=Paper=
{{Paper
|id=Vol-3282/icaiw_aiesd_10
|storemode=property
|title=Sticky Floor and Glass Ceiling in Ecuador. The Evolution of the Gender Wage Gap, 2010-2021
|pdfUrl=https://ceur-ws.org/Vol-3282/icaiw_aiesd_10.pdf
|volume=Vol-3282
|authors=Diego Linthon-Delgado,Lizethe Méndez-Heras,Gino Cornejo-Marcos
|dblpUrl=https://dblp.org/rec/conf/icai2/Linthon-Delgado22
}}
==Sticky Floor and Glass Ceiling in Ecuador. The Evolution of the Gender Wage Gap, 2010-2021==
https://ceur-ws.org/Vol-3282/icaiw_aiesd_10.pdf
Sticky Floor and Glass Ceiling in Ecuador. The
Evolution of the Gender Wage Gap, 2010-2021
Diego Linthon-Delgado1 , Lizethe Méndez-Heras2,* and Gino Cornejo-Marcos2
1
Universidad de Guayaquil, Guayaquil, Ecuador
2
Universidad Ecotec, Samborondón, Ecuador
Abstract
This research focuses on the evolution of the Gender Wage Gap (GWG) and its components (endowment
effect and coefficients effect) over the entire wage distribution in Ecuador between 2010 and 2021. It was
used Melly’s decomposition method based on quantile regressions along the wage distribution with Data
from Ecuador’s National Survey of Employment, Unemployment, and Underemployment (ENEMDU
for its name in Spanish). There was found strong evidence of a Sticky Floor (a larger GWG at the tenth
percentile than the median) and of a Glass Ceiling (a larger GWG at the ninetieth percentile than the
median) for Ecuador. The results show that women with low and high income have limited mobility and
invisible barriers which prevent them from getting better positions and salaries. The ninety-nine (99)
regressions also suggest a GWG favoring men, which is explained by a discrimination factor (coefficient
effect). Public policy must be geared towards fighting discrimination against women to reduce gender
wage inequality.
Keywords
Gender, Wage Gap, Discrimination, Quantiles Regressions
1. Introduction
Over the past decades, concerns about inequality have grown significantly worldwide, par-
ticularly those related to gender because of the Covid-19 pandemic. According to the World
Economic Forum gender inequality indices, in 2021, the global gender gap favoring men reached
32.3%. Meanwhile, over the Political Empowerment and the Economic Participation and Op-
portunity areas, it got up to 78% and 42%, respectively. The lowest inequality level was over
Educational Attainment (5) and Health and Survival 4% [1].
The International Labor Organization (ILO) estimated that globally Covid-19 caused a drop
in the weekly worked hours ratio (2.4%), over employment (2.5%), and workforce participation
(1.9%) in 2020 compared to 2019. Furthermore, this indices recovery will be slow since, according
to some of ILO’s estimates, the levels we had before the pandemic will not be reached by 2023.
Covid-19 also provoked a dropped in employment of 2.8% for men and 2.2% for women between
ICAIW 2022: Workshops at the 5th International Conference on Applied Informatics 2022, October 27–29, 2022, Arequipa,
Peru
*
Corresponding author
$ diego.linthondel@ug.edu.ec (D. Linthon-Delgado); lmendez@ecotec.edu.ec (L. Méndez-Heras);
gcornejo@ecotec.edu.ec (G. Cornejo-Marcos)
0000-0003-2115-2807 (D. Linthon-Delgado); 0000-0002-3885-4584 (L. Méndez-Heras); 0000-0002-7541-203X
(G. Cornejo-Marcos)
© 2022 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
Workshop
Proceedings
http://ceur-ws.org
ISSN 1613-0073
CEUR Workshop Proceedings (CEUR-WS.org)
169
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
2019 and 2020. However, women already had a significantly lower employment rate than men
by 2019 (45.2% against 69.4%), and by 2022 there have not been any relevant changes in rates.
