Attached.

Running head: LITERATURE REVIEW

1

Higher Education and Labor Force in Mexico
Name
Institution

LITERATURE REVIEW

2

Higher Education and Labor Force in Mexico
In Latin America, and more particularly Mexico, there has been a growing academic
and policy interest in the link between skills being taught in classrooms and its impact on the
labor force among the youth. This is because, on the one side, there is a higher education
attainment rates among youth aged between 20 to 30 years in Mexico. But on the other side,
there is high number of unfilled job positions because of lack of qualified candidates.
Furthermore, it is not just academic qualifications that has an impact on labor force, but also
lack of prior work experience. In this essay, three articles on this subject will be reviewed.
A paper titled “Education and Earnings Inequality in Mexico,” by Ulrich Lachler
discusses this link between education and labor force. It suggests that investment in education
has taken place in a more efficient manner. However despite this, here has been a significant
increase in earning inequality in that there has been a sharp decrease in the earnings of both
highly educated and less educated youth (Lachler, 2000)). This is a surprise because
education has a great equalizing aspect. The author, in this regard, notes that this is a
phenomenon that has been observed in developing countries in recent years. The paper also
finds that this presents a challenge for the country’s long-term growth in that the increased
disparity in income equality reduces the return from investment in higher education, and as
such, means devoting more resources in higher education for economic efficiency.
The World Bank report, Mexico Policy Note 4: Labor Markets for Inclusive Growth,
finds that Mexico currently has low unemployment rates and has its most skilled workforce,
but has had a history of decreased labor productivity that hinders its economic growth. The
Latin American country is experiencing low productivity despite increased attainment in
higher education (World Bank, 2012). There is need to create a more efficient workforce by
increasing the relevance of workforce skills to the market so to align workforce demand by

LITERATURE REVIEW

3

firms and industries. Mexico can therefore improve its allocative efficiency by focussing on
training and education.
Lastly, in the paper “Labor Earnings, Misallocation, and the Returns to Education in
Mexico,” Santiago Levy and Luis Lopez-Calva observe that Mexico has over the past two
decades experienced economic growth and significant progress in education. The earnings of
those workers with nori educational qualifications has, however, dipped. The explanation for
this development is that there has been allocation of resources to less productive firms,
because thee firms are efficient with more productive workers than more educated ones
(López-Calva & Algazi, 2016). The increase in the supply of highly educated youth has
coincided with misallocation of resources to less productive firms, effusively increasing the
gap between supply and demand for highly skilled workers. This widening gap, the paper
concludes, prevents Mexico from taking maximum advantage of its investment in its
workforce.

LITERATURE REVIEW

4
References

Lachler, U. (2000. Education and earnings inequality in Mexico. World Bank Policy
Research Working Paper.
López-Calva, L. F., & Levy Algazi, S. (2016). Labor Earnings, Misallocation, and the
Returns to Education in Mexico.
World Bank. (2012). Labor Markets for Inclusive Growth (Mexico Policy Note 4). World
Bank.


IDB WORKING PAPER SERIES Nº

IDB-WP-671

Labor Earnings, Misallocation,
and the Returns to Education in Mexico

Santiago Levy
Luis Felipe López-Calva

Inter-American Development Bank
Department of Research and Chief Economist
February 2016

Labor Earnings, Misallocation,
and the Returns to Education in Mexico

Santiago Levy*
Luis Felipe López-Calva**
* Inter-American Development Bank
** World Bank

February 2016

Cataloging-in-Publication data provided by the
Inter-American Development Bank
Felipe Herrera Library
Levy, Santiago.
Labor earnings, misallocation, and the returns to education in Mexico / Santiago
Levy, Luis Felipe López-Calva.
p. cm. — (IDB Working Paper Series ; 671)
Includes bibliographic references.
1. Wages-Mexico. 2. Resource allocation-Mexico. 3. Education-Economic aspectsMexico. I. López-Calva, Luis Felipe. II. Inter-American Development Bank. Department
of Research and Chief Economist. III. Title. IV. Series.
IDB-WP-671

http://www.iadb.org
Copyright © 2016 Inter-American Development Bank. This work is licensed under a Creative Commons IGO 3.0 AttributionNonCommercial-NoDerivatives (CC-IGO BY-NC-ND 3.0 IGO) license (http://creativecommons.org/licenses/by-nc-nd/3.0/igo/
legalcode) and may be reproduced with attribution to the IDB and for any non-commercial purpose, as provided below. No
derivative work is allowed.
Any dispute related to the use of the works of the IDB that cannot be settled amicably shall be submitted to arbitration pursuant to
the UNCITRAL rules. The use of the IDB's name for any purpose other than for attribution, and the use of IDB's logo shall be
subject to a separate written license agreement between the IDB and the user and is not authorized as part of this CC-IGO license.
Following a peer review process, and with previous written consent by the Inter-American Development Bank (IDB), a revised
version of this work may also be reproduced in any academic journal, including those indexed by the American Economic
Association's EconLit, provided that the IDB is credited and that the author(s) receive no income from the publication. Therefore,
the restriction to receive income from such publication shall only extend to the publication's author(s). With regard to such
restriction, in case of any inconsistency between the Creative Commons IGO 3.0 Attribution-NonCommercial-NoDerivatives license
and these statements, the latter shall prevail.
Note that link provided above includes additional terms and conditions of the license.
The opinions expressed in this publication are those of the authors and do not necessarily reflect the views of the Inter-American
Development Bank, its Board of Directors, or the countries they represent.

