Nonparametric bounds on the returns to language skills
ABSTRACT This paper applies the theoretical literature on nonparametric bounds on treatment effects to the estimation of how limited English proficiency (LEP) affects wages and employment opportunities for Hispanic workers in the United States. I analyse the identifying power of several weak assumptions on treatment response and selection, and stress the interactions between LEP and education, occupation and immigration status. I show that the combination of two weak but credible assumptions provides informative upper bounds on the returns to language skills for certain subgroups of the population. Adding age at arrival as a monotone instrumental variable also provides informative lower bounds. Copyright © 2005 John Wiley & Sons, Ltd.
IZA DP No. 1098
Nonparametric Bounds on
the Returns to Language Skills
D I S C U S S I O N P A P E R S E R I E S
zur Zukunft der Arbeit
Institute for the Study
Nonparametric Bounds on
the Returns to Language Skills
Universitat Pompeu Fabra,
CREA and IZA Bonn
Discussion Paper No. 1098
P.O. Box 7240
This paper can be downloaded without charge at:
An index to IZA Discussion Papers is located at:
Any opinions expressed here are those of the author(s) and not those of the institute. Research
disseminated by IZA may include views on policy, but the institute itself takes no institutional policy
The Institute for the Study of Labor (IZA) in Bonn is a local and virtual international research center
and a place of communication between science, politics and business. IZA is an independent nonprofit
company supported by Deutsche Post World Net. The center is associated with the University of Bonn
and offers a stimulating research environment through its research networks, research support, and
visitors and doctoral programs. IZA engages in (i) original and internationally competitive research in
all fields of labor economics, (ii) development of policy concepts, and (iii) dissemination of research
results and concepts to the interested public.
IZA Discussion Papers often represent preliminary work and are circulated to encourage discussion.
Citation of such a paper should account for its provisional character. A revised version may be
available on the IZA website (www.iza.org) or directly from the author.
IZA Discussion Paper No. 1098
Nonparametric Bounds on the Returns to Language Skills∗ ∗
This paper applies the theoretical literature on nonparametric bounds on treatment effects to
the estimation of how limited English proficiency (LEP) affects wages and employment
opportunities for Hispanic workers in the United States. I analyze the identifying power of
several weak assumptions on treatment response and selection, and stress the interactions
between LEP and education, occupation and immigration status. I show that the combination
of two weak but credible assumptions provides informative upper bounds on the returns to
language skills for certain subgroups of the population. Adding age at arrival as a monotone
instrumental variable also provides informative lower bounds.
JEL Classification: C14, J24, J31
Keywords: nonparametric bounds, language skills, Hispanic workers, labor market
Departamento de Economía y Empresa
Universitat Pompeu Fabra
Edificio Jaume I
Ramón Trias Fargas 25-27
Tel.: +34 93 542 2610
Fax: +34 93 542 1746
∗ I would like to thank Chuck Manski, Chris Taber, Joe Altonji and Bruce Meyer for their guidance and
useful comments. I also thank Francesca Molinari and three anonymous referees for their suggestions.
All remaining errors are mine.
Hispanics reached 13.3 percent of the total population in the United States in 2002 (37.4
million).1 Hispanic workers earn less than Non-Hispanic Whites and are more likely to be
unemployed. This has been attributed to assimilation problems for immigrants (2 in 5
Hispanics were foreign born in 2002), low levels of schooling (more than 2 in 5 Hispanics
aged 25 and older had not graduated from high school), discrimination, and limited English
proficiency (almost 1 in 3 Hispanics admitted speaking English less than “very well”). This
paper explores the role of limited language skills in explaining wages and unemployment
for Hispanic workers in the US. From a policy perspective, we may be interested in
learning whether, and by how much, investing in language acquisition would help labor
market outcomes for Hispanics converge to those of Non-Whites.2
Previous studies have estimated the effect of language skills on earnings through OLS
regressions.3 Concerns about a possible “ability bias” led some to incorporating IV
estimates,4 but these results tend to be unstable or implausibly large, and the credibility of
the instruments is often questionable. The size of reported OLS and IV estimates varies
considerably across studies and methods.
1 Ramirez and De la Cruz 2003.
2 In this paper I focus on how English proficiency translates into labor market outcomes. In terms of policy,
we would also be interested in learning how language training translates into the actual mastery of the
language. This is outside the scope of this paper. For a study of the determinants of language proficiency see,
for example, Chiswick and Miller (1995).
3 For example, see Reimers 1983, McManus, Gould and Welch 1983, Grenier 1984, McManus 1985, Carliner
4 For example, see Chiswick and Miller 1995 or Bleakley and Chin 2002.
I propose a “conservative” approach in the spirit of Manski (Manski 1990, 1995, 1997,
Manski and Pepper 2000), according to which a worst-case analysis would precede the
study of alternative weak assumptions and their identifying power. Through the
introduction of weak but credible assumptions, I provide identification regions for the
effects under consideration, so that any point estimates obtained under stronger
assumptions should lie within these regions. The results are informative, in the sense that
the upper bounds provided are lower than some point estimates reported in previous
The assumptions I explore include monotone treatment response, as introduced by
Manski (1997), and monotone instrumental variables, in the sense of Manski and Pepper
(2000). I also stress the interactions between LEP and education, occupation and years of
residence in the U.S., and study the possibility of positive sorting into treatments.
All the assumptions considered are informative, i n the sense that they improve the
identification problem with respect to the worst-case situation in which no prior
information is available. However, only when we assume positive selection into treatments
(MTS) and a monotone response function (MTR) do we get bounds narrow enough to be
interesting. The lower bound for the effect of LEP when only MTR and MTS are imposed
is always zero, and the upper bound is slightly higher than the estimate obtained when
assuming that selection into treatments is random (as in OLS). I also propose the use of
immigrants’ age at arrival in the US as a monotone instrumental variable (MIV). This
assumption reduces the MTR+MTS bounds by raising the lower bounds above zero.