Gender inequality in Latin America has had a long concerning story with data, resulting in old
multiple academic pieces of research aiming at analyzing the wage gap3. Ecuador is no stranger
to this matter4. As seen in the following part of this work, factual evidence shows a wage gap
in Ecuador, considerably persistent benefiting men. Most empiric research on the determiners
of this gap in Ecuador has been based on the Blinder [2] and the Oaxaca [3] methodology, a
pioneer back in their time, which focuses on trying to distinguish two main groups of causes.
On the one hand, some originated from specific, quantifiable differences over the study subject
(i.e., academic level).
On the other hand, we have those whose root is discrimination5. Based on this method,
the bibliography used for the national analysis tends to conclude that the gender gap results
from discrimination and not due to employees’ different characteristics. Likewise, part of the
literature finds that the wage gap shows a downtrend, although most papers only analyze short
terms.
This study analyzes and identifies critical factors in the evolution of the wage gap along
with its observable and non-observable components in Ecuador for the period 2010-2021. This
research contributes to the empirical literature in two fundamental aspects. First, we analyze
the gap for all wage distribution (quantiles 1 to 99). This is especially relevant since it helps
approach the "Sticky Floor" and "Glass Ceiling" issues6. It also provides evidence of Covid-19’s
impact on the evolution of the wage gap and its components.
On top of this introduction, this article is built as follows. In the first section, we will conduct
relevant research on gender wage inequality in Ecuador. In the second section, we present
the traditionally used Blinder [2] and Oaxaca [3] decomposition methodologies, and Melly [4]
applied to this research. In the third section, we describe the data of the taken sample for the
National Survey of Employment, Unemployment, and Underemployment (ENEMDU) carried
out by the National Institute of Census and Statistics (INEC for its name in Spanish) for the
2010-2021 period. In the fourth section, we will review the results of applying the methodology
for Ecuador’s case. Finally, we present conclusions and recommendations.
2. Reviewing the literature
Amid the first research on the gender wage gap in Ecuador, we find the one conducted by Ñopo et
al. [5], who analyzed wage inequality by gender and ethnicity during the 2003-2007 period. These
authors found that the gender wage gap benefits men less than the ethnic wage gap. Regarding
gender, the wage gap showed a stable pattern, fluctuating between 7% and 12% between 2003
and 2007. Moreover, this gap was mainly explained by ’discrimination’ due to women having
higher human capital7 than men. When analyzing the wage distribution, they found evidence
of the ’Sticky Floor’ and ’Glass Ceiling’ patterns, meaning women with high or low income
have little mobility or face invisible barriers preventing them from achieving better jobs and
salaries. They used annual figures from the ENEMDU and applied a technique designed by one
of the authors [6], which consists in decomposing the wage gap into four additive components:
1) differences between the combination of some characteristics that non-minorities do have
170
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
but that minorities do not have, 2) the opposite to the first component, differences between
the combination of some characteristics that minorities do have but that non-minorities do not
have, 3) differences between observable characteristics between both comparable groups, and,
4) differences between the non-observable characteristics (discrimination).
Pérez et al. [7] focused on applying four decomposition methodologies for ethnicity and
gender wage gap known as 1) multiple linear regression, 2) Blinder-Oaxaca’s method, 3) Machado
et al. [8] quartiles decomposition method and 4) Juhn et al. [9] quantiles decomposition method
to analyze the evolution of the wage gap. These authors use data from the Survey of Urban
and Rural Employment and Unemployment (ENEMDUR for its name in Spanish) for 2007
and the ENEMDU for 2013. Using the first methodology, they found that women earned an
average wage 13% lower than men. The second methodology, Blinder-Oaxaca’s, was used
by Pérez et al. [7]. They realized that in 2013 the wage gap benefitted men, mainly due to
discrimination. Implementing Machado et al. [8] methodology, they noticed that the gap is
significantly higher on the lower side of the wage distribution (quantiles 1, 10, and 25) than
on the higher side (quantiles 75, 90, and 99). Finally, to assess the changes over time, by using
the same methodology, they found evidence of a settling ’Sticky Floor’ phenomenon, although
they did not observe any changes over the ’Glass Ceiling’ patterns. In summarizing, they found
that the gender wage gap dropped during the studied year and the discrimination factor had a
similar trend.