Abstract *
Over the last two decades Mexico has had an open trade regime, experienced
macroeconomic stability, and made substantial progress in education. However,
average workers’ earnings have stagnated and earnings for workers with more
schooling have declined, compressing the earnings distribution and lowering the
returns to education. We hypothesize that these developments are explained by
large and persistent of distortions that misallocate resources towards less
productive firms, since these firms are substantially less intensive in educated
workers than more productive ones. We show that at the same time that the
relative supply of workers with more years of schooling has increased,
misallocation of resources toward less productive firms has persisted. These two
trends have generated a widening mismatch between the supply and demand for
educated workers. We decompose worker earnings into observable and
unobservable firm and individual worker characteristics, and simulate a
counterfactual earnings distribution in the absence of misallocation. Under the
counterfactual, earnings differentials across schooling levels would increase, as
would the returns to education. In parallel, earnings differentials, rather than
narrowing overtime, would widen. We conclude arguing that the persistence of
distortions that misallocate resources toward lower-productivity firms impedes
Mexico from taking full advantage of its investments in the human capital of its
workers.
JEL classifications: J24, J23, O17, L11
Keywords: Earnings, Misallocation, Returns to education, Human capital

*

Santiago Levy is with the Inter-American Development Bank and Luis Felipe López-Calva with the World Bank. Their
opinions do not necessarily coincide with those of the institutions they are affiliated with. The authors are very grateful to
Matias Morales for excellent research assistance and to Luca Flabbi, Rafael de Hoyos, Samuel Freije, Julian Messina,
Hugo Nopo and Norbert Schady for very useful comments on an earlier draft. They also thank, without implicating,
Rodrigo Negrete and Tomas Ramirez from Mexico’s Instituto Nacional de Estadística, Geografía e Informática for their
help interpreting the employment surveys.

1

Contents

1.

Introduction

2.

The Wage Premium and the Returns to Education in Mexico

3.

Definitions and Data
3.1
3.2

4.

Misallocation and Firm Formality and Informality
4.1
4.2
4.3
4.4

5.

Size and Type Distribution of Firms
Large and Persistent Misallocation
Productivity Differences between Formal and Informal Firms
Evolution of Firm Size and Type between 1998 and 2013

Workers’ Demographics, Schooling, and Earnings
5.1
5.2

6.

Definitions
Data

Descriptive Statistics
Returns to Education: Preliminary Overview

Misallocation, Earnings and the Returns to Education, 2006
6.1
6.2
6.3
6.4
6.5

Formal-Informal Firm Differences
Employee Mobility across Firm Size and Type
Earnings Regressions by Educational Level with Workers’ Fixed Effects
Counterfactual Earnings Distributions and Returns to Education
General Equilibrium Effects

7.

The Wage Premium, 1996–2015

8.

Conclusions

Appendix: The Problem of Missing Observations

2

Figures
1.
2.
3.
4.
5.
6.
A1.
A2.
A3.
A4.

Sample Employee Earnings, 1996–2015
Years of Schooling and Share of Informal Employment, Economically Active
Population and Sample, 1996–2014
Returns to Education, 1996–2015
Observed and Simulated Earnings Distributions, 2006
Excess Supply of Workers Relative to Those with a University Education, 1996–2015
Earning Premiums: University versus Completed Primary, 1996–2015
Rates of Under-reporting of Earnings, ENE-ENOE, 1996–2015
Rates of Under-reporting of Earnings in Mexico’s National Household Income and
Expenditure Survey (ENIGH), 1996–2012
Earnings by Educational Level, ENIGH, 1994–2012
Evolution of Returns to Schooling by Educational Level, ENIGH, 1996–2012

Tables
1.
2.
3.
4.
5.
6.
7.
8.
9:
10.
11.
12.
13.
14.
15.
A1.

Formality Status of Firms and Workers
Size and Type Distribution of Firms and Employment, 2008
Measures of Dispersion of Firm Productivity, 1998–2013
Productivity Differences by Firm Type, 1998–2013
Size and Type Distribution of Firms and Employment, 1998–2013
Workers’ Descriptive Statistics, 1996–2015
Characteristics of the Sample of Workers
Returns to Education, Sample, 2006
Distribution of Employees by Education and Firm Size, 2006
Bonus and Contracts by Education and Firm Size, 2006
Mobility of Individual Workers across Firm Size and Type, 2005.Q3–2007.Q1
Earnings Regressions by Educational Level, Anchor 2006.Q2,
Observed and Simulated Earnings, 2006.Q2 Anchor
Observed and Simulated Returns to Education, 2006.Q2 Anchor Panel 2005.Q3–
2007.Q1
Excess Supply of Workers by Schooling Category, 2006
Employee Earnings, 2006

3

1. Introduction
Over the last two decades, Mexico has made notable efforts to increase the schooling of its
workers, in the hope that accumulating human capital will lead to higher earnings and better
jobs, covered by labor and social insurance regulations. Indeed, there has been a significant
increase in schooling levels: in 1996, working-age Mexicans (18 or older) had on average 4.7
years of education; by 2015 that figure had almost doubled to 9.2 years. Similarly, in 1996 less
than 19 percent of working-age Mexicans had completed high school; by 2015, 33 percent had
high school degrees.
But despite these achievements and the fact that these two decades were characterized by
macroeconomic stability and a large opening to international trade, hopes for higher earnings and
better jobs have not materialized. As documented in this paper, the share of jobs covered by
labor and social insurance regulations has remained essentially constant. And average hourly
earnings, after recuperating from the 1995 financial crisis, have in fact fallen slightly, as a result
of an absolute decline in the earnings of workers with more years of schooling. This has by-andlarge offset the expected increase in average earnings associated with the change in the schooling
composition of the labor force.
This paper argues that misallocation of resources—as evidenced by large differences in
the productivity of resources across firms—explains these phenomena. The paper presents a
preliminary exploration of the impact of misallocation on labor earnings and the returns to
education. We have two basic hypotheses. First, in the particular case of Mexico, distortions
result in too many resources allocated toward low-productivity firms that demand less-educated
workers, depressing the earnings of workers with more education. 1 Second, over the last two
decades these distortions have persisted even while the composition of the labor supply has
changed toward more educated workers, implying a growing mismatch between the supply and
demand for workers with more years of schooling.
Our paper can be seen as a bridge between the literature on misallocation and the
literature on returns to schooling in the particular context of Mexico, where misallocation looms
1

