Upload
nockkk
View
232
Download
0
Embed Size (px)
Citation preview
1
Factors that influence Women’s Economic Participation in Mexico
ABSTRACT
This paper analyses women’s economic participation (WEP) in Mexico. The paper tests an econometric model
about the influence of various socio-economic factors on WEP. The results show that the main barriers to
female participation are the lack of education and the absence of diverse work possibilities in industries such
as manufacturing and hoteling. Other factors mentioned on the literature that also affect the WEP were number
of children per household, population density, the Gini coefficient and poverty on each municipality.
Therefore, these variables contribute on increasing WEP and should be the focus of any attempt to increase
their labour force participation in the formal sector.
Keywords:
Gender Economics, Urbanization, Demographic Economics, Geographic Labor Mobility, Demographic
Trends.
Classification JEL: J1; J10; J16, J61, J11
RESUMEN Este artículo analiza la participación económica de las mujeres (PEM) en México. El artículo pone a prueba un
modelo econométrico sobre la influencia de diversos factores socioeconómicos en la PEM. Los resultados
muestran que los principales obstáculos para la participación de las mujeres son la falta de educación y de
otras posibilidades de trabajo en industrias como la manufacturera y la hotelera Otros factores mencionados en
la literatura que también influyen en la PEM son el número de niños por hogar, la densidad poblacional, el
coeficiente de Gini y la pobreza en cada municipio. Por lo tanto, estas variables contribuyen a incrementar la
PEM y deben ser consideradas para aumentar su participación en la fuerza laboral en el sector formal.
Palabras clave:
Economía de Género, Urbanización, Economía Demográfica, Movilidad Geográfica de la mano de obra,
Tendencias Demográficas
Clasificación JEL:J1; J10; J16, J61, J11
Rafael Garduño Rivera
División de Economía-Sede Región Centro
Centro de Investigación y Docencia Económicas, A.C. (CIDE)
Tecnopolo Norte s/n, Parque Industrial Tecnopolo II,
Col. Hacienda Nueva
Aguascalientes, Ags. 20313
Email: [email protected]
2
1 INTRODUCTION
Since the 1950's, the social and economic importance of women has been increasingly accepted
worldwide (Mehra 1997). In a global context, studies such as Klasen and Lamana (2009) show a
general increase in the economic participation rates of women over the analyzed period. In
Mexico the first claims on gender equality occurred in the 1970's, where several feminist groups
were created to support the economic, educative and social aspects of the equality movement
(Bartra, Fernández Poncela y Lau 2002). The women's economic participation (WEP) in the
country has increased from 32.9% in 1987 to 41% in the last quarter of 2010. This reflects the
economic situation of Mexican households and some degree of women's empowerment (INEGI
2010). In this study I examine the demographic and economic factors affecting the WEP in
Mexico, and its evolution over a twenty year period (from 1990 to 2010). I am specifically
concerned with this topic from an economic growth and, consequently, poverty alleviation
perspective (Booz & Company 2012, World Bank 2012).
Female participation in the labor market has been highly studied because of the economic and
social implications. An important focus from recent literature has been the gap between male
and female schooling levels, and its repercussion on women´s ability to find jobs. Barro (1996)
using around 100 countries, estimates that the effect of higher education for women on economic
growth was practically zero. Meanwhile, more recent studies indicate that the effect of higher
education for women leads to higher WEP rates. Among the latest studies on this topic Klasen
and Lamanna (2009) stands out. They found that for a wide temporal range and including
information from over 130 countries, the effect of educative discrimination is determinant on the
employment opportunities for women and, indirectly, on growth patterns. This result agrees with
those found by Knowles, Lorgelly and Owens (2002) and Aguayo and Lamelas (2011). Aguayo
and Lamelas (2011) study Mexico and find that for the 2000-2005 period, education plays a
decisive role on WEP, although it will not estimate the female occupational levels.
3
In classical and modern literature, having children is seen as a factor that facilitates or hinders
WEP, depending on the country. Schockaert (2005) demonstrates the divergent effect of children
over WEP. Evidence for North-American women’s conditions allow them to conciliate their
reproductive and professional spheres while children in Latin-American countries represent an
obstacle to getting a (better) job or obtaining a higher wage for women. For this perspective,
Crespo Garrido (2012) studies the different public policies taken from a sample of European
countries to reconcile women’s family and professional lives. Meanwhile, García and de
Oliveira (1997) express that motherhood is seen as "their main source of identity" by an
important segment of society, varying between social classes.
Marital status has traditionally been highly related to women's employment rates. In a recent
study, Grantham (2012) concludes that even for mid-nineteenth century France, marital status
mattered for women entering the labor market, as reflected in the divergent signs associated with
the coefficients of married and widowed women. Anderson and Dimon (1998) provide a detailed
analysis of the economic behavior of married women, demonstrating that the earnings of the
husband or other members whom contribute to familial income are negatively related to the
probability of women joining the labor market.
Meanwhile, literature has not been focused on the barriers associated with women's entry into
the labor market, but instead, on the conditions women have to deal with once they get a job.
England et al. (2006) present evidence indicating that the feminization of occupations lowers the
wages of these positions; they conclude that the initial composition of the workforce by gender
was a determinant for this phenomenon, and that it has prevailed due to the “inertial relationship
between highly female occupations and low wages”. In contrast, Kriesi et al. (2010) report that
even for highly developed countries such as Switzerland, gender plays a fundamental role in job
opportunities, and therefore wage levels. Besides the characteristic of lower wages associated
with female employments, the lack of social security puts the familial economy at a higher risk.
Dominguez-Villalobos and Brown-Grossman (2010) demonstrate that the export-oriented
4
industries have contributed to lower wages in absolute and relative terms with respect to
women’s wages relative to those of men's.
Another factor that restricts the WEP for Mexican women is the machismo. Although machismo
is not reinforced by any law, as in Islamic countries, its effects prevail even after 40 years of
feminist demands (Sidani 2005). Contreras and Plaza (2010) argue that for the Chilean case,
machismo represents an obstacle for WEP, leading them to occupy the last places on this
category among OECD countries. Cerruti and Zenteno (2000) show that, for the Mexican case,
the probability of women’s joining the labor market decreased if they are living with a partner
and he is the household head.
While a significant aspect of the Mexican economy, this paper does not discuss women
participation in the informal sector. It neither discusses in depth the gender gaps in employment
and wages. Proof of those gender gaps has been carefully researched in other studies1.
This paper looks at the different factors that can affect women’s participation and explains the
disparities of WEP among Mexican regions. The hypothesis is that the regional disparities of
women’s participation are based on education, presence of maquiladoras, urbanization and other
non-observed factors. The questions that this study addresses are: What regional factors
influence women’s participation and cause disparities in their participation across
municipalities? Additionally, this paper measures the progress and the current state of WEP
during the last decades and the evolution of the differences across regions.
Women are the largest social group in Mexico (51.17% of population by 2010 census), and are a
potential source of labour for national economic growth (Booz & Company 2012). Thus, a better
understanding of the causes that increase WEP will help to identify measures to foster a greater
inclusion in the labour market and construct a more dynamic economy.
