The objective of the paper is to analyze the effect of trade liberalization on the wage gap by gender and its components in Mexico. For this, the behavior of the wage gap is analyzed for different regions of the country, classified according to the degree of exposure to trade liberalization. Using the quantitative methodology of Melly (2005) the breakdown of the salary gap by quantile used the micro data of the National Survey of Occupation and Employment (ENOE) for the years 2005 and 2016. The results reveal the existence of wage gap by gender in the different regions, that its magnitude differs regionally throughout distribution and that the wage gap has declined over time in all regions of Mexico, but the wage disadvantage still persists against women and this is higher in the lower deciles. Most of the differences are not explained by productive characteristics, although the proportion of inequality associated with discrimination factors declined in 2016 in relation to the previous reference year.
In the last decades, Mexico has undergone great political, demographic and economic transformations, in the latter aspect one of the most relevant elements was the implementation of an economic model, characterized by a greater orientation towards the external sector, which materialized with the incorporation to the General Agreement on Tariffs and Trade (GATT) in 1986; The North American Free Trade Agreement (NAFTA) in 1994, and more than 20 more trade agreements in the following years.
30 years since the incorporation of Mexico into the GATT (now the World Trade Organization), major economic transformations have been generated, both spatial and sectoral, in the first case through the strengthening of geographical spaces that have advantages of localization in the new economic model and that can be defined as the region of high exposure, and in the second, the relative growth of activities that have achieved a better insertion in global value chains, such as the automotive sector and machinery and equipment in general within the manufacturing activities.
In addition to these spatial and sectoral changes, there have been changes in the Mexican labor market, one of the most important being the increasing participation of women, [
In terms of wage inequality or gender pay gap, there has been a great deal of research, both for Mexico and for other countries1, measuring the magnitude of the wage gap, its components, and relevant factors, but insufficient trade openness can have on the wage gap and its components.
The motivation of this study is to explore whether trade liberalization has impacted on the gender wage gap in Mexico.
Therefore, the objective of this research is to analyze the wage gap and its decomposition along the distribution of the different regions of Mexico, under the criterion of the degree of exposure to the commercial opening in the period 2005-2016. Applying the methodology of Melly [
The hypotheses to be tested are:
H1: The regions with the greatest exposure to trade liberalization have lower gender wage gaps relative to less exposed regions.
H2: Women in the highest deciles will have a lower wage gap by gender in relation to the lower deciles.
H3: The unexplained component of the wage gap, which is often associated with discrimination, will be lower in regions with higher exposure relative to less exposed regions.
H4: The above behaviors are expected to show a trend towards convergence, as well as the effects, are more pronounced in 2016 compared to 2005.
The theoretical arguments underlying the above hypotheses are 1) the degree of exposure of the regions to trade liberalization increases competition between firms and workers with the same characteristics, leading to a reduction in the wage gap between men and women; 2) increased competition imposes higher costs on discriminatory behavior; 3) higher female workers have higher incomes and have more information that leads them to take better job choices and reduce wage differentials in relation to women with lower human capital and lower incomes; effects of competition are not immediate but act over time.
To test the hypotheses, the wage gap and its decomposition along the wage distribution of the different regions of Mexico will be analyzed by applying the quantitative methodology of Melly [
The rest of the document is structured in four sections. The first presents the theoretical review on trade openness, inequality, and empirical evidence, the second presents the data and methodology, while the results are presented in section three and finalized with general conclusions and reflection.
Artecona et al. [
Discriminatory behaviors practiced by different agents, such as consumers and producers, are reflected in the theory of taste for discrimination developed by Becker [
Artecona et al. [
Therefore, assuming there is competition in the product market if a company does not discriminate and its rivals do, it would face higher production costs and end up losing market share or being expelled.
In summary, if we take Becker’s model of reference, trade liberalization increases competition and reduces the wage gap between men and women. To the extent that economies are more exposed to trade liberalization, foreign producers will enter the most profitable domestic markets, and countries or regions less exposed to trade liberalization will be affected by competition and they will have to reduce their costs and increase their productivity to stay in the market, so they can no longer afford the luxury of discrimination. Therefore, the gender wage gap would tend to decrease in the countries or regions most exposed to trade liberalization.