Aligned with the works that used Blinder-Oaxaca’s methodology with bias selection correc-
tions, we have Linthon-Delgado et al. [10], who used data from ENEMDU for September 2020,
estimated that the gender wage gap was about 35.6 percent points in favor of men where the co-
efficients component (discrimination) explained this gap since women had a better endowment
of human capital.
So far, the analysis for literature in Ecuador had as its primary source the ENEMDU. Antón
et al. [11] found similar results by using data from the administrative records of the Labor and
Entrepreneurial Dynamics Laboratory (LDLE for its name in Spanish) from INEC in 2016. They
used three methodologies to estimate the gender wage gap in Ecuador’s public and private
sectors. The first is a multiple linear regression model considering gender as a Dummy variable.
They estimated that women earn a lower salary than men. Their second methodology was
Blinder-Oaxaca’s decomposition. With this method, they found that the gender wage gap
is mainly explained by discrimination within the private sector, while in the public sector,
they could not find any evidence of gender wage inequality. The third methodology was a
decomposition of quantiles [12], and they applied it to the 25, 50, and 75 quantiles, and they
found that women earned less than men in all three quantiles, meaning that ’Sticky Floor’ and
’Glass Ceiling’ phenomena were present. A limitation of this research is that they restrain the
sample to people with a third (university) and fourth (postgraduate) levels of education.
The research analyzing the wage gap over different distribution points shows the coexistence
of the ’Sticky Floor’ and ’Glass Ceiling’ patterns, although there is a wider discrepancy in its
magnitude and evolution. Some say that ’Sticky Floor’ is similar in size to the ’Glass Ceiling,’
and others think the ’Glass Ceiling’ is a much bigger phenomenon than the ’Sticky Floor.’ The
main conclusion of all this research is that the gender wage gap is due to discrimination, not
human capital differences.
171
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
3. Methodology
The most used methodology to decompose wage gaps is Blinder [2] and Oaxaca [3]. This allows
to decompose wage gaps into two groups, A and B, into two components: i) the observable
component, which comes from the differences in the socio-demographic characteristics of
individuals, and ii) the non-observable, which is associated with discriminatory policies or
behaviors against certain groups.
This method consists of estimating Mincer’s regressions [13] for a wage on both groups, A
and B, in order to obtain the wage returns of the observable characteristics of each group:
𝑛
∑︁
ln 𝑊𝑖𝐴 = 𝛽0𝐴 + 𝛽𝑗𝐴 𝑋𝑗𝑖
𝐴
+ 𝜇𝐴
𝑖 (1)
𝑗=1
𝑛
∑︁
ln 𝑊𝑖𝐵 = 𝛽0𝐵 + 𝛽𝑗𝐵 𝑋𝑗𝑖
𝐵
+ 𝜇𝐵
𝑖 (2)
𝑗=1
Where subindices “𝑖” and “𝑗” represent employees and coefficients, respectively. ln 𝑊 is the
natural logarithm of wages, 𝑋 represents the human capital component for employees, 𝛽 are
the regressors, and reflect the returns from the labor market to the characteristics of employees
or the ’prices’ of services associated with them [14] and is the error term.
Stressing equations 1 and 2, the Blinder-Oaxaca’s decomposition results in the following
expression:
(︁ )︁ ∑︁
¯𝐴 𝐵 ¯𝐵 𝐴 ¯𝐴 𝐵 ¯𝐵
∑︁ ∑︁ ∑︁
𝛽𝑗𝐴 𝑋 ¯
(︀ 𝐴 𝐵
)︀
𝑗 − 𝛽 𝑗 𝑋 𝑗 = 𝛽 𝑗 𝑋 𝑗 − 𝑋 𝑗 + 𝑋 𝑗 𝛽𝑗 − 𝛽𝑗 (3)
𝑗 𝑗 𝑗 𝑗
(︁ )︁
Where the first term, 𝑗 𝛽𝑗𝐴 𝑋
∑︀ ¯𝐴 −𝑋 ¯ 𝐵 represents the observable component, meaning
𝑗 𝑗
the part of the wage gap that can be explained by the differences
)︁ observed in the characteristics
∑︀ ¯ 𝐵 (︁ 𝐴
of groups A and B, and the second term 𝑗 𝑋 𝑗 𝛽𝑗 − 𝛽𝑗 expresses the non-observable
𝐵
component, in other words, the part of the wage gap that can be explained by the differences
in the coefficients related to each of the estimations of the equations for employees’ wages in
groups A and B, typically considered as the discrimination effect.