1 i.e. (Duryea, et al. 2007)
5
2 WOMEN’S SITUATION IN MEXICO
The situation of women in Mexico has shown promises of transformation during the last 30
years. This section analyses aspects such as fecundity, economic and work environment,
government awareness and women’s economic participation to observe if they had suffered a
transformation too.
2.1 Fertility
Between 1970 and 2010 an important variation in the demographic growth was found. This
change refers to the fecundity in women aged 12 and over. INEGI (2010) shows that in 2009 the
number of children per woman was 2.39 while in 1970 was 3.1; a reduction of almost one child
per women during the last 40 years. The main factors that INEGI identify as causes of this
variation are the level of education and economic conditions. They believe that the higher the
level of education the lower the average number of children.
Despite these fecundity changes, INEGI (2010) noted that by the end of 1990, the lower
fecundity level was not reflected in a more equitable percentage of women and men participating
in the work place –326 men were working for every 100 women—.This effect was only
perceived starting in 2009, when the ratio of economic participation among women and men was
151 men working for every 100 women, a reduction of 115% from the 1990 levels.
2.2 Recession and Crisis
Since1970, Mexico engaged in a process for liberalising and opening up its economy. This was
in response to the downturn and recession that hit Mexico, like many Latin American countries,
during the eighties. The deterioration of living standards made it increasingly necessary for the
population –especially the ones with low incomes— to look for new and different strategies to
sustain their household income levels. Levine (1993) explains that the 1982 crisis led more
family members into the work force, which meant school-age boys looking for after-school
employment and housewives looking for jobs that could be done at home. Women were obliged
to participate in order to contribute to the family budget along with continuing domestic work.
6
Furthermore, the growth of the informal sector was a result of the formal sector’s stagnation.
The former was characterised by high women’s participation. For his part, Campos-Vazquez
(2010) analyses the effects of macroeconomic shocks on employment during the lasts crises,
finding that most vulnerable were young and unskilled populations, but that women’s
participation seems not to be affected by previous crisis. Using a twenty years database, from
1990 to 2010, this paper analyses the variations of our data on three major crises (1994, 2001
and 2008), and the effect of the crises on it.
2.3 Type of Work
With respect to the type of work undertaken by women, INEGI (2009) shows that it was mostly
concentrated on health and educative services. Both types of work are tightly correlated with
demographic density, indicating that the higher the population, the higher the number of women
economically active. Among the traditional activities that have characterized WEP are working-
at-home, in maquiladoras and in restaurant, hotel and health services. The work undertaken by
women at home or in the factory was low remunerated due to the low skill requirements and did
not significantly contribute to household income. In relation to this information, ILO (2009)
reports that the expected balance after financial crisis will be an increase in participation of
women in economic life, in more vulnerable jobs, not resulting necessarily in significant
increases in the monetary income contributions.
2.4 Governmental Programs
Since 1970 women played a far greater role in the nation's economy. After the Women’s
International Conference in Mexico in 1975, many sectors and social organizations started
focusing on gender issues (Mehra 1997). INEGI (1994) called this time the “decade of women’s
incorporation to development”. Governmental programs and policies were created to open social
environments and foster women’s participation in the labor market. A high number of women
joined the educational system and the increase in opportunities resulted in education, health and
welfare services gains (INEGI 1994).
7
Moreover, since 2000 the gap on health and childcare services for women was reduced through
the creation of the social security program "Seguro Popular" and, in 2007, the childbearing
program "Estancias infantiles", whose purpose is to improve women’s labor conditions by
diminishing the associated cost implied by requiring child care.
2.5 Women’s Age and Participation
Some of the studies consider age as one of the main factors that affect the WEP. Age is used as a
life cycle indicator because it is related to family responsibilities and with female participation in
the labour market. Figure 1 shows the gender participation rates in Mexico by age in 2009.
Figure 1: Gender Participation Rates by age, 2009
Source: author’s calculation using INEGI´s Encuesta Nacional de Empleo Urbano (1999) data.
As observed, WEP never reaches the same level as men’s participation. The smallest gap
between male and female is in the age group of 12-14. In this group (12-14) the gap is lower due
to both genders mostly being in school. Male participation is already higher since they tend to
6.5
41.2
78.0
93.2 95.8 96.1 95.8 95.0 91.7
86.1
71.0
59.3
48.8
39.1
28.4
19.3
1.9
16.3
38.6
47.0 46.5 47.5 47.8 45.0
38.3
29.9
20.7
14.2 9.6
6.7 4.3 2.6
0.0
10.0
20.0
30.0
40.0
50.0
60.0
70.0
80.0
90.0
100.0
Men Women
8
leave school sooner than women to start working2. The highest levels of WEP are within the 20 -
49 years group. At these ages most of the women are co-habiting and raising their family;
therefore, during this period they keep working while they care and educate their children.
However, Pacheco and Parker (2001) establish that woman participation within informal market
have risen since 1987, especially for older women, increasingly due to the lack of social security.
It is also noticeable how economic participation changes when I control for age. First, both sexes
seem to behave in a similar way, showing an accented increase in economic participation before
the age of 29, which coincides with this education status. Subsequently, from 30 to 50 years old
there is a stationary stage on economic participation characterized by economic participation
rates around 95% for males and 45% for females. After finishing this stage, a decreasing phase
takes place, presenting a more marked downturn trend in the female’s case.
2.6 Evolution of Women’s Economic Participation
García and Oliveira (1994) mention that, during 1950, WEP of women 12 years old and higher
was 13%3. In 1970 the WEP increased to 16% and in the 1979 their participation had reached
21%4. By 1987, INEGI (2000) shows that the WEP had rose to 32.9%; and it increased
constantly until reaching 39.5% at the end of the second quarter of 2000, and, by 2010 up to
41%. In comparison with the male contribution there is still a large difference; just in the last
quarter of 2010, Male Economic Participation is situated at 76.3%. Men participation has always
been high and without much variation. Oliveira and García (1990) consider that men’s
participation has been reduced due to longer permanence of men at school and the improvement
of retirement possibilities.
WEP has kept rising. During the Tequila Crisis the WEP’s participation increased (1995, 38%)
even though the level did lower in 1994. This effect is similar to the one after 1982, where the
2 García and Oliveira (1994) show, in their analysis of 1982 to 1987, that there was an increase of the population
with incomplete schooling or primary school education. They attribute this increase to the crisis, which caused
students to leave school to work especially to the population with low incomes. 3 Based on the General Population Census (Censo General de Poblacion).
4 Based in ECSO (Occupational Continuous Inquiry).
9
WEP increased even in the rural areas. The re-appearance of the effect marks the most important
socio-demographic and economic processes that happened during the last decades in Mexico: a
decrease in fertility, an increase of women’s level of schooling, and the worsening of the crisis
(García y de Oliveira 1994).