Therefore, it was considered important to have an estimate of the magnitude of the wage gap and its decomposition between men and women employed in regions of Mexico according to their exposure to trade liberalization, if at all, in order to confirm if the female group obtains a lower average wage than the male population for reasons different to its productive characteristics. Faced with this concern, we present below the findings of different studies that address the problem at the international level and for Mexico.
The issue has been analyzed at the international level from a regional perspective [
The analysis of the effect of trade liberalization on the wage gap by gender in regions and/or sectors of Mexico has been addressed in different studies. Artecona et al. [
Dominguez y Brown [
For their part, Rodríguez y Castro [
Tello et al. [
Popli [
For their part, Arceo and Campo [
In another study, Rodríguez and Castro [
Castro et al. [
For its part, Rodríguez [
In summary, the review of empirical evidence outlined in this section covers from 1987 to 2015, considering different sources of information, economic sectors, cities, and regions, as well as with different methodological approaches. In general, the documents conclude that during the period of commercial and subsequent opening, the wage gap persists against women, although it is observed that it decreases over time, as the non-explaining part (associated with gender wage discrimination) contributes to explain most of the wage gap. On the other hand, the results of the studies reviewed do not provide elements that indicate that trade liberalization favored the reduction of gender wage inequality in Mexico. Finally, the review allows identifying the insufficiency of recent studies that analyze the effects of trade liberalization on gender pay inequality using methodologies that carry out a measurement and decomposition of the gender wage gap along the distribution.
The study is based on the analysis of the individual data of the ENOE carried out quarterly by the National Institute of Statistics and Geography (INEGI). In the present work, the third quarter of 2005 and 2016 is considered, which avoids the effects of seasonal factors and which is intended to cover the most recent available information period of the survey and, at the same time, to extend the analysis of studies similar antecedents.
The study population is women and men subordinate and paid between 16 and 65 years of age. Years of education are used, as well as the potential experience of individuals, as well as the monthly hourly wage deflated with the Consumer Price Index (CPI) 2016.
There are different studies that classify Mexico, Hanson [
Source: Aguilera and Castro 2016.
Map 1. Spatial distribution of the states that make up the regions according to their exposure to trade liberalization in Mexico.
To meet the objective of this research, wage equations are estimated using a quantile regression technique, for different segments of the wage distribution, in order to decompose the wage differences between women and men, in their components explained and not explained by the characteristics of human capital.
Koenker and Bassett [
Melly [
Melly [
It is assumed that the quantiles of Y are linear to X and the parametric general extension is direct. An estimator of the unconditional distribution function is presented in the presence of covariates, which is used to decompose the distribution differences, which is analogous to the Oaxaca-Blinder decomposition [
In Equation (1), Y reflects the explanatory variable of the logarithm of hourly wages, X are the explanatory variables for the individual i, in this case, education and potential experience F Y ( q ) and is the cumulative function of the variable Y of q, F(y) q, represents the density of Y, at the same point, F( θ ) and-1 reflects the inverse of the distribution function, commonly called the quantile function, evaluated at 0 < θ <1. F y ( q | X i ) , represents the unconditional accumulation of the distribution function Y evaluated at q, which receives X = Xi.
q ^ t ( θ ) = inf { q : n t − 1 ∑ i : T i = t F ^ Y ( t ) ( q | X i ) ≥ θ } (1)
The quantile of unconditional distribution can be estimated considered by a simple quantile q ^ t ( θ ) is more precise than the simple quantile. However, it is a counterfactual estimator; it is used to decompose the differences in distribution and estimate the effects by quantile.
q ^ c ( θ ) = inf { q : n t − 1 ∑ i : T i F Y ( 0 ) ( q | X i ) ≥ θ } (2)
The disparity in the logarithm of hourly wages is primarily explained by differences in characteristics (continuing education and potential experience) between men and women. The second term is the part of the inequality associated with the difference in wages to the characteristics, that is to say, it corresponds to the part of the inequality of the logarithm of wages, due to the differences (quantile) of the characteristics between men and women. This is explained by the unconditioned quantile function based on a binary treatment of t (women 0 and men 1) and would reflect the counterfactual function.
q ^ 1 ( θ ) − q ^ 0 ( θ ) = [ q ^ 1 ( θ ) − q ^ c ( θ ) ] + [ q ^ c ( θ ) − q ^ 0 ( θ ) ] (3)
The decomposition of the wage differential with the Oaxaca [
To explain changes in wage structure, standard economic theory focuses on average wage dynamics rather than changes in the wage distribution, thus ignoring differences in the lower or upper part of the wage distribution. In addition, average wage statistics can hide important functions in the wage structure. The contribution of this research is to obtain wage differentials throughout the distribution, by quantiles at the regional level according to their exposure to the commercial opening in Mexico in the recent decade.