In this paper the word “distortions” is to be interpreted very broadly, as any market or regulatory failure or frictions in
output, labor, and credit markets that cause wedges between the marginal revenue products of labor and capital across
firms. Distortions can result from the interaction of many policies related to, among other things, taxation, credit, labor
and social insurance regulations (including the nature of their enforcement). Distortions can also come from the absence
of policies to correct for market failures, artificial barriers to entry or special subsidies to firms or sectors, or high
registration or transaction costs.

4

large. The literature on misallocation starts from the premise that in the absence of distortions,
individuals efficiently distribute themselves between entrepreneurs, employees, and selfemployment. In turn, entrepreneurs hire the efficient number of employees given their abilities.
The resulting distribution of individuals across occupations, and of firms across sizes, maximizes
the productivity of the economy and the returns to factors. But if distortions are present, the
distribution of individuals across occupations and the size distribution of firms are suboptimal:
some individuals who should be employees are entrepreneurs (or vice versa), while some firms
are larger (or smaller) than they should be given their underlying productivity. In parallel, firms
may change the nature of the contracts offered to their workers. The implied misallocation of
capital and labor lowers aggregate productivity and distorts the returns to factors (Hsieh and
Klenow, 2009; Restuccia and Rogerson, 2008).
The empirical evidence for Mexico summarized below shows that misallocation is very
relevant. While there is an important debate as to the exact nature of the distortions that induce
this phenomenon, three results are robust: distortions result in large productivity losses; operate
in the direction of allocating too much capital and labor to low-productivity firms that are less
intensive in educated workers; and are persistent through time.
On the other hand, the literature on the returns to education has focused on understanding
the relative importance of supply and demand factors in determining the distribution of earnings
across educational levels. In Mexico’s case, attention has focused on the fact that the earnings
differential between workers with more and fewer years of education, at times called the wage
premium, has narrowed over the last decade, if not before. 2 This finding is puzzling because, on
the one hand, human capital is thought to be a constraint on growth in Mexico; and, on the other,
the finding is the opposite of the trend found in the United States—by far Mexico’s largest
trading partner—where the wage premium has actually increased (Autor, Katz, and Kearney,
2008; Goldin and Katz, 2007).
In an immediate sense, of course, the fact that given the composition of the demand for
labor, the earnings of workers with more years of education fall as their relative supply increases
suggests a normal market adjustment. But this explanation is almost tautological, and begs the
2

Lustig and López-Calva (2010) find that there has been a steady decline in the wage premium between skilled and
unskilled workers at least since 2002. Robertson (2007) suggests that the decline started at the end of the 1990s. CamposVázquez et al. (2010) and Campos, Esquivel and Lustig (2012) find that returns to schooling started to decline after
1994.

5

question as to why demand for workers with more education has lagged. Moreover, this
explanation does not square with the empirical evidence for Mexico. As documented below,
during the last two decades the earnings of workers who completed junior high school have
increased relative to workers with a university education, despite the fact that the supply of the
former has increased faster than the latter. This suggests that, in addition to supply-side
considerations, the determinants of the schooling composition of the demand for labor have
played a critical role.
Our bridge between the literature on misallocation, and that on returns to education,
involves focusing on firms as an important observable determinant of workers’ earnings, in a
context where as a result of misallocation, the number, type, and size of firms participating in the
demand side of the labor market is strongly distorted in the direction of low-productivity firms
that are intensive in less-educated workers. The perspective taken here is that given workers’
observable and unobservable characteristics, their earnings partly depend on the nature of the
firms that employ them. In this context, we explore how the size distribution of firms (measured
by the total number of workers) and the type distribution of firms (measured by the contractual
composition of their workforce) affect the distribution of employees’ earnings and the returns to
schooling.
This line of inquiry is relevant because of three empirical regularities documented below.
First, controlling for firm type, larger firms are more intensive in educated workers than smaller
ones. Second, controlling for firm size, firms that offer their workers contracts with labor and
social insurance regulation coverage are more intensive in educated workers than other firms.
Third, there is a strong positive correlation between firms that are large and firms that offer their
workers contracts with labor and social insurance regulation coverage. When, as a result of
distortions, too few resources are allocated to these firms—as we document to be the case in
Mexico—the schooling composition of the demand for labor will tilt in the direction of workers
with fewer years of education.
Firms that offer contracts to their workers with coverage of labor and social insurance
regulations are typically referred to as formal firms, while those that do not offer such coverage
are known as informal firms. In turn, workers are referred to as formal or informal depending on
whether they are employed by the former or the latter set of firms. Using that terminology, one
can state that in Mexico distortions result in too many resources allocated to informal firms, and
6