2.7 Regional Women’s Economic Participation
In 1980, the metropolitan areas in the centre (Mexico City [D.F. and part of State of Mexico] and
Guadalajara [Jalisco]) was reaffirmed as the area of high economic expansion and consequently
high presence of both women’s and men’s economic participation (de Oliveira y García 1990,
351). The eighties witnessed an increase in the number of maquiladoras, especially in the
northern cities of Chihuahua and Ciudad Juarez, in the state of Chihuahua; Tampico and
Matamoros, in Tamaulipas; and Torreon, in Coahuila. In the central region, the cities that
reported the increases in WEP were Guadalajara, in Jalisco; and León, Celaya, and Irapuato in
Guanajuato (García y de Oliveira 1994, de Oliveira y García 1990). Rendón (1990) reports that
the increase in WEP was concentrated on low skill workers, which coincides with the increase in
maquiladoras—due to maquiladoras hiring a large number of female workers. However, it is
important to note that the growth in the WEP in the state of Quintana Roo, almost 20%, from
11% to 30%, as a result of the growth of tourism5. Surprisingly, the states of Guerrero, Oaxaca
and Chiapas, which present a low WEP most of the years before and after, occupied the second,
third and fourth place in this year, respectively6.
In 1990, the effect of the opening of the economy in 1986 and the rise in the activity of
maquiladoras is noticeable in the northern states (Coubes 2003). This explains why the five
states with the highest WEP, except D.F. (30.66%) are northern states at the border (Baja
California [BC] 27.4 %, Nuevo Leon 23.64 %, Chihuahua 23.49%, Baja California Sur [BCS]
5 In terms of foreign exchange earnings, Tourism often ranks third in importance behind petroleum and
manufacturing. Today, Cancun, Quintana Roo is one of the most frequented tourist destinations in Mexico. 6 The most probable explanation is that the 1980 census overestimated the WEP as it also included many women in
domestic activities in rural areas (B. García 1988).
10
22.59%, and Tamaulipas 22.27 %). Nevertheless, the effect vanishes at municipal level as we
move away from the US-Mexico border line. For the case of Quintana Roo, a WEP of 22.95%
shows that the effect of the high investment in tourism is still apparent. Baja California Sur did
not benefit from the maquiladora sector, it experienced an increase in the social and communal
services (from 3,332 women employed in 1980 to 5,753 in 1990), commerce (from 2,031 to
4,469), and restaurants and hotel services (from 781 to 2,675). To observe how spatial patterns
of economic activity have evolved in the different regions, I utilize the WEP at the municipal
level, which is the share of women economically active on the total female population for
municipality i. Figure 2 demonstrates the positive effect being situated near the border with the
principal commercial partner of México (the U.S.) had on WEP.
Figure 2: WEP percentage increase (1990-2010)
Source: author’s calculation using INEGI´s XI Censo General de Población y Vivienda data, 1990 and 2010.
In 1996, regional disparities prevailed and even increased. Although the overall WEP in Mexico
is still growing —mainly because of the states with maquiladoras— the differences persist. The
northern states and the three main cities continued with the highest WEP. Moreover, the spill-
over effect seems to be lost among the states around Jalisco and D.F., and increased in the
Border States. In addition, the D.F., which has been the region with more participation during
11
the three decades (70s, 80s, and 90s), lost its position and Baja California Sur (BCS) and Nayarit
have risen above it. The cause of these results could be the industrial restructuring and relocation
that Sanchez-Reaza (2000) finds:
Firms facing foreign competition, plausible over-crowding effects in Mexico City and the opportunity
to export to the larger market of the US, could be deciding to move away from the City. Some firms
could be relocating in the northern export-oriented Border States. Some others in need of being close
to the largest domestic market and to the federal government could only be moving away to the
neighbouring Estado de Mexico or to nearby states such as Tlaxcala, Puebla, and Queretaro . (pg. 7)
The year 2000 presented a complicated scenario for most countries, especially for those highly
involved in the world's economy, such as Mexico. The two major crises that Mexico suffered in
such a short period put serious restrictions on employment generation. The burst of the dotcom
bubble in 2001 indirectly affected the Mexican economy, causing a contraction of 5 percentage
points of GDP in the first trimester of 2001. Despite this decrease on GDP, Mexico has shown a
revitalized economy since the fourth trimester of 2001, increasing the speed of GDP growth and
the financial stability of the country.
Furthermore, the Mexican political scene changed when the Partido Acción Nacional candidate
Vicente Fox Quesada won the 2000 elections. The government transition offered a redefinition
of the strategy used until that moment for poverty alleviation, this one with a stronger gender
based component to accelerate the inclusion of women in the Mexican economy.
After a growth period, the strong connection between Mexico and the U.S. caused the Mexican
economy to decline due to the effects of the 2009 financial crisis. The crisis especially hurt the
most vulnerable sectors of the society, among them women. The 10 percentage points decrease
on GDP during the first trimester of 2009 was one of the strongest shocks in recent history and
was accompanied by a stagnation of export oriented manufacturing enterprises.
With the world's most developed economies stagnated, Mexico showed a reactivated economy
since June 2009, allowing for the creation of new firms and the expansion of existing ones.
Several counter cyclic measures were created aiming to alleviate the effect of the shock, and
especially, the poverty caused by the financial crisis that affected principally women. Moreover,
12
insecurity created an inhospitable environment for attracting investment. The "war against
drugs", started in 2006 by the former Mexican President Felipe Calderón, increased the cost
associated with establishing an enterprise in Mexico, which affected also the WEP.
2010 shows a similar pattern to the observed one in 1990. WEP still stood higher in the northern
states, especially Baja California Norte and Baja California Sur. This could be due to a
combined event: on one hand, both States are close to the U.S., and this proximity makes them
attractive to new investments; while on the other hand, these states are renowned for being some
of the most preferred for international and domestic tourists. Jalisco, Guanajuato, Estado de
Mexico and D.F. present a high WEP associated with overall economic concentration.
After two decades of structural changes, political alternation and economic crises, Mexico shows
a similar picture for women´s incorporation to formal economic market to the one presented in
1990. In Figure 2 we appreciate an interesting fact: municipalities with a high WEP in 1990
presented a higher growth rate of WEP compared to municipalities with a small WEP in 1990. In
other words, initial WEP has effects on the speed at which women are included in the formal
labor sector.
3 EMPIRICAL MODEL
Based on the theory and the regional differences previously showed, this section presents a
model to observe the factors that influence WEP. A twenty years period will be analysed in this
model, from 1990 to 2010 with decennial gaps. This allows for observation of the effects of the
variables over the WEP in a wide temporal frame. It will account for the different political and
economic contexts in those years, including the changing political scene. The model to be used
is the following:
where i indexes municipalities and t indexes time (1990, 2000, and 2010). Let WEP denote the
share of women economically active on the total female population for municipality i at time t.
13
Let denote a vector of time-constant municipality-specific variables that influence the output
per worker, be a vector of time-varying explanatory variables and is the idiosyncratic
error.7 Overall, there are 7,131 observations related to all the 2,377 Mexican municipalities and 3
years (1990, 2000, and 2010). Table 1 shows variable definition and summary statistics.
