessment
of the effects of low-skilled immigration.
Utilities:
Electricity
Utility natural gas services
Telephone services, local charges
Water and sewage maintenance
Cable TV
Garbage and trash collection
Medical services:
Hospital and other medical care services
Physicians’ services
Dental services
Eyeglasses and eye care
Services by other medical professionals
Education:
College tuition and fees
Elementary and high school tuition
Child day care
Other tuition and fees
Household services:
Food away from home
Babysitting
Domestic service
Nontraded Goods and Services
Appendix A
2
1
1
3
3
3
3
3
3
3
3
3
3
3
3
3
3
2
Group
Traded Goods and Services
Food:
Cereals
Bakery products
Beef and veal
Pork
Other meats
Fish and seafood
Fresh milk and cream
Processed dairy products
Fresh fruits
Fresh vegetables
Processed fruits
Processed vegetables
Sugar and sweets
Carbonated drinks
Apparel and textiles:
Apparel
Footwear
Textile house furnishing
Gadgets:
Household appliances
TV and sound equipment
TABLE A1
Classification of Goods and Services
2
2
1
1
1
2
1
2
2
1
1
2
2
1
1
1
1
1
3
Group
416
2
1
3
3
2
2
3
3
3
3
3
1
2
3
1
1
Group
Traded Goods and Services
Toys, hobbies, etc.
Photographic supplies and equipment
Watches
Sporting goods and equipment
Information technology, hardware and services
Supplies:
Maintenance and repair commerce
Toilet goods and personal care
Laundry and cleaning products
Household paper products
Other:
Intrastate telephone services
Airline fares
Other intercity transportation
Fuel oil
Furniture and bedding
Tires
New vehicles
Prescription drugs and medical supplies
Nonprescription drugs and medical supplies
Reading materials
School books and supplies
Tobacco products
3
3
3
3
1
1
2
3
3
3
3
2
2
2
2
2
1
2
1
1
3
Group
Note.—Group p 1 if low-skilled immigrant share in industry’s employment is greater than 9 percent in 1980; group p 2 if low-skilled immigrant share in industry’s employment is between 4 and
9 percent in 1980; Group p 3 if low-skilled immigrant share in industry’s employment is less than 4 percent in 1980.
Other household services (including gardening)
Appliance and furniture repair
Care of invalids, elderly at home
Other apparel services (including shoe repair)
Laundry and dry cleaning
Other:
Beauty parlors
Barber shops
Tenants’ insurance
Automobile insurance
Automotive repair
Automotive maintenance and service
Intracity transportation
Admissions (movies etc.)
Legal fees
Cemetery lots and funeral expenses
Financial services
Nontraded Goods and Services
TABLE A1
(Continued )
417
j
(1)
(2)
.405
(.285)
Yes
Yes
.972
90
(3)
.659
(.287)**
Yes
No
.970
81
Note.—OLS estimates. City and decade fixed effects are included in all the regressions. Robust standard errors are reported in parentheses.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
.566
(.240)**
No
Yes
.967
90
.366
(.054)***
No
Yes
.976
90
(4)
.280
(.063)***
Yes
Yes
.980
90
(5)
.299
(.070)***
Yes
No
.979
81
(6)
ln[(LS Immigrants ⫹ LS Natives)/
Labor Force]
Dependent Variable
ln(LS Immigrants/Labor Force)
TABLE B1
First Stage: 1986, 1990, and 2000
ji1970 /Immigrantsj1970 ) # LS Immigrantsjt ]
Region#decade fixed effects
Includes California
R2
Observations
ln [
冘 (Immigrants
Appendix B
TABLE B2
First Stage: Alternative Functional Form
Dependent Variable: LS Immigrants/Labor Force
冘 (Immigrants
(1)
(2)
(3)