in too many informal workers. In short, distortions generate large informality. Critically,
however, what matters for our analysis is how firms behave, and not how they are labelled.
Informality is a manifestation of distortions that result in a market equilibrium with too many
low-productivity firms, low demand for educated workers, and jobs not covered by labor and
social insurance regulation. Firm labels could change without changing the underlying
distortions that determine firm behavior, in general, and the schooling composition of their
demand for labor, in particular. 3 That said, and following standard practice, we will refer in this
paper to firms and workers as formal and informal, but with the understanding that the focus is
on the underlying phenomenon of misallocation, and not on the formal-informal labels.
A large literature has focused in understanding the role of taxation, social insurance and
labor regulations, credit frictions, market failures, and other factors like registration and
transaction costs in generating misallocation; see IDB (2010) for a summary. In the case of
Mexico, Busso, Fazio and Levy (2012) and Levy (2008) have emphasized the role of labor and
social insurance regulations; Leal (2014) the role of taxation, and López-Martin (2015) the role
of credit. In all these cases, misallocation results in too many low-productivity firms employing
too many workers without coverage of labor and social insurance regulations—in other words, a
large informal sector. In parallel, an emerging literature is focusing on the links between
misallocation and human capital, using models where the size distribution of firms is
endogenous. Torres-Coronado (2015) analyzes the impact of size-dependent firm taxes on the
returns to skills, and Busso, Neumayer, and Spector (2015) study the interaction between firm
size and the distribution of skills. Finally, Bobba, Flabbi, and Levy (2016) focus on the impact of
labor market distortions on the returns to education in a search-bargaining model.
In this paper we do not model the frictions or market or regulatory failures that generate
misallocation and distort the size and type distribution of firms. Rather, we follow a three-step
approach to test our hypothesis. In the first step, we estimate a model of individual workers’
earnings that controls for all observable worker characteristics but focuses on estimating the
coefficients associated with observable firm characteristics.
3

This observation is relevant because policy can change the formal-informal firm labeling without affecting firm
behavior. An example would be so-called “formalization programs” that offer subsidies to firms to register with social
security authorities, but do not change underlying distortions in output, credit, tax, and labor markets faced by these
firms, and therefore do not change their demand for labor. In this case, firm formality would increase while misallocation
would persist. Similarly, self-employed workers may be offered subsidies to induce them to formalize, in which case
labor informality could decline, but again without any changes in firm behavior.

7

In the second step, we consider the implications of eliminating misallocation, interpreted
here as eliminating firm informality, only from the point of view of individual workers. To do
this, we construct a counter-factual earnings distribution keeping constant individual workers’
unobservable characteristics, as well as observable characteristics like years of schooling, age,
gender, and location, but assuming that the size and type distribution of firms mimics that of
formal firms. Our purpose is to measure how workers’ earnings are affected when as a result of
misallocation there are too many informal firms in the demand side of the labor market,
independently of workers’ education and abilities. In this context, we show that eliminating firm
informality increases average earnings and changes the composition of the demand for labor,
augmenting the demand for more educated workers relative to those with fewer years of
schooling; and thus increasing the returns to schooling. The mean of earnings across all
educational levels is higher and the distribution widens. Put differently, the distortions that
misallocate resources toward informal firms act like a penalty on earnings that is paid by all
workers but proportionately more by the more educated. Misallocation matters more to educated
workers than to workers with little schooling.
In the third and final step we consider the aggregate implications of eliminating
misallocation. We show that, given the supply of workers from each educational level, if the
schooling composition of the demand for labor in the economy were that of formal firms, there
would be an excess supply of workers with few years of education. We measure the size of
excess supply and, critically, show that it would increase overtime. Next, for given values of the
elasticity of substitution between workers of different schooling levels, we compute the changes
in earnings required to absorb excess supply. We then compare the observed path of the ratio of
earnings of workers with more versus less years of schooling (i.e., the wage premium) with
alternative paths where there is no firm informality and where earnings adjust in each period to
clear the market. We find that in the absence of firm informality the wage premium is
substantially higher, and that the difference vis-à-vis the observed premium increases overtime.
These results, in turn, suggest that the returns to education would be even higher than what our
counter-factual simulations indicate.
Of course, if the distortions that misallocate resources toward the informal sector were
removed, there would be further changes in the economy that our approach fails to capture. As
the earnings of employees relative to those of the self-employed increase, there would be
8

changes in the distribution of workers between these two occupational categories. Participation
rates would also change, as would the equilibrium level of unemployment. Further, even the size
distribution of formal firms would change, probably in the direction of larger firms. Clearly, a
model with more structure than what we present here is needed to fully measure these general
equilibrium effects. Nonetheless, our preliminary results do show that by distorting the schooling
composition of the demand for labor, misallocation is lowering the returns to education in
Mexico, and that the persistence of misallocation in the face of increased schooling of the labor
force goes a long way toward explaining the observed downward trend of the wage premium.
Section 2 of this paper briefly reviews the literature on the behavior of the wage premium
and the returns to education in Mexico. Section 3 defines firm and worker informality and
describes our data. Section 4 documents misallocation and describes the size and type
distribution of firms, while Section 5 presents stylized facts on workers’ earnings and schooling.
Section 6 analyzes the impact of observable firm characteristics on workers’ earnings and
constructs a counterfactual scenario where firm informality is absent. Section 7 sheds light on the
evolution of the wage premium over the 1996-2015 period. Section 8 presents concluding
remarks.