Between the 1989 and 2004 censuses, 48 new municipalities were created by splitting some of
the old municipalities. To analyse the same municipalities through the years, I merged the new
municipalities back to their 1988 boundaries.8
In the following part I explain the different variables used in the model and their importance for
testing our hypothesis. I also describe the method used to construct each variable9.
Table 1 Variable Definitions and Summary Statistics for each of the years analysed Variable Definition 1990 2000 2010
n Observations 2377 2377 2377
WEP % of economically active women 0.1050 0.2202 0.2302
(0.066) (0.0947) (0.0962) [0 , 0.7788] [0.0164 , 0.8088] [0.0174 , 0.6691]
WoPrim % of women with no formal education 0.2347 0.1855 0.1299
(0.1430) (0.1178) (0.0888) [0 , 0.914] [0.016 , 0.7858] [0.007 , 0.5950]
Incprim % of women with not completed primary school 0.3263 0.2773 0.2152
(0.0856) (0.079) (0.0676)
[0.0023 , 0.76] [0.0215 , 0.6627] [0.033 , 0.5178]
Comprim % of women with completed primary school 0.2067 0.2301 0.2069
(0.0689) (0.0602) (0.0491)
[0.0086 , 0.7247] [0.0412 , 0.6774] [0.0668 , 0.7011]
Comsec % of women with completed secondary school 0.1297 0.1904 0.0125
(0.0683) (0.0692) (0.0158)
[0 , 0.3633] [0.0026 , 0.5212] [0 , 0.1387]
Comhigh % of women with high school education 0.0742 0.1065 0.1829
(0.0708) (0.0879) (0.103)
[0 , 0.5269] [0 , 0.6172] [0.0023 , 0.7426]
whh % of women head of the household 0.0970 0.1208 0.1437
(0.0322) (0.035) (0.0361)
[0.0126 , 0.371] [0.0099 , 0.3387] [0.0277 , 0.3439]
chxhh average number of children per Household 1.7500 1.4100 1.0200
(0.36) (0.37) (0.28) [0.54 , 3.44] [0.37 , 5.71] [0.29 , 3.8]
Wsep % of separated women 0.0161 0.0285 0.0409
(0.008) (0.0129) (0.0167) [0 , 0.0612] [0 , 0.0784] [0 , 0.1209]
Wdiv % of divorced women 0.0050 0.0054 0.0077
(0.0039) (0.0049) (0.0069)
[0 , 0.0438] [0 , 0.0553] [0 , 0.0617]
Wsing % of single women 0.3486 0.3338 0.3152
7 where affects all observations for time period t, affects all observations for cross-sectional
unit i, and affects only observation it. I assume ( ) ( )
8 I obtained the list of new municipalities and from where they were created (INEGI 2006). For those created from
more than one municipality, I calculate the percentage of how many people (or how much land) was taken from the
former municipalities [information provided by (SEGOB 2005)]. I then allocate the information (e.g. total labor,
total population, number maquiladoras) of the new municipality back to the former municipality based on those
percentages. 9 I want to emphasise that this study is based on secondary information which in some cases did not replicate the
effect of the agents analysed over the WEP.
14
(0.0468) (0.0401) (0.0375)
[0.1558 , 0.5399] [0.1711 , 0.5006] [0.1533 , 0.4602]
Wcoh % of women in cohabitation 0.0851 0.1027 0.1405
(0.0669) (0.067) (0.0681)
[0 , 0.6954] [0 , 0.6244] [0 , 0.5709]
Wmar % of married women 0.4759 0.4484 0.4125
(0.0693) (0.0731) (0.0777) [0.0148 , 0.7326] [0.0201 , 0.6617] [0.0082 , 0.62]
Wwid % of widowed women 0.0621 0.0787 0.0811
(0.0229) (0.0266) (0.0269)
[0.0122 , 0.2276] [0.0269 , 0.2826] [0.0236 , 0.3068]
ln(distance) Distance from the municipality to the nearest border crossing point 6.8 6.8 6.8
(0.88) (0.88) (0.88)
[0 , 7.77] [0 , 7.77] [0 , 7.77]
ln(density) Population density 3.75 3.85 3.90
(1.52) (1.6) (1.67)
[-1.56 , 9.81] [-1.84 , 9.78] [-1.86 , 9.69]
maq % of women working in the manufacturing sector 0.0216 0.0451 0.0259
(0.0381) (0.053) (0.055) [0 , 0.7458] [0 , 0.5717] [0 , 0.7153]
hotels % of women in the touristic sector 0.0048 0.0133 0.0128
(0.0067) (0.0112) (0.013)
[0 , 0.0918] [0 , 0.1063] [0 , 0.1826]
ln(income) Average income by municipality (2010 pesos) 2.05 2.15 2.31
(1.43) (1.27) (1.17)
[0 , 5.43] [0 , 7.66] [0 , 6.11]
food poverty % of total population in food poverty situation 0.3736 0.4439 0.3165
(0.1762) (0.2408) (0.1869)
[0.014 , 0.949] [0.016 , 0.968] [0.01 , 0.832]
Gini Inequality measured with the Gini coefficient 0.4010 0.4615 0.3739
(0.0367) (0.0691) (0.0485)
[0.138 , 0.617] [0.243 , 0.705] [0.252 , 0.565]
2000 dummy variable for year 2000 0 1 0
(0) (0) (0)
[0 , 0] [1 , 1] [0 , 0]
2010 dummy variable for year 2010 0.0000 0.0000 1.0000
(0) (0) (0) [0 , 0] [0 , 0] [1 , 1]
Note: Reported statistics are mean, (standard error), and [minimum, maximum] values
3.1 Dependent Variable
I use the share of female economically active population over 12 years old, this variable is
known by the INEGI as women’s PEA (Economically Active Population). I chose the years of
1990 until 2010, with decennial gaps, because these were the years in which the INEGI did the
most recent censuses. These years allow for an analysis of the changes from one decade to
another. In addition, these years have the most detailed regional data and permit an observation
of precise differences in WEP as well as in the right hand side variables.
3.2 Independent Variables
Base on the different theories, I selected the variables that most probably influence female labor
participation. For this model I first include the variables of education, women headed
households, maquiladoras, hotels service sector, population density, average income, the Gini
15
coefficient, level of poverty, number of children per household, marital status, year dummies and
road distance to the U.S. border.
3.2.1 Education
Many studies have been aimed at demonstrating the effect of education on the supply side of the
labor market. With regard to the Mexican case, Aguayo and Lamelas (2011) observed that from
2000 to 2005 the levels of education increased proving that employer preferences for young and
more educated employees had also grown.
The educational effect, and especially the difference between men and women, has part of its
roots in the past exclusion of women for educational opportunities. It is possible to observe these
differences when I analyze the gap between men and women in the average level of schooling10
.
While Chiapas, Guerrero and Oaxaca have the lowest levels in average education for women,
Distrito Federal, Sonora, Tamaulipas and Nuevo León present the highest levels. I can anticipate
a positive correlation between WEP and education, which could lead to stronger economic
development for the region.