Blinder-Oaxaca’s methodology is strongly present in the field; however, it only allows the
calculation of the mean and not of all the wage distribution [9, 12, 8].
We used Melly’s methodology [4], which allows us to decompose the wage gap for the entire
distribution. This method calculates the conditional distribution by using quantile regression.
Melly [4] proposes to estimate the unconditional distribution through quantile regressions and
integrate the conditional distribution over a covariables range.
Melly’s decomposition methodology allows us to obtain a numerically equivalent estimator
when the number of simulations used in Machado et al. [8] decomposition tends to infinity and
is also computationally faster to obtain since when it does not depend on simulations [4].
Therefore, Melly [4] proposes the following process to estimate by integrating the uncondi-
tional counterfactual distributions:
172
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
• Step 1. Estimate through quantile linear regression all the conditional distributions.
• Step 2. Integrate the conditional function over the covariable range to obtain the uncon-
ditional distribution function.
• Step 3. Obtain the unconditional quantile function of interest by inverting the uncondi-
tional distribution function.
• Step 4. Invert the counterfactual distribution of interest to obtain the counterfactual
−1
quantile of interest 𝑄𝐶𝐹,𝜃 = 𝐹𝑊 𝐶 (𝜃) with women’s wage structure, indicated as follows:
𝐹
∫︁
𝐹𝑊 𝐶 (𝑊 ) = 𝐹𝑊𝑀,𝜃|𝑋𝑀 (𝑊 |𝑋)𝑑𝐹𝑥𝐻 (𝑋) (4)
𝑀,𝜃
Decompose the changes in wage density [15]:
∆𝜃 = 𝑄𝑀,𝜃 − 𝑄𝐶 (5)
[︀ ]︀ [︀ 𝐶 ]︀
𝐻,𝜃 + 𝑄𝑀,𝜃 − 𝑄𝐻,𝜃
After having estimated the counterfactual, it is now possible to decompose the gender
wage gap of the unconditional quantile function.
Equation 5 shows the decomposition of the gender wage gap for different quantiles,
where the first term on the right of the equation represents the effect of the characteristics
(observable components) and the second the effect of the coefficients (non-observable
components), commonly understood as the discrimination effect.
4. Description of the data
Data used for this research were taken from the ENEMDU, which is made on a quarterly basis
by INEC. We used figures for September within the 2010-2021 period. We limited the sample to
people between 18- and 65 years old living in an urban area. Table 1 shows the main statistics
for the total sample and Table 2 shows the main statistics for employees.
The structure of the age groups is very similar between men and women; about 50% of the
people in the sample are between 25-49 years old. This percentage grows by about 5% for all
employees. Almost 56% of people from the sample are married or live in Free Union, but when
it comes to the employees’ group, the percentage for women in either of these marital statuses
is, on average, 13% lower than for men. Concerning the head of household, the percentage of
men is on average 33% higher than women for the total sample; this percentage reaches up
to 38% when only employees are considered; this suggests there is a gender social division
of household roles in Ecuador just as the literature pinpointed [16, 17, 18]. The descriptive
statistics of these three variables (age, married, head of household) reveal no significant change
throughout the studied years.
In the labor context, we can see a rise in the average hourly wage from 0.81 in 2010 to 1.11 in
2021 for men and from 0.81 in 2010 to 1.15 in 2021 for women; meaning that by the last year of
the sample, there is a wage gap benefiting women. This wage gap in favor of women could be
connected to the fact that the percentage of women with higher education is, on average, 13%
higher than their male peers.