2. The Wage Premium and the Returns to Education in Mexico
The general consensus regarding earnings and the returns to schooling in Mexico is that the
premium paid to higher-skilled labor increased with the take-off of the North American Free
Trade Agreement (NAFTA) in 1994 and then began to decline, with the returns to schooling
following a similar trend.
Bouillon (2002) defines the wage premium as the ratio of wages of workers with more
than a high school education to those with primary education or less, and finds that the premium
rose between 1984 and 1994. Lächler (1998) finds an increase in the dispersion of earnings
across different schooling levels from 1984-1994, and also finds that wages rose for workers
with a high school education or more but fell for less-educated workers. Esquivel and RodríguezLópez (2003) use a different definition of wage premium, focusing on skilled and non-skilled
workers (defined as non-production and production workers), and consider only manufacturing
employment. In accordance with Bouillon (2002) and Lächler (1998), they find that the wage
premium rose after 1988, but plateaued in the mid-1990s. Robertson (2007) defines the wage
9

premium as in Esquivel and Rodríguez-López (2003), and finds that it rose steeply prior to 1994,
continued to rise slowly until 1999, and declined in the period up to 2005.
With regard to the returns to schooling, Campos, Esquivel, and Lustig (2012) find that
relative returns to skilled workers (defined as those holding a high school degree or more versus
junior high or less) increased between 1989 and 1994 but declined thereafter. Benita (2014)
looks only at a subsequent period, from 2005 to 2012, and finds that the wage premium (defined
in this case as the wages of university versus high school-educated male workers) declined for
younger workers (ages 25-29), increased for older ones (ages 45-49), and remained constant for
the oldest (ages 50-59). His findings suggest a large elasticity of substitution between workers
with different levels of education (university and high-school workers appear to be
interchangeable to employers).
Lustig, López-Calva, and Ortiz-Juarez (2014) focus on the 1990s and 2000s and define
the wage premium as the returns to primary, secondary, and tertiary education versus no
schooling or incomplete primary schooling. They find a decline in the returns to education,
especially during the 2000s. This is in line with the decline in relative returns for high-skilled
workers that Campos, Esquivel, and Lustig (2012) find for the 1994-2010 period. Lustig, LópezCalva, and Ortiz-Juarez (2014) also discuss the various explanations given for this behavior: an
increase in the supply of workers with higher educational attainment; a decline in the demand for
skilled labor; a decline in the quality of higher education; and/or a mismatch in the supply and
demand of skills. The authors note that none of these factors has been unambiguously identified.
Three observations are relevant from this brief review. First, studies vary in scope
(manufacturing industry, the export sector, the overall economy), data sources, time periods, and
groups used to define wage/skill premiums (by schooling levels or by workers’ roles in the
production process). 4 Second, despite this dispersion, on the whole there is agreement that after
an initial widening following the start of NAFTA, the difference in mean earnings between
workers with more versus fewer years of education narrowed, and that as a result the returns to
education have declined.

4

In some cases it is unclear whether “workers” refers only to employees or also to the self-employed; to female and male
workers or only to the latter; or to formal and informal workers, or only to the former. In terms of data, studies use, inter
alia, household surveys, employment surveys, administrative data from social security institutions, and sectoral surveys,
particularly of manufacturing.

10

Third, even though not always explicitly noted, it is clear that average earnings over the
last several decades have been strongly influenced by transitory macro shocks. In particular, it is
not surprising that earnings rose in the late 1980s as the economy emerged from that decade’s
debt crisis, but then declined sharply immediately after the 1995 peso crisis. As a result, to
isolate the effects of transitory macro shocks, studies have focused on the structure of earnings
across educational groups, trying to disentangle the relative importance of changes in the
composition of supply (like demographics and educational investments) from changes in the
composition of demand (like NAFTA or technical change). In this context, it is somewhat
surprising that studies have by-and-large ignored the role of one of the more salient characteristic
of the Mexican economy: that as a result of misallocation, most firms, and workers are informal.
The remainder of this paper makes a first attempt to assess the effects of misallocation on
relative earnings and the returns to education.

3. Definitions and Data
3.1 Definitions
We distinguish between self-employed workers and workers engaged with firms. The latter—
employees—can be so under salaried or non-salaried contractual relationships, a key distinction
in Mexico’s institutional context. A salaried worker is a subordinated employee. The hiring firm
is obligated by law to pay him or her at least the minimum wage, observe regulations regarding
promotions and dismissals, and contribute to the worker’s social insurance benefits. In turn, the
worker has various rights, among them to unionize and to sue the boss if dismissed for an
unjustified cause. Non-salaried workers, on the other hand, may be associated with a firm, but
are not subordinated to it. Examples include workers who sell door-to-door, workers on a
temporary contract performing a non-recurrent task and, very importantly for the case of Mexico,
workers who are relatives and collaborate in a family firm. Critically, the law does not obligate
firms to contribute to the social insurance benefits of non-salaried workers, or to observe
regulations regarding dismissal, promotions, or minimum wages. Further, non-salaried workers
cannot take a firm to court for dismissing them because there is no relation of subordination, nor
can they unionize.
The distinction between salaried and non-salaried employment is the basis for the
distinction between formal and informal workers. We define a formal worker as a salaried
11

employee covered by labor regulations regarding minimum wages, unionization, and dismissal,
among other things, and who benefits from social insurance paid by the firm that hires him or
her. All other workers, including the self-employed, are informal. If the law were fully enforced,
all salaried workers would be formal. However, this is not the case in Mexico (Busso, Fazio, and
Levy, 2012). This implies that informal workers consist of the self-employed, non-salaried
employees, and salaried employees in firms that do not comply with the law.
The formal-informal distinction is not as sharp in the case of firms because they mix
salaried and non-salaried contracts and at times violate the law. Table 1 identifies five possible
combinations (all observed in the data). One implication of firms with mixed contracts is that
they make it difficult to identify the formal and informal sectors with precision. Clearly, the
former consists of at least those workers and firms in column two, while the latter consists of at
least those workers and firms in columns three, six, and seven. But there are some firms that are
neither purely formal nor informal, mixing salaried and non-salaried employees but complying
with the law (column four), or partly violating it (column five).