I use four different variables to measure the level of education for municipality i at time t:
Incomplete Primary school (Incprim), Complete Primary school (Comprim) Complete
Secondary school11
(Comsec) and Complete High School and Higher (Comhigh) as a share of
the female population with 15 years and over; in order to capture the diverse impact of education
over the WEP. I omit the share of female without education, Without Primary (WoPrim), from
the model as the benchmark category.
3.2.2 Women Headed Households
Concerning the structure of the families and its impact on women's participation, I include the
variable of women headed household (whh). For this variable I use the share of women headed
households within a municipality. Despite the fact that the use of this variable has been highly
10
Author’s calculation using INEGI´s Encuesta Nacional de la Dinámica Demográfica (2009) data. 11
A.K.A. Elementary School in the U.S.
16
questioned due to the downward bias that results from the idiosyncrasy of the population of
Mexico and the lack of a formal definition at pooling time (INEGI 2005), I have used it to
represent the increasing role women have had in Mexican society.
Choosing whh as a variable to explain WEP follows the hypothesis that the greater the number
of households headed by women, the greater the women's participation in the formal economy.
This causality should arise because, first, many women in whhs stay single most of their
childbearing years (15-49),12
with the responsibility to sustain the house in terms of the domestic
and extra-domestic work. Second, in female-headed houses with adult men who are inflicted by
unemployment, incapacity, alcoholism, small wages, economic crisis or other factors cause
women to be the principal income provider (INEGI 1999).
3.2.3 Number of Children per Household and Marital Status
These two variables are present in practically all the literature aimed at analysing the WEP. The
appearance of both variables are due to the focal value they have: while children are seen as
‘barriers’ to extra-domestic job due to the higher opportunity cost associated with them, marital
status gives an estimate of the family members who contribute to the household income. When
we include the number of children (ages 12 and below) per household, we can expect a lower
economic participation of women (Schockaert 2005). Also of importance are the presence of
other members that help in domestic activities and the services available to support childcare13
(Anderson y Dimon 1998, Harkness y Waldfogel 1999). In order to observe the effect of
children on the WEP, this study considers the number of children per household (chxhh), which
is the average number of children (ages 12 and below) per household per municipality. I expect a
negative nexus between CA and WEP.
12
In 2002, 92.2% of women household heads were living without a partner. Predominantly widows (39.3%);
although there is a large group of single (16%), separated and divorce women (34.7%). In contrast, 95% of the men
headed households where living with their partner (INEGI 2005). 13
Unfortunately I could not obtain the necessary data to test Federal program "Estancias infantiles" implemented by
former Mexican President Felipe Calderón or the presence of other member on those households.
17
Another variable to consider is the marital status. For this, I use the share of separated (Wsep),
divorced (Wdiv), single women (Wsing), cohabiting (Wcoh) and married (Wmar) women;
omitting widowed (Wwid) as the benchmark category, since it is the category with less
participation in 2010. Table 2 shows the groups with more WEP during the last decades.
Theoretically, I expect that the higher the rate of separated, divorced and widowed women, the
higher women’s participation in the labor force; and the higher the rate of married women, the
lower the WEP.
Table 2: % of WEP for each Marital Status, 1990-2010
Marital Status 1990 2000 2010
Divorced 56.0% 67.6% 68.3%
Separated 43.6% 59.6% 61.8%
Single 24.7% 33.6% 32.6%
Total Women 19.2% 29.8% 32.5% Cohabitation 14.5% 25.9% 30.4%
Married 14.0% 25.1% 29.7%
Widowed 16.7% 22.5% 22.6%
Source: INEGI, (1990) (2000) (2010).
3.2.4 Distance to the U.S. border & Population Density
A time-constant municipality-specific variable was included in the model to account for a
portion of the unobserved heterogeneity in the data. Distance to the U.S. border ( ( ) ),
a continuous variable that reflects the road distance (in kilometers) from municipality ‘i’ to the
closest U.S. border crossing point. To create this variable, I first obtain the name of the
municipality capital14
(INEGI, 2008). Second, I calculate the road distance from each of the
municipalities to the different U.S. border crossing points, by entering the destination and origin
points in the webpage “Traza tu Ruta” provided by the Secretaría de Comunicaciones y
Transportes (2008). Finally, I chose the shortest distance for each municipality form the different
14
When one municipality includes more than one city or town, one of them is selected as cabecera municipal (head
city or seat of the municipal government).
18
distances provided by each U.S. border crossing point.15
I also include the population per sq.
kilometer in municipality i for period t, (
), ( ) .
I include the population density ( ( ) ) and the road distance to the nearest U.S. border-
crossing point ( ) as Market Size and Transportation Cost proxies, respectively. The
population density reflects the net centripetal and centrifugal force in the model since it
represents both market size and congestion costs. Market size may facilitate economies of scale
and induce a twofold attraction towards the market center in terms of backward and forward
linkages (Krugman, 1991). Distance to the U.S. border act as centrifugal force since it is related
to the transportation cost of the new accessible market (the United States). Therefore, I expect
that population density and distance to the U.S. border influence the WEP.
3.2.5 Maquiladoras
During the last decades, the economy in Mexico has significantly changed as a result of
structural adjustment and stabilisation policies—especially due to conditional programs
implemented by the IMF and World Bank in developing countries (Afshar y Barrientos 1999).
After the economic crisis of 1982, Mexico abandoned the model of government dependence in
favour of a more ‘neo-liberal’ model with a free market approach.
Chronologically, I can identify two periods. First, the Import Substitution Industrialisation (ISI)
(1930-1985) period, where both tariff and non-tariff barriers were used in order to promote
industrialisation. Now Mexico is using export lead growth approach, observed through the
numerous agreements that Mexico have signed with several countries (most notably is NAFTA).
These changes brought the creation and growth of the maquiladoras, which play a significant
role in increasing women’s participation rates caused by the high number of women employed in
rural and urban areas (Fontana and Wood 2000, Wilson 1991). It is necessary to emphasise that
15
For Municipality heads that do not appear as origin point, I calculate the distance of the nearest available city or
town and add the road distance from that point to the municipality head of interest, which I calculate manually by
using a map of Mexico.
19
this type of growth was the result of Mexico’s ‘competitive advantage’ in low skill labour, low
regulations and vicinity to the US market.
The model takes the share of workers occupied by the maquiladora sector in every single
municipality16
(maq) from 1990 until 2010. This study anticipates a positive correlation between
manufacturing presence and WEP as has been observed in diverse countries (i.e. Bangladesh)
due to female-intensive manufactured exports (Fontana and Wood 2000).
3.2.6 Hotels
Since 1970's, tourism has been highlighted as a focal point for the development of Mexico. The
fundamental role that tourism plays on employment is indispensable in explain the dynamics that
follows the population. Tourism is a control variable since there are certain states, such as
Quintana-Roo and Guerrero, with gross domestic product largely drawn from touristic activities
(Diaz-Briquets y Weintraub 1991). This sector has typically high levels of female employment,
due to the demands of the work. Consequently, the participation of women is especially high in
these states17
. To observe the effect of the touristic sector on the model, I include the share of
active women over 12 years old hired in the hotels service sector.