Figure 1 shows the observed gender wage gap for the whole sample per quantiles (1 to 99) for
the years 2010, 2015, and 2021. We can see a U shape, which means men have higher salaries
173
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
Table 1
Descriptive statistics. People between 18-65 years old (2010-2021)
All
2010 2015 2021
Man Woman Man Woman Man Woman
Age 37.4 38.2 37.3 37.6 37.9 38.9
Married 58.0 57.6 62.0 59.2 52.4 50.5
Head of Household 55.2 15.7 59.8 20.0 51.4 21.8
18-24 23.3 20.7 21.6 20.1 22.3 19.7
25-34 23.7 23.4 24.7 25.9 23.9 22.7
35-49 30.0 31.8 32.1 31.6 29.1 30.6
50 23.1 24.1 21.6 22.3 24.7 27.1
Work
Log hourly wage 0.81 0.82 1.15 1.15 1.11 1.15
Weekly hours 44.9 38.7 42.9 37.6 40.5 35.6
School level
Below elementary 1.4 2.4 1.2 1.8 1.0 1.3
Basic 23.8 25.5 24.3 23.1 16.9 17.6
High-School 42.2 39.7 44.5 42.0 46.1 43.6
Higher education 32.7 32.5 30.0 33.2 36.0 37.5
N 12026 13434 19690 21408 6002 6696
than women in the low part of the wage distribution (’Sticky Floor’) and the high part too
(’Glass Ceiling’). Firstly, the ’Glass Ceiling’ phenomenon shows a higher magnitude, suggesting
that women with higher salaries face the largest inequality compared with their male peers.
Secondly, women have better compensation in the middle part of the wage distribution, which
can result from better human capital skills for women during the last years. Thirdly, excluding
2021 (a pandemic period), we can see a practically constant uptrend for ’Sticky Floor’ and
’Glass Ceiling’; in other words, between 2010 and 2015, there was a wage inequality process
between men and women with the lowest income, while women with the highest wages did
not experience higher inequality with men between 2010 and 2015. Covid-19 seems to have
affected the gender wage gap since 2021, showing a more even U shape than the previous two
periods; this confirms the existence of a ’Sticky Floor’ and ’Glass Ceiling.’
In summarizing, the descriptive statistics analysis of the gender wage gap in Ecuador for the
2010-2021 period showed the following fundamentals: i) an uptrend in the ’Sticky Floor’ and
’Glass Ceiling’ patterns, ii) women have better income in a great part of the wage distribution,
iii) women with low school levels are in an evident disadvantage compared to men, iv) regarding
employees in the higher education group, men have better salaries in the low and the high side
of the distribution, v) Covid-19 seems to have increased the total female labor participation, but
it dropped the participation of married women, vi)the period affected by Covid-19, the ’Sticky
Floor’ and ’Glass Ceiling’ phenomena show a drop.
174
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
Table 2
Descriptive statistics. Employees between 18-65 years old (2010-2021)
Employees
2010 2015 2021
Man Woman Man Woman Man Woman
Age 39.0 39.1 38.7 38.8 39.5 40.2
Married 66.6 53.4 69.6 55.9 60.9 48.6
Head of Household 63.3 21.0 67.0 26.3 59.3 28.7
18-24 15.4 12.3 14.1 12.1 13.7 10.3
25-34 25.8 27.4 27.1 28.5 26.2 26.0
35-49 34.7 37.2 36.7 37.9 34.4 37.5
50 24.2 23.1 22.1 21.5 25.7 26.2
Work
Log hourly wage 0.81 0.82 1.15 1.15 1.11 1.15
Weekly hours 44.9 38.7 42.87 37.6 40.5 35.6
School level
Below elementary 0.9 1.3 0.9 1.1 0.4 1.0
Basic 25.0 20.2 26.2 20.5 18.3 15.9
High-School 42.6 35.3 43.9 37.7 46.8 39.9
Higher education 31.6 43.2 29.0 40.8 34.5 43.2
N 9636 6668 16356 11778 4686 3694
5. Results
We will go through the results of the decomposition of the gender wage gap in observable
factors (endowment effect) and non-observable (coefficients effect) after having applied Melly’s
methodology.