Table 1. Formality Status of Firms and Workers
(2)
Only salaried,
compliant with
law
Firm

Worker

Contracts between Firms and Employees
(3)
(4)
(5)
Only nonMixed, but
Mixed, but not
salaried, not
compliant with fully compliant
obligated by law
law
with law

Formal and
legal

Informal and
legal

Formal

Informal

Semi-formal
and legal

Semi-formal
and semi-legal

Salaried formal
and
Non-salaried
informal

Salaried
compliant
formal; the rest
informal

(7)
Only salaried,
not compliant
with law

(8)
Selfemployed
workers

Informal and
illegal

Not
applicable

Informal

Informal

Source: Prepared by the authors.

3.2 Data
Our analysis relies on Mexico’s employment surveys and economic census, and focuses on the
period 1996-2015, after the start of NAFTA in 1994 and the financial crisis of 1995. From 1996
to 2004 the survey was known as the National Employment Survey (Encuesta Nacional de
Empleo—ENE); after 2005 it was known as the National Occupation and Employment Survey
12

(Encuesta Nacional de Ocupación y Empleo—ENOE). We refer to it here as the ENE-ENOE, a
nationally-representative quarterly survey on type of employment (public or private employees,
or self-employed); labor status (formal, informal, or unemployed); location (municipality); size
of firm where workers are employed; workers’ age, gender, and years of schooling; and other
dimensions of a job like a written contract and yearly bonus payments. The ENE-ENOE also
records hours worked and earnings net of taxes and contributions, whether they take the form of
wages, salaries, commissions, or bonuses. We use data from the second quarter of each year and
apply the average of the corresponding monthly price indices. All earnings are measured per
hour in prices of May 2008.
The relevant educational categories for Mexico are primary (six years, usually ages 612); junior high (three years, usually ages 13-15); high school (three years, usually ages 16-18);
and university (four years or more). We use these categories to classify workers in seven groups:
incomplete primary, complete primary, incomplete junior high, complete junior high, incomplete
high school, complete high school, and university.
Importantly for the econometric analysis in Section 6, the employment surveys have a
panel structure that allows for following the same worker through five consecutive quarters.
Since in each quarter we can identify firm size and worker’s labor status, we can measure
individual worker transitions in the course of a year across firm size and formal-informal status.
An issue with the ENE-ENOE is that some workers fail to report earnings (CamposVazquez, 2013). To correct for this problem, we match workers with and without earnings on
observable characteristics like gender, years of education, age, location, size of firm, and
formality status. We take a random sample of workers with reported earnings in each of these
categories, and randomly impute their earnings to workers with the same observable
characteristics but without reported earnings. 5 The Appendix provides more details.
While the ENE-ENOE is a household-based employment survey, the Economic Census is
a firm-based data set published every five years. We have four available censuses for 1998,
2003, 2008, and 2013, a subset of the years for which we have employment data. The census
collects information on economic activity in urban areas in fixed establishments of all sizes
5

To check the robustness of our procedure we compute earnings with data from Mexico’s household survey which, as
documented in the Appendix, does not have the same under-reporting problem as the ENE-ENOE. While the definitions
of earnings do not match perfectly between the two surveys, the trends and structure of earnings, including the decline in
the returns to schooling, are confirmed.

13

(henceforth, firms). Economic activity in rural areas, and on the streets in urban areas (street
vendors, street markets, and so on) is not captured. Public sector employment is also excluded.
As a result, the census only captures about 45 percent of the occupied population captured in the
ENE-ENOE. However, as detailed below, we restrict the ENE-ENOE sample to private sector
employees in firms in urban areas, ensuring that they are a subset of the employees captured in
the census.
The census has data on the total number of workers and the aggregate of earnings and
social security payments at the firm level, which allows us to classify firms according to the
typology described in Table 1 (Busso, Fazio, and Levy, 2012). We also group firms by size into
four categories: 0 to 5 workers (henceforth, very small firms), 6 to 10 (small firms), 11 to 50
(medium firms), and 51 or more (large firms).
Unfortunately, the census does not have data on individual workers within each firm, and
thus provides no information on hours worked, gender, or years of education. On the other hand,
the ENE-ENOE identifies the size of the firm where the worker is employed, so, critically, we
can classify workers by years of schooling, firm size, and formal-informal status. 6 However, our
data do not allow for mapping formal and informal workers as identified in the ENE-ENOE into
the formality status of firms as identified in the census. Given that in our analysis we focus on
the earnings of individual workers by years of schooling, firm size, and formality status, we
mostly rely on ENE-ENOE data. However, we use the census to measure misallocation and the
size and type distribution of firms between 1998 and 2013.

4. Misallocation and Firm Formality and Informality 7
This section documents four sets of findings. First, most firms are small and informal, and they
absorb a large share of employment. Second, in Mexico there is large and persistent
misallocation of resources. Third, misallocation systematically operates in the direction of
channeling too many resources to informal firms. Lastly, between 1998 and 2013 the number of
formal firms declined, while that of informal ones increased. Over the same period employment
grew substantially more in informal firms.