3.2.7 Labor Markets
To capture the effect of the labor market on the WEP, I include the remuneration per worker, the
Gini coefficient and the share of people living in food poverty. Remuneration per worker is
generated as total remuneration paid18
in a municipality divided by the number of workers
registered in that year for that region. The Gini coefficient and the share of people living in food
poverty19
are calculated by CONEVAL (2010) using the income information provided by
INEGI’s 1990, 2000 and 2010 Household and Population censuses.
16
INEGI. Sistema de Cuentas Nacionales de México; and INEGI (1998). 17
Author’s calculation using INEGI’s 2009 Economic Censuses 18
Remunerations are presented in real thousand pesos from 2003 19
Food poverty is defined by CONEVAL as the inability to obtain a basic food basket, even if you make use of all
disposable income at home just to buy goods from the basket.
20
3.2.8 Year Dummies
A set of year dummy variables were included in the regression. This accounts for socio-
economic factors which might have evolved over time (such as the opening of the Mexican
economy) and are not explained by the other independent variables included in this model.
4 EMPIRICAL RESULTS
Table 3 reports the regression results using panel data of 7,131 observations related to all 2,377
Mexican municipalities over three years (1990, 2000, and 2010). To test for geographically-
specific omitted variables, separate regressions on the same model were run using random
effects (RE; column 1) and another regression using fixed effects (FE; column 2), which
considers the unobserved characteristics of individual municipalities. A test for endogeneity was
conducted by including the individual means of all time varying variables in the RE regression
(column 3) and performed a Wald test on joint significance of the coefficients of mean input
variables (Table 4). Since the joint test rejects the null at p-value = 0.0000, I conclude that the
RE is inconsistent, as it is not capturing an important part of the average individual municipal
effects. As a result of these findings, I based my results in the FE model (column 2). However
the FE model dropped the time invariant variables (distance to the U.S.). To obtain this
coefficient I did a matrix calculation to get the residuals from the FE regression and regressed
the residuals on the time invariant variable (with no constant).20
This approach gave the
coefficients for these variables as reported in Table 3 column 4.
Our findings in education coincide with those expressed in previous studies: the greater the years
in the classroom, the higher the probability of joining the formal labor market. In fact, I found
that complete secondary school (Comsec) is negatively related with WEP, while incomplete
primary and high school (Incprim and Comhigh, respectively) are positively related with WEP.
All four variables, as a group, reflect the same fact: increasing access to school for women, and,
20
See section 11.4, Hausman and Taylor-Type Models (Wooldridge, 2001, p. 325)
21
as a consequence, school years of study will result in an increase of WEP, which could lead to
economic growth and poverty alleviation.
The results presented in Table 3 shows that women headed households (whh) do not
significantly influence women's participation. This is partly due to many women household
heads receiving important transfers and grants (more than 25%) from their husbands (when they
are separated), parents, and/or brothers. It is necessary to underline that 15% of them do not
receive incomes (INEGI, 99, pg. XIV).
All four coefficients related to marital status confirm my hypotheses. The coefficient associated
with divorced, single and married women (Wdiv, Wsing and WMar, respectively) is negative and
significant at 5% level, indicating that being divorced, single or married reduces the probability
of entering the formal labor market in comparison with the benchmark category: divorced
women (Wdiv). And that separate and cohabiting woman participate the same as divorce women
in the formal labor market.
As expected, the presence of maquiladoras and hotels has a positive effect on WEP. These
results support the hypothesis that states that had high investment in these two sectors, and
consequently high demand of female labor, had a high WEP. It also confirms our previous
findings in Figure 2 that borderline states such as Baja California, Chihuahua, Coahuila, Nuevo
León, Sonora and Tamaulipas, which had been historically recognized for being the principal
destination of foreign direct investment (FDI) attract more WEP. WEP follows a similar
dynamic to that presented by industrial concentration in Mexican regions (Gómez-Zaldívar and
Ventosa-Santaulària 2009, Baylis, Garduño-Rivera and Piras 2012). Regional disparities persist
for WEP and lead women to enter the labor market only to those regions that manage to attract
investment, especially in the maquiladora and hotel sector.
The coefficient of population density ( ( )) proves to be significant at 5% level and
positive. The result shows that the WEP is linked to the population —in terms of inhabitants per
22
km2—. This conforms that the more densely populated region, the more women participation in
the formal labor market.
Within the labor markets’ variables: The coefficient of remuneration per worker was not
significant, indicating that the impact of the remuneration per worker did not influence the WEP.
But the coefficients of the Gini and share of people living in food poverty are positive
significant, indicating that the bigger the income disparity and the number of people in food
poverty in a region the more women participation in the formal labor market on that region.
The coefficient of the number of children per household (chxhh) is positive and significant;
indicating that, on average, one extra child in a household increases the WEP in 2%. I attribute
this effect on the economic performance in Mexico during the last two decades and, particularly,
to the loss of purchasing power that forced any family member, with the capacity to work, to
find a job in order to contribute to the family income. Since 1980 the purchasing power in
Mexico had been reduced in 65% (see Figure 3), which compelled Mexican families to increase
the number of members who contribute to the family’s income, especially as the number of
household members, without the capacity to work, increase.
Figure 3: Real Wage by Year, base year 2003
Source: author’s calculation using SAT’s (2011) data.
Furthermore, the program "Estancias infantiles" implemented by the former President Felipe
Calderón, seems to have reduced the barrier between having children and WEP. According to
0.00
20.00
40.00
60.00
80.00
100.00
120.00
1980 1985 1990 1995 2000 2005 2010
23
governmental information, by the end of 2012 there were 9,473 child care centers, benefiting
over 270 thousand women. These child care cost allow women to get a job and contribute to
familial income (Presidencia de la República 2013).
Different from expected, the coefficient of the distance to the U.S. border was not significant.
This indicates that the importance for a high WEP in a region does not rest on how close this
region is to the border with the U.S. market but on the investment in the maquiladora and hotel
sector.
Finally, we observed that the coefficients of the year dummies were both significant, indicating
that the pattern of WEP has increased in 2000 and 2010 relative to 1990. This confirms that
socioeconomic factors, different from the ones included in the model (i.e. the opening of the
Mexican economy), have evolved over time and influenced a higher WEP.
5 CONCLUSIONS
This paper has analysed the factors that influence women’s economic participation in Mexico’s
economy. In order to observe the effect of the variables the analysis has focused on a period of
twenty years, from 1990 to 2010. The model considered the rate of the economically active
female population within a municipality as a dependent variable. From the results of this model,
I found evidence that supports our hypotheses: First, I found that women with complete
secondary studies have a negative relationship with WEP; whereas women who have completed
their primary and high school studies have a positive propensity to enter the formal labor market.
I consider the negative sign on women with complete secondary studies is due to the lack of
skills of the population with these characteristics.
Second, there is not a significant relationship between women headed households and WEP,
indicating that women in this group are likely to be in the labor market when they are the family
head as well as when they are not. Hence, an increase in the share of woman headed households
is not reflected in a higher rate of women’s economic participation.
24
Third, I found that the poorer the community the more likely it is that women will participate.