2005 | 2016 | |||||||
---|---|---|---|---|---|---|---|---|
Regions according to their exposure to trade liberalization | Regions according to their exposure to trade liberalization | |||||||
México | High | Medium | Low | México | High | Medium | Low | |
Hourly wage (mexican pesos 2016) | 42.38 | 45.26 | 40.91 | 41.75 | 37.09 | 38.79 | 36.10 | 37.44 |
Hours | 44.80 | 44.66 | 44.80 | 45.13 | 44.26 | 44.21 | 44.27 | 44.31 |
Married | 0.604 | 0.603 | 0.605 | 0.602 | 0.589 | 0.607 | 0.576 | 0.603 |
Head of household | 0.461 | 0.464 | 0.460 | 0.461 | 0.445 | 0.457 | 0.436 | 0.455 |
Educational years | 10.15 | 10.03 | 10.18 | 10.34 | 10.99 | 10.80 | 11.03 | 11.23 |
Experience | 19.75 | 19.35 | 20.12 | 19.06 | 20.90 | 20.41 | 21.37 | 20.00 |
Full time | 0.750 | 0.774 | 0.742 | 0.723 | 0.750 | 0.768 | 0.748 | 0.720 |
Qualified | 0.380 | 0.350 | 0.389 | 0.412 | 0.462 | 0.418 | 0.479 | 0.492 |
Sex | 0.607 | 0.624 | 0.601 | 0.591 | 0.583 | 0.596 | 0.578 | 0.572 |
14 - 25 | 0.238 | 0.250 | 0.228 | 0.253 | 0.201 | 0.220 | 0.189 | 0.212 |
26 - 35 | 0.277 | 0.281 | 0.273 | 0.279 | 0.252 | 0.247 | 0.254 | 0.260 |
36 - 45 | 0.254 | 0.252 | 0.256 | 0.248 | 0.252 | 0.255 | 0.248 | 0.257 |
46 - 55 | 0.161 | 0.153 | 0.167 | 0.155 | 0.195 | 0.190 | 0.200 | 0.181 |
56 - 65 | 0.0700 | 0.0633 | 0.0748 | 0.0655 | 0.100 | 0.0880 | 0.109 | 0.0897 |
High region | 0.315 | 0.303 | ||||||
Medium region | 0.563 | 0.568 | ||||||
Low region | 0.121 | 0.129 | ||||||
Agriculture, forestry and fishing | 0.00857 | 0.00916 | 0.00708 | 0.0139 | 0.00726 | 0.00687 | 0.00597 | 0.0138 |
Extractive industry and electricity | 0.00743 | 0.00639 | 0.00688 | 0.0127 | 0.00581 | 0.00563 | 0.00497 | 0.00993 |
Manufacturing industry | 0.189 | 0.253 | 0.171 | 0.102 | 0.176 | 0.251 | 0.156 | 0.0861 |
Building | 0.0788 | 0.0853 | 0.0726 | 0.0904 | 0.0805 | 0.0846 | 0.0754 | 0.0936 |
Commerce | 0.226 | 0.206 | 0.235 | 0.233 | 0.216 | 0.203 | 0.222 | 0.219 |
Restaurants and accommodation services | 0.0684 | 0.0634 | 0.0624 | 0.109 | 0.0892 | 0.0791 | 0.0861 | 0.127 |
Transport, communications, mails and storage | 0.0728 | 0.0608 | 0.0812 | 0.0648 | 0.0676 | 0.0565 | 0.0757 | 0.0582 |
Financial profesional services | 0.0837 | 0.0796 | 0.0886 | 0.0718 | 0.103 | 0.0877 | 0.114 | 0.0916 |
Social services | 0.0862 | 0.0796 | 0.0890 | 0.0906 | 0.0827 | 0.0792 | 0.0816 | 0.0960 |
Miscellaneous services | 0.120 | 0.111 | 0.123 | 0.127 | 0.118 | 0.109 | 0.119 | 0.131 |
Government | 0.0601 | 0.0457 | 0.0628 | 0.0847 | 0.0545 | 0.0383 | 0.0586 | 0.0744 |
Observations | 16,348,681 | 5,156,261 | 9,207,502 | 1,984,918 | 18,006,488 | 5,463,979 | 10,219,252 | 2,323,257 |
Source: Own elaboration based on ENOE data, respective years.