6
7

We do this following the methodology developed by INEGI (2014).
This section borrows heavily from Busso, Fazio, and Levy (2012) and Busso and Levy (2016).

14

4.1 Size and Type Distribution of Firms
Table 2 serves to make four observations on the distribution of firms and employment in 2008.
First, 89.7 percent of all firms are very small (up to five workers); and 90.1 percent are purely
informal (legal and illegal). Second, purely informal firms have 2.9 workers, on average, but
account for more than half of all employment captured in the census (54.5 percent). On the other
hand, purely formal firms have 32.2 workers on average (and 22.3 if we extend the definition of
formal firms to include mixed firms). Altogether these firms account for 45.5 percent of
employment captured in the census.
Table 2. Size and Type Distribution of Firms and Employment, 2008
(in percent)

Firms*
[0-5 workers]
[6-10]
[11-50]
[50+]
Total
Workers**
[0-5]
[6-10]
[11-50]
[50+]
Total

Legal &
Formal

Legal &
Informal

1.21
0.69
0.87
0.29
3.06

65.60
1.11
0.56
0.19
67.46

0.77
1.09
3.84
14.67
20.37

23.60
1.63
2.39
7.74
35.36

Legal &
Semi-formal

Semi-legal &
Semi-formal

Ilegal &
Informal

Total

1.18
0.33
0.15
0.03
1.69

2.49
1.23
1.10
0.30
5.12

19.25
2.46
0.90
0.05
22.66

89.73
5.82
3.58
0.86
100.00

0.79
0.50
0.57
1.19
3.05

1.69
1.92
4.72
13.75
22.08

10.94
3.66
3.36
1.18
19.14

37.79
8.80
14.88
38.53
100.00

Source: Busso, Fazio, and Levy (2012). *3.643 million; **17.655 million.

Second, note that most very small firms are informal but legal. When a firm has few
workers, it is easy to monitor effort, coordinate activities, and reach agreements to distribute
profits. This is the case of cooperatives or, more relevant in Mexico, firms where workers are
related (family firms). In these cases it may be efficient for the firm to establish non-salaried
contracts. Indeed, as shown in Table 2, this is the most common contractual structure of firms in
Mexico.
Our third observation is that as the number of workers in the firm increases, it is more
difficult to coordinate tasks unless there are relationships of subordination, and to observe
individual effort. Problems of shirking and free-riding thus soon appear. Moreover, if the
production technology calls for a fixed place of work and close coordination between tasks

15

performed by different workers at the same time, salaried contracts will be more efficient. 8 As a
result, on average, larger firms have proportionately more salaried contracts than smaller ones:
firms with only salaried workers, formal and informal, constitute only 25.7 percent of all firms
but employ 39.5 percent of all workers.
Our fourth observation is that considering only firms with salaried contracts, formal firms
are larger than informal ones: on average, 32.2 versus 4.1 workers. This results from the
imperfect enforcement of laws regarding salaried contracts. Since the probabilities of being
detected by the authorities are proportional to the size of the firm, firms with illegal salaried
contracts tend to be small, as the expected marginal costs of labor increase sharply with size
(Anton, Hernandez, and Levy, 2012).
This brief discussion about the type and size distribution of firms is central to our
analysis of labor earnings and the returns to education in Mexico. This is because firms’
demands for workers of various schooling levels depend on those firms’ size and
formal/informal status. Transportation services can be provided by a hundred self-employed
workers driving their own trucks, or by a single firm with a hundred salaried employees. In both
cases there will be a hundred trucks and a hundred drivers, but in the latter case there will be a
need for a logistics engineer doing dispatches and a sales manager. Tortillas can be produced
with simple technologies in small establishments with unskilled labor, or in large plants needing
engineers; the same holds for apparel and food processing, among many manufacturing
activities. And the same holds for retail commerce: it can be carried out in small stores
employing workers with only basic literacy and numeracy or through large supermarket chains
requiring, say, industrial designers. In general, the complexity of tasks and the division of labor
increase with firm size, and generate a need for more educated workers. But the issue is not only
size: a small informal firm producing jeans for sale in a street market will be less likely to need
an accountant than a formal firm of the same size selling jeans to a large retailer.
As discussed, the census data unfortunately contain no information on the schooling
composition of workers. However, in Table 9 later in this paper we use the ENE-ENOE data to
document two additional empirical regularities that are also central to our analysis: that larger

8

These differences matter for earnings: for salaried workers they take the form of wages (a fixed amount of money per
unit of time), while for non-salaried workers they take the form of payments per product regardless of the time required
to produce it, profit-sharing, or commissions based on sales.

16

firms are more intensive in workers with more years of schooling than smaller ones; and that
controlling for size, informal firms are less intensive in educated workers than formal ones.
4.2 Large and Persistent Misallocation
We next turn to evidence of misallocation. Following Hsieh and Klenow (2009), we define total
revenue productivity, TFPRis, as the value of the output produced by firm i in sector s with one
peso of capital and labor (a weighted average of the marginal revenue products of the labor and
capital used by that firm). In turn, TFPQis is the physical productivity of resources (a weighted
average of the marginal products of labor and capital). In the absence of any distortions that
would misallocate resources across firms, revenue productivity would be the same for all firms
in a given sector and across all sectors. This implies that the greater the dispersion of TFPR, the
greater the degree of misallocation. Table 3 presents three measures of the dispersion of TFPR
and TFPQ. 9
Table 3. Measures of Dispersion of Firm Productivity, 1998-2013
1998
TFPR TFPQ
Standard deviation
p75 - p25
p90 - p10

1.15
1.55
2.95

1.75
2.44
4.58

2003
TFPR TFPQ
1.14
1.50
2.91

1.77
2.38
4.57

2008
TFPR TFPQ
1.23
1.60
3.14

1.90
2.60
4.91

2013
TFPR TFPQ
1.24
1.55
3.14

1.85
2.48
4.77

Source: Busso and Levy (2016).