.297
(.075)***
.217
(.073)**
.831
(.218)***
⫺.279
(.075)***
No
Yes
.967
90
⫺.211
(.073)***
Yes
Yes
.976
90
-2.168
(.627)***
Yes
No
.971
81
ji1970 /Immigrantsj1970 ) # LS
j
Immigrantsjt
冘
[ j (Immigrantsji1970 /Immigrantsj1970 ) # LS
Immigrantsjt]2
Region#decade fixed effects
Includes California
R2
Observations
Note.—OLS estimates. Included years are 1980, 1990, and 2000. City and decade fixed effects are included
in all the regressions. Robust standard errors are reported in parentheses.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
TABLE B3
Countries Included in the Construction of the Instrument
Albania
Argentina
Australia
Austria
Azores
Belgium
Belize
Bolivia
Brazil
Bulgaria
Canada
Chile
China
Colombia
Costa Rica
Cuba
Czechoslovakia
Denmark
Dominican Republic
Ecuador
Egypt
El Salvador
England
Finland
France
Germany
Greece
Guatemala
Haiti
Honduras
Hungary
India
Iran
Ireland
Israel
Italy
Jamaica
Japan
Jordan
Korea
Latvia
Lebanon
Lithuania
Mexico
Netherlands
New Zealand
Nicaragua
Northern Ireland
Norway
Pakistan
Panama
Paraguay
Peru
Philippines
418
Poland
Portugal
Romania
Russia
Scotland
South Africa
Spain
Sweden
Switzerland
Syria
Trinidad and Tobago
Turkey
Uruguay
Venezuela
Vietnam
Wales
Yugoslavia
419
Observations
Excludes
Main explanatory variable: ln[(LS Natives ⫹ LS Immigrants)/Labor Force]
Observations
Excludes
⫺.618
(.241)**
Top 3 cities’
share of
Mexicans
in labor
force
486
(1)
450
⫺.628
(.213)***
Gardening
and other
services
450
(3)
(5)
⫺.640
(.274)**
Top 4 cities’
share of
Mexicans
in labor
force
468
(2)
486
⫺.576
(.267)**
LA, NY,
and Miami
(3)
(6)
450
450
450
⫺.679
⫺.673
⫺.535
(.230)***
(.235)***
(.174)***
Shoe repair Laundry and Barber shops
dry cleaning
(4)
B. Specifications Excluding Cities with Many Immigrants
450
⫺.602
(.206)***
Housekeeping
(2)
A. Specifications Excluding Immigrant-Intensive Services One by One
⫺.711
(.245)***
Babysitting
(1)
TABLE C1
Robustness Checks to Instrumental Variables Price Regressions
Dependent Variable: ln(Price Index)
Main explanatory variable: ln[(LS Natives ⫹ LS Immigrants)/Labor Force]
Appendix C
360
⫺.567
(.161)***
1980
(1)
540
⫺.501
(.179)***
Uses 1980
immigrant
variables and
1986 prices
(2)
⫺.427
(.162)***
Replaces LS
immigrants
in labor force
with all LS
immigrants
540
(3)
Note.—Services included in the regressions are babysitting, housekeeping, gardening, dry cleaning, shoe repair, and barber shops. All regressions include city, industry, and region#decade fixed
effects. Standard errors clustered at the city#decade level are reported in parentheses. The four cities with the largest share of Mexicans in the labor force are Los Angeles, San Diego, Dallas, and
Houston. All the specifications are estimated using IV.
* Significant at the 10 percent level.
** Significant at the 5 percent level.
*** Significant at the 1 percent level.
Observations
Excludes
Main explanatory variable: ln[(LS Natives ⫹ LS Immigrants)/Labor Force]
C. Specifications That Use Different Time Periods/Variables
TABLE C1 (Continued)
low-skilled immigration and u.s. prices
421
References
Altonji, Joseph G., and David Card. 1991. “The Effects of Immigration on the
Labor Market Outcomes of Less-Skilled Natives.” In Immigration, Trade, and
the Labor Market, edited by John M. Abowd and Richard B. Freeman. Chicago:
Univ. Chicago Press (for NBER).