We also observed that the effect of population density on the WEP has proved to be highly
significant and positive. Further research is required to shed light on the exact link between
population density and female participation. A likely causality arises from increased work
opportunities in a densely populated setting, which include more possibilities of work for
women.
Regarding the control variables, maquiladoras and hotels were both positive and highly
significant. The behaviour of both variables seems to respond to the process of economic
openness and public policies created to increase FDI in Mexico. This means that the attraction of
foreign or national investors will result in higher demand for female labor in the formal sector.
Sustained public policies seem to be determinant for the increased role that they play over the
creation of manufacturing and touristic enterprises and, indirectly, female economic
incorporation.
Summarising the results, I observe that the main barriers to female participation are a lack of
education and undiversified possibilities of work in the less dense section of the economy. This
study reinforces the results obtained by Rendón (1990)and García and Oliveira (1994), namely
that crises, higher education, population density and manufacturing create more employment
opportunities for women.
Thus far, this study has analysed the different variables that affect women’s labor participation
and increasing regional employment disparities. However, it is necessary to remember that this
study did not consider some factors which are mentioned in the literature, such as household size
and childcare responsibilities. Moreover, the use of secondary information could have decreased
the effect of some important variables that would have shown its weight and giving more precise
data.
6 APPENDIX
Table 3: WEP Regressions
25
(1) (2) (3) (4)
RE FE RE_means Residuals
Incprim 0.00581 0.0754** 0.0754**
(0.42) (3.03) (3.03)
Comprim 0.0758*** 0.00520 0.00520
(5.07) (0.16) (0.16)
Comsec -0.158*** -0.211*** -0.211***
(-8.40) (-7.56) (-7.56)
Comhigh 0.364*** 0.271*** 0.271***
(17.66) (6.09) (6.09)
Whh 0.0894** 0.0492 0.0492
(2.67) (0.97) (0.97)
Wsep -0.130 -0.0674 -0.0674
(-1.60) (-0.58) (-0.58)
Wdiv 1.065*** -0.669* -0.669*
(5.08) (-2.25) (-2.25)
Wsing -0.243*** -0.202** -0.202**
(-5.43) (-2.69) (-2.69)
Wcoh -0.315*** -0.151 -0.151
(-6.76) (-1.94) (-1.94)
Wmar -0.318*** -0.219** -0.219**
(-7.10) (-2.92) (-2.92)
Maq 0.704*** 0.501*** 0.501***
(46.70) (21.06) (21.06)
hotels 1.201*** 0.643*** 0.643***
(14.92) (5.22) (5.22)
ln(density) 0.00818*** 0.00866* 0.00866*
(12.39) (2.23) (2.23)
Ln(income) 0.000109 0.00188 0.00188
(0.14) (1.61) (1.61)
gini 0.0779*** 0.0787*** 0.0787***
(5.05) (4.30) (4.30)
food poverty -0.0160* 0.0338** 0.0338**
(-2.19) (3.18) (3.18)
chxhh -0.000842 0.0223*** 0.0223***
(-0.26) (4.29) (4.29)
ln(distance) -0.000358 0 0.00189 -0.0000825
(-0.33) (.) (1.70) (-0.88)
y2000 0.0725*** 0.0986*** 0.0986***
(26.86) (21.64) (21.64)
y2010 0.0397*** 0.0797*** 0.0797***
(9.13) (11.21) (11.21)
avg_ food poverty -0.0906***
(-5.71)
avg_gini 0.0202
(0.61)
avg_ chxhh -0.0263***
(-3.77)
avg_Wmar -0.166
(-1.69)
avg_Wcohab -0.218*
(-2.13)
avg_Wdiv 2.800***
(6.71)
avg_maq 0.317***
(10.42)
avg_Comhigh -0.0201
(-0.39)
avg_whh -0.0566
(-0.80)
avg_Comprim 0.0355
(0.98)
avg_Incomprim -0.103***
(-3.41)
avg_Compsec 0.0413
(0.88)
avg_Wsep -0.0726
(-0.44)
26
avg_Wsing -0.0731
(-0.75)
avg_hotels 0.974***
(6.01)
avg_ln(density) -0.000815
(-0.21)
avg_ln(income) -0.00592***
(-3.74)
_cons 0.259*** 0.140* 0.318***
(6.04) (2.08) (5.21)
N 7131 7131 7131 7131
R-sq 0.672 0.000
adj. R-sq 0.506 -0.000
t statistics in parentheses. * p<0.05 **, p<0.01, *** p<0.001
Table 4: Wald test on joint significance of the coefficients of mean input variables . test avg_shareprimincom avg_shareprimcomp avg_shareseccomp avg_sharemediasup avg_sharemjf1
avg_shareseparadas avg_sharedivorciadas avg_sharesoltero avg_sharecohabitation avg_sharecasadas
avg_sharemanuf avg_host avg_ln_dens avg_ln_r_income avg_gini avg_alimentaria avg_niñosxhogar
>
( 1) avg_shareprimincom = 0
( 2) avg_shareprimcomp = 0
( 3) avg_shareseccomp = 0
( 4) avg_sharemediasup = 0
( 5) avg_sharemjf1 = 0
( 6) avg_shareseparadas = 0
( 7) avg_sharedivorciadas = 0
( 8) avg_sharesoltero = 0
( 9) avg_sharecohabitation = 0
(10) avg_sharecasadas = 0
(11) avg_sharemanuf = 0
(12) avg_host = 0
(13) avg_ln_dens = 0
(14) avg_ln_r_income = 0
(15) avg_gini = 0
(16) avg_alimentaria = 0
(17) avg_niñosxhogar = 0
chi2( 17) = 396.87
Prob > chi2 = 0.0000
7 Bibliography
Afshar, Haleh, and Stephanie Barrientos. Women, globalization and fragmentation in the developing world. St. Martin's Press, 1999.
Aguayo, Eva, and Nélida Lamelas. “Educación y empleo: desigualdad de género en las regiones mexicanas. 2000-2005.” Revista Estudos Feministas, 2011: 733-750.
Anderson, Joan B., and Denisse Dimon. “Married Women's Labor Force Participation in Developing Countries: The Case of Mexico.” Estudios Económicos Vol. 13, No. 1, 1998: 3-34.
Barro, Robert J. “Determinants of Economic Growth: A Cross-Country Empirical Study.” National Bureau of Economic Research, 1996.
Bartra, Eli, Anna M. Fernández Poncela, and Ana Lau. Feminismo en México, ayer y hoy. 2002. Baylis, Kathy, Rafael Garduño-Rivera, and Gianfranco Piras. "The distributional effects of NAFTA in Mexico: Evidence from a panel of
municipalities." Regional Science and Urban Economics, 2012: 286-302.
Booz & Company. “Empowering the Third Billion Womenand the World of Work in 2012.” 2012. Campos-Vazquez, Raymundo. “The effects of macroeconomic shocks on employment: the case of Mexico.” Estudios Económicos, 2010: 177-
246.
Cerrutti, Marcela, and René Zenteno. “Cambios en el papel económico de las mujeres entre las parejas mexicanas.” 2000.