Two facts follow from Table 3. First, there is a large dispersion in revenue productivity
across firms, implying substantial misallocation of resources. For example, in 2013 the firm in
the 75th percentile of the revenue productivity distribution was 55 percent more productive than
the one in the 25th percentile. 10 Second, this dispersion persisted (and in fact increases slightly)
over the 15 years considered, indicating the persistence of misallocation of capital and labor.
9

Computations are done at the six digit sector level and include firms of all sizes in manufacturing, services, and
commerce. There are 559 sectors in 1998, 699 in 2003, 707 in 2008, and 735 in 2013. Comparisons of TFPR and TFPQ
are only made for firms within the same sector. The numbers in Table 3 are averages across all sectors. As expected,
dispersion of TFPR is smaller than that of TFPQ, because when firms produce more physical output they sell at lower
prices (so that revenue-based measures of productivity tend to underestimate variation in producers’ physical
efficiencies). See Syverson (2011).
10
The difference between firms in the 90th and 10th percentile is 214 percent. This compares with a difference of 92
percent for the same range in the manufacturing sector of the United States, as reported by Syverson (2004). Importantly,
Syverson’s computations are carried out at the four-digit level, and one would expect smaller differences at the six-digit
level. IDB (2010) compares the dispersion of revenue productivity between manufacturing firms in the United States and
Mexico at the four digit level, and finds that dispersion is substantially higher in Mexico. Busso, Madrigal, and Pages
(2013) find that dispersion is also higher in Mexico compared to other Latin American countries.

17

4.3 Productivity Differences between Formal and Informal Firms
Table 4 compares productivity across firm types. 11 In all years, formal firms are more productive
than all other firms. If we focus only on pure informal firms, legal and illegal, ignoring mixed
firms, it turns out that depending on the year considered, their physical productivity is between
158 and 28 percent lower than that of pure formal firms. If we focus on revenue productivity, the
differences are between 60 and 10 percent. Note that legal informal firms, which as shown in
Table 2 constitute the majority of firms in Mexico, are always the least productive of all, and that
productivity differences between these firms and illegal informal ones are significant. That said,
there are two critical results for our purposes. First, the fact that informal firms, legal or illegal,
have systematically lower revenue productivity than formal ones indicates that the effect of
distortions in Mexico is to allocate too many resources to informal firms. Second, the fact that
this result is observed in all periods considered shows that distortions operate systematically in
the same direction.

Table 4. Productivity Differences by Firm Type, 1998-2013
(In percent, relative to formal legal firms)

Legal &
Semi-formal

1998
TFPQ
TFPR
-0.635
-0.360
(0.0064) (0.0042)

2003
TFPQ
TFPR
-0.497
-0.340
(0.0058) (0.0038)

2008
TFPQ
TFPR
-0.672
-0.418
(0.0072) (0.0046)

2013
TFPQ
TFPR
-0.370
-0.394
(0.0066) (0.0041)

Legal &
Informal

-1.585
(0.0043)

-0.582
(0.0028)

-1.475
(0.0040)

-0.644
(0.0026)

-1.130
(0.0043)

-0.467
(0.0027)

-1.404
(0.0035)

-0.690
(0.0022)

Semi-legal
&
Semi-formal

-0.096
(0.0039)

-0.010
(0.0025)

-0.012
(0.0035)

-0.049
(0.0022)

0.035
(0.0041)

-0.053
(0.0026)

-0.058)
(0.0036)

-0.185
(0.0023)

-0.541
(0.0045)

-0.201
(0.0029)

-0.285
(0.0040)

-0.102
(0.0026)

-0.488
(0.0044)

-0.208
(0.0028)

-0.414
(0.0041)

-0.157
(0.0025)

Illegal &
Informal
Observations
R-squared

2,138,976
0.379
0.066

2,398,341
0.407
0.071

2,192,322
0.381
0.053

2,694,7120
0.371
0.069

Source: Busso and Levy (2016).

11

These ordinary least square regressions capture the log of firm i productivity in sector s over the log of average
productivity of all firms in that sector (i.e., we only compare firms in the same sector). Regressions include controls
for firm size and age and, without implying causality; show that some firm types are systematically more productive
than others. Standard errors are in brackets. All coefficients are significant at the 99 percent level; see Busso and
Levy (2016), who also perform these regressions separately for each size group and report similar results, but with
larger productivity differences among smaller firms.

18

4.4 Evolution of Firm Size and Type between 1998 and 2013
Table 5 presents data on the size and type distribution of firms over the 1998-2013 period.
Notice first that average firm size is constant. But despite this constancy, there are large
compositional changes: the number of small and very small firms increases more than that of
medium-size and large firms. In parallel, the number of formal firms falls (by over 12 percent)
while that of informal firms rises quite substantially, by 66 percent.
Table 5. Size and Type Distribution of Firms and Employment, 1998-2013
(Thousands)
Totals
Firms
Employment
Average firm size
Firms by size
[0-5]
[6-10]
[11-50]
[51+]
Firms by type
Formal*
Informal**
Employment by firm size
[0-5]
[6-10]
[11-50]
[51+]
Employment by firm
type
Formal*
Informal**

1998

...