Barsky, Robert, Mark Bergen, Shantanu Dutta, and Daniel Levy. 2003. “What
Can the Price Gap between Branded and Private Label Products Tell Us about
Markups?” In Scanner Data and Price Indexes, edited by Robert Feenstra and
Matthew Shapiro. Chicago: Univ. Chicago Press (for NBER).
Borjas, George J. 1995. “The Economic Benefits from Immigration.” J. Econ.
Perspectives 9 (Spring): 3–22.
———. 2003. “The Labor Demand Curve Is Downward Sloping: Reexamining
the Impact of Immigration on the Labor Market.” Q.J.E. 118 (November):
1335–74.
———. 2006. “Native Internal Migration and the Labor Market Impact of Immigration.” J. Human Resources 41 (Spring): 221–58.
Borjas, George J., Richard B. Freeman, and Lawrence Katz. 1996. “Searching
for the Effect of Immigration on the Labor Market.” A.E.R. Papers and Proc.
86 (May): 246–51.
Card, David. 1990. “The Impact of the Mariel Boatlift on the Miami Labor
Market.” Indus. and Labor Relations Rev. 43 (October): 245–57.
———. 2001. “Immigrant Inflows, Native Outflows, and the Local Labor Market
Impacts of Higher Immigration.” J. Labor Econ. 19 (January): 22–64.
———. 2005. “Is the New Immigration Really So Bad?” Econ. J. 115 (November):
F300–F323.
Card, David, and John E. DiNardo. 2000. “Do Immigrant Inflows Lead to Native
Outflows?” A.E.R. Papers and Proc. 90 (May): 360–67.
Card, David, and Ethan G. Lewis. 2007. “The Diffusion of Mexican Immigrants
during the 1990s: Explanations and Impacts.” In Mexican Immigration, edited
by George J. Borjas. Chicago: Univ. Chicago Press.
Cortes, Patricia. 2006. “Language Skills and the Effect of Low-Skilled Immigration on the US Labor Market.” PhD diss., Massachusetts Inst. Tech.
Federman, May N., David E. Harrington, and Kathy J. Krynski. 2006. “Vietnamese
Manicurists: Displacing Natives or Finding New Nails to Polish?” Indus. and
Labor Relations Rev. 59 (January): 302–18.
Khananusapkul, Phanwadee. 2004. “Do Low-Skilled Female Immigrants Increase
the Labor Supply of Skilled Native Women?” PhD diss., Harvard Univ.
Lach, Saul. 2007. “Immigration and Prices.” J.P.E. 115 (August): 548–87.
LaLonde, Robert, and Robert H. Topel. 1991 “Labor Market Adjustments to
Increased Immigration.” In Immigration, Trade, and the Labor Market, edited by
John M. Abowd and Richard B. Freeman. Chicago: Univ. Chicago Press (for
NBER).
Leamer, Edward E. 1995. “The Heckscher-Ohlin Model in Theory and Practice.”
Princeton Studies in International Finance, Working Paper no. 77 (February),
Princeton Univ.
Lewis, Ethan G. 2005. “Immigration, Skill Mix, and the Choice of Technique.”
Working Paper no. 05-8 (May), Fed. Reserve Bank Philadelphia.
Munshi, Kaivan. 2003. “Networks in the Modern Economy: Mexican Migrants
in the U.S. Labor Market.” Q.J.E. 118 (May): 549–99.
Ottaviano, Gianmarco, and Giovanni Peri. 2006. “Rethinking the Effects of Immigration on Wages.” Working Paper no. 12497 (September), NBER, Cambridge, MA.
422
journal of political economy
Saiz, Albert 2003. “Room in the Kitchen for the Melting Pot.” Rev. Econ. and
Statis. 85 (August): 502–21.
———. 2007. “Immigration and Housing Rents in American Cities.” J. Urban
Econ. 61 (March): 345–71.