information presented in
The descriptive information of the data allows identifying the existence of regional differences relevant in the labor markets. In the region with the highest exposure, the real hourly wage is higher, as it would correspond to a region with greater labor competition due to a strong link with the foreign market. The information also recorded a fall in real hourly wages in the period (2005-2016), a fact that is registered in all regions. On the other hand, in the region with the greatest exposure, the manufacturing sector has a greater participation, and in this region, workers have fewer years of education than the rest of the regions, which seems to support the argument of the Stolper-Samuelson theorem, Although the lower relative participation of women in this regional labor market seems not to go in that direction.
In the following section, a comparison of the wage gap by gender in each one of the regions of Mexico is made, as well as its decomposition.
The review of previous studies on the wage gap in Mexico indicate that: 1) gender pay gaps are relevant, 2) the measurement and decomposition of wage inequality has been done mainly through methodologies that consider only the mean values of the distribution, 3) there are few studies that incorporate recent data and analyze the behavior of the wage gap in the last decade, 4) regional analyses that include spatial delimitations associated with the degree of exposure to the outside have been made without considering a product regionalization of a methodological exercise, and 5) the origin of wage discrimination is not sufficiently explained, aspects that are covered in this document and constitute the main contribution of the same.
The first objective is to determine the gender wage gap by considering different points of the distribution since it is reasonable to assume that the gender wage gap does not necessarily have the same magnitude as its average value.
From the information provided by the
of the reduction Is different by regions and deciles, highlighting the higher deciles of the region of high exposure.
In relation to the first hypothesis, based on the information available, there do not seem to be any elements to support compliance, since the region with the highest exposure does not present the lowest relative wage gap by gender. These results go online as found by [
On the other hand, it can be observed that the wage gap is significantly lower in the upper deciles relative to the lower part of the distribution, a fact that provides evidence in favor of hypothesis 2, and that it would not find correspondence with Popli [
Once the gender wage gap in each of the regions has been exposed, the characterization and coefficients are decomposed using the quantile regression technique by Melly [
In order to identify the contribution of the characteristic effects (human capital endowments) and those related to the coefficients (yields to endowments), the wage gap by deciles for the year 2005 is decomposed and presented in the
From the information provided by the decomposition can be established: 1) the wage gap is the product of the two effects found, where the corresponding coefficients are dominant in all regions, 2) the contribution of the characteristic effects and coefficients To the total wage gap is different by deciles, with the highest coefficient effect at the top of the distribution, even higher than the total wage gap, which is compensated for by negative characteristic effects, indicating that women have more human capital than the men; 3) the behavior of the effects coefficients and characteristics are different between regions throughout the distribution, while in the region of high exposure the coefficient effect is increasing by deciles, in the region of medium exposure the effect remains stable
and in the less exposed the coefficient decreases, so there seems to be a direct relationship between exposure to the openness and magnitude of the coefficient effect along the distribution, indicating an inverse behavior than expected a priori.
From the information provided by the breakdown of the wage gap by sex for the different regions, it is possible to affirm that there are no elements available to support that the unexplained component (coefficient effect) of the wage gap decomposition is smaller In the region of greatest exposure, which rejects hypothesis 3, indicating that the greater competition generated by trade liberalization, at least in the case of Mexico, does not constitute a strong enough element to reduce wage discrimination by gender Associated to the coefficients), an outcome that is shared by Artecona and Cunningham [
It is important to note that there is a reduction of the coefficient effect on the wage gap over time and a greater convergence with the rest of the regions, which would partially support the hypothesis 4.
In general, it can be deduced that in Mexico and the different regions, gender wage gaps have been reduced from 2005 to 2016, where most of the reduction is not explained by productive characteristics, but by unexplained factors, although this element continues to explain most of the wage gap.