27
CONEVAL. Consejo Nacional de Evaluación de la Política de Desarrollo Social. November 2010. http://www.coneval.gob.mx/ (accessed July
19, 2013).
Contreras, Dante, and Gonzalo Plaza. “Cultural Factors in Women's Labor Force Participation in Chile.” Feminist Economics 16, no. 2 (2010): 27-46.
Coubes, Marie-Laure. “Evolución del empleo fronterizo en los noventa: Efectos del TLCAN y de la devaluación sobre la estructura
ocupacional.” Frontera Norte julio-diciembre, año/vol. 15, número 030, 2003: 7-32. Crespo Garrido, María. “Economic development, family wealth and settlement of the personal life and work.” Comunitania International
Journal of Social Work and Social Sciences, 2012: 43-61.
de Oliveira, Orlandina, and Brígida García. “Expansión del trabajo femenino y transformación social en México: 1950-1987.” México en el umbral del milenio, 1990: 345-374.
Diaz-Briquets, Sergio, and Sidney Weintraub. Regional and Sectoral Development in Mexico as Alternatives to Migration. Vol. II. Oxford: West
view Press, 1991. Domínguez-Villalobos, Lilia, and Flor Brown-Grossman. “Trade Liberalization and Gender Wage Inequality in Mexico.” Feminist Economics
16, no. 4 (2010): 53-79.
Duryea, Suzanne, Sebastian Galiani, Hugo Nopo, and Claudia C. Piras. “The educational gender gap in Latin America and the Caribean.” Working Paper, IDB Research Department, Inter-American Development Bank, Washington, DC, 2007.
England, Paula, Paul Allison, and Yuxiao Wu. “Does bad pay cause occupations to feminize, Does feminization reduce pay, and How can we tell
with longitudinal data?” 2006. Fontana, Marzia, and Adrian Wood. "Modeling the Effects of Trade on Women, at Work and at Home." World Development 28, no. 7 (2000):
1173-1190.
García, Brígida. Desarrollo económico y absorcion de fuerza de trabajo en México: 1950-1980. México: El Colegio de México, 1988. García, Brigida, and Orlandina De Oliveira. “Motherhood and Extradomestic Work in Urban Mexico.” 1997: 367-384.
García, Brígida, and Orlandina de Oliveira. Trabajo femenino y vida familiar en México. El Colegio de México, 1994.
Gómez-Zaldívar, Manuel, and Daniel Ventosa-Santaulària. "Liberación comercial y convergencia regional del ingreso en México." Trimestre Económico LXXVI, no. 301 (2009): 212-235.
Grantham, George. “Occupational, Marital, and Life-Cycle Determinants of Women's Labor Force Participation in Mid Nineteenth-Century
Rural France.” Feminist Economics 18, no. 4 (2012): 97-119. Harkness, Susan, and Jane Waldfogel. “The family gap in pay: evidence from seven industrialised countries.” Centre for Analysis of Social
Exclusion, London, 1999.
INEGI. Censo de Población y Vivienda 2010. INEGI, 2010. —. Censos Económicos 2009. Aguascalientes, 2009.
—. Encuesta Nacional de Empleo Urbano, 1980-1997. 1999.
—. Encuesta Nacional de la Dinámica Demográfica (ENADID). 2009. http://www.inegi.org.mx/est/contenidos/proyectos/encuestas/hogares/especiales/enadid/default.aspx (accessed January 15, 2013).
—. Hombres y Mujeres en México. Fourth Edition. INEGI, 2000.
—. Industria Maquiladora de Exportación. Aguascalientes: INEGI, 1998.
—. La Participación de la Población en la Actividad Económica en Mexico. Aguascalientes: INEGI, 1994.
—. Los Hogares con Jefatura Femenina. Aguascalientes: INEGI, 1999.
—. Los Hogares con Jefatura Femenina. Aguascalientes, Ags, 2005. —. Principales resultados por localidad 2005 (ITER) from II Conteo de Población y Vivienda 2005. 2006.
http://www.inegi.gob.mx/est/contenidos/espanol/sistemas/conteo2005/localidad/iter/default.asp (accessed June 1, 2008).
—. XI Censo General de Población y Vivienda. INEGI, 1990. —. XII Censo General de Población y Vivienda. INEGI, 2010.
International Labour Office. Global Employment Trends for Women. Geneva: ILO, 2009.
Klasen, Stephan, and Francesca Lamanna. “The Impact of Gender Inequality in Education and Employment on Economic Growth: New Evidence for a Panel of Countries.” Feminist Economics, 2009: 91-132.
Kriesi, Irene, Marlis Buchmann, and Stefan Sacchi. “Variation in job opportunities for men and women in the Swiss labor market 1962-1989.”
Research in Social Stratification and Mobility 28, no. 3 (2010): 309-323. Levine, Sarah. Dolor y Alegria: women in social chane in urgan Mexico. Wisconsin: University of Wisconsin Press Madison, 1993.
Mehra, Rekha. “Women, Empowerment, and Economic Development.” Annals of the American Academy of Political and Social Science 554
(Nov. 1997): 136-149. Pacheco, Edith, and Susan Parker. “Movilidad en el mercado de trabajo urbano: evidencias longitudinales para dos periodos de crisis en
México.” Revista Mexicana de Sociología 63, no. 2 (2001): 3-26. Presidencia de la República. México: Presidencia de la República. 08 Enero 2013. http://www.presidencia.gob.mx/articulos-prensa/beneficios-
del-programa-estancias-infantiles/ (accessed Febrero 20, 2013).
Rendon, Teresa. “Trabajo Femenino remunerado en el siglo veinte. Cambios Tendencias y Perspectivas.” In Trabajo Femenino y Crisis en México. Transformaciones y Tendencias Actuales, by Elia Bautista, & Hilda R. Davila Ibañez. Xochimilco: Universidad Autónoma
Metropolitana, 1990.
Sanchez-Reaza, Javier. “The impact of trade policies on Mexican regional growth.” The London School of Economics and Political Sciences, Department of Geography, MIMEO, 2000.
Schockaert, Ingrid. “Travail féminin et fécondité en Amérique latine: Un état de la question.” Population Vol. 60, No. 1/2, 2005: 157-177.
SEGOB, Secretaría de Gobernación. Enciclopedia de los Municipios de México. Edited by Instituto Nacional para el Federalismo y el Desarrollo Municipal. 2005. http://www.e-local.gob.mx/wb2/ELOCAL/ELOC_Enciclopedia (accessed May 13, 2008).
Servicio de Administración Tributaria (SAT). Cuadro Histórico de los Salarios Mínimos (1982 - 2010). 2011 November 2011.
http://www.sat.gob.mx/sitio_internet/asistencia_contribuyente/informacion_frecuente/salarios_minimos/45_7369.html (accessed March 20, 2013).
Sidani, Yusuf. “Women, work, and Islam in Arab societies.” Women in Management Review, 2005: 498-512.
Wilson, Fiona. Sweaters: gender, class and workshop- based industry in Mexico. London: Macmillan, 1991. World Bank. The Effect of Women's Economic Power in Latin America and the Caribbean. 2012.