The evidence found is in line with the findings of Popli [
The objective of the paper is to analyze the effect of trade liberalization on the wage gap by gender and its components in Mexico. For this, the behavior of the wage gap is analyzed for different regions of the country, classified according to the degree of exposure to trade liberalization. Using the methodology of Melly [
The results reveal the existence of a gender wage gap in the different regions that its size differs regionally, along with the distribution and that the wage gap has diminished over time in all regions of Mexico, but the wage disadvantage against women and this is higher in the lower deciles. Most of the differences are not explained by productive characteristics, although the proportion of inequality associated with discrimination factors declined in 2016 in relation to the previous reference year.
In the document, different hypotheses were put forward to contrast the effect of trade liberalization on the wage gap and its components, particularly on the one related to differences in returns, and which is associated with elements of discrimination to explain the wage gap.
The first hypothesis argues that the regions with the greatest exposure to trade liberalization have lower gender wage gaps relative to less exposed regions.
In relation to this first hypothesis, based on the available information, there does not appear to be any elements to support its compliance, since the region with the highest exposure does not present the lowest relative gender pay gap for any of the years, a result that is shared by previous studies that considering different periods, sources of information and methodologies do not find sufficient evidence to maintain that the commercial opening favors the women in the remunerations and reduces the wage gap by gender.
The second hypothesis states that women in the highest deciles will have a lower gender wage gap in relation to lower deciles. In this respect, the data provide evidence in favor of this hypothesis, given that the wage gap is lower in the upper deciles of the distribution, a fact that is fulfilled for the two reference years, a result that differs from previous studies where they find elements to sustain that In the case of Mexico, the so-called “glass ceiling” phenomenon is presented in the wage distribution by gender.
Hypothesis 3 states that the unexplained component of the wage gap, which is often associated with discrimination, will be lower in regions with higher exposure relative to less exposed regions. In this respect, the results indicate that, based on the information provided by the breakdown of the gender wage gap for the different regions, it is possible to affirm that there are no elements available to support that the unexplained component (coefficient effect) of the wage gap decomposition is lower in the region of greatest exposure, which rejects hypothesis 3, a fact that can be interpreted as an indication that the greater competition associated with trade liberalization, at least for the case of Mexico, is not a strong enough element to reduce gender pay discrimination, a result that is shared by previous studies.
Finally, hypothesis 4 states that the above behaviors (hypotheses 1 - 3) will be the most pronounced effects in 2016 in relation to 2005 and that a trend towards regional convergence will be observed, given the permeability of the effect of trade opening over time in the different regions. In this respect, the results of the wage gap estimates and their decomposition do not provide elements for not rejecting hypotheses 1 and 3, which would not confirm this hypothesis, there are elements to affirm the presence of greater convergence in wage gaps regional, and this would be partially fulfilled.
The reduction in the wage gap in all regions regardless of their degree of exposure is possibly indicating the presence of other factors, not necessarily linked to trade liberalization, such as the policy of wage restraint, among others, which are affecting on the behavior of the gender wage gap homogeneously throughout Mexico.
The document provides elements that allow a better knowledge about the effects of trade liberalization on the wage gap and its components along the wage distribution for the case of Mexico in the last decade. The results do not seem to contribute to the fact that trade liberalization generates a reduction in the gender pay gap and that the most relevant component of the reduction is associated with performance factors, which are linked to discrimination; however, it is pertinent to recognize the limitation that for the study requires considering in the analysis two reference years, it would be desirable to include more years of the study. The above identifies the need to advance in the thematic, with the purpose of having more knowledge of the effect of the commercial opening on the wage gap and its decomposition. In the analysis developed in this document, it considers salaried and remunerated workers from all economic sectors (public and private); however, a reasonable extension of this line of research would be to reduce the analysis to the manufacturing sector, since it is the economic activity most directly related to trade liberalization; a second extension would be to perform a decomposition of the different components associated with the endowment of human capital to identify the contribution and evolution in the explanation of the wage gap along the distribution.
Pérez, R.E.R. and Lugo, D.C. (2017) Trade Liberalization and Gender Wage Gap in Mexico. Modern Economy, 8, 1167-1185. https://doi.org/10.4236/me.2017.810081