Figure 1 shows that during the three studied years, 2010, 2015, and 2021, the relevance of the
observable components (characteristics) and the non-observable (coefficients) for the gender
wage gap varies through the quantiles. In 2010, there is no wage gap for people with the lowest
income (until quantile 19), while after quantile 20 and up to the 80th, women’s salaries are
slightly higher than men’s, and then forward the gap expands in favor of men. In other words,
it can be said that in 2010 there was an ongoing ’Glass Ceiling’ phenomenon. More importantly,
results show that if the characteristics of employees had determined wages, the wage gap would
have benefitted women along the entire wage distribution. In fact, the observed wage gap had
such a trend due to discrimination.
Discrimination mainly affected women in the high part of the distribution. In 2015, the
observed wage gap showed a U trend, meaning it favored men in the high and low part of the
distribution; these phenomena are known as the ’Sticky Floor’ and ’Glass Ceiling’. Likewise,
in 2010, the observed wage gap in 2015 is mainly explained by discrimination; in other words,
men’s human capital had better compensation than women. Finally, in 2021, a year impacted
by Covid-19, we can see more even wages, the result of a drop in the ’Sticky Floor’ and ’Glass
Ceiling’ patterns compared with previous years (2010 and 2015). In the last mean of the wage
distribution, women have better average wages than men, but this is due to them having broader
175
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
Figure 1: Decomposition of the gender wage gap. Total, 2010, 2015 y 2021
human capital.
Furthermore, results show that the coefficients effect is higher than the endowment effect
along the distribution. This allows us to reinforce the gender discrimination hypothesis present
in the Ecuadorian labor market. However, the discrimination effect is higher for women in the
high part of the wage distribution, meaning that for the best-paid employees’ group, women’s
human capital is worth less in the labor market than men’s. This was particularly true in 2010
and 2015 since, in 2021, discrimination affected both women in the high part and women in
the low part of the distribution. These results also mean that if women were to bring the same
human capital as men to the labor market, the wage gap benefitting men would be way higher
176
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
than observed. In a non-discriminatory scenario, the wage gap would benefit women along the
entire wage distribution.
We can see in Figure 1 two key aspects: 1) men’s wages have benefitted from discrimination,
and 2) women have broader human capital than men. When we compare 2021 with 2010, we
can see a slight upturn in the ’Sticky Floor’ and an essential downturn in the ’Glass Ceiling’
Figure 2: Decomposition of the gender wage gap. People with low school levels, 2010,2015 y 2021
177
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
phenomena. We can say that when it comes to gender wage equality matters, women with
the lowest income are worse than ten years ago, and women with the highest income have
experienced an improvement. Finally, women were paid in the middle part of the distribution
in the three studied years.
In the same sense, Figure 2 shows the gender wage decomposition for people with low school
levels (primary or less). In 2010 the gender wage gap (observed) slightly benefitted men in
quantiles 1 to 83; ever since it has benefitted women. Moreover, this gap is mainly explained by
the coefficients effect, suggesting positive discrimination against women. In 2015, the wage gap
mostly favored men over women in all quantiles; although the gap was higher in the higher
part of the distribution, this uneven wage gap was also the result of discrimination. In 202, the
wage gap trend is like the one in 2010, suggesting that gender wage inequality dropped during
Figure 3: Decomposition of the gender wage gap. People with higher education, 2010, 2015 y 2021
178
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
the Covid-19 pandemic.
On the other hand, the decomposition of the gender wage gap showed that the coefficients
effect prevails for the higher education employees’ group (Figure 3). In 2010, the coefficient
effects grew significantly right after the 70th quantile, meaning that discrimination affects
women with higher education and high income. In 2015, the pattern was similar to 2010, where
the main difference is that it also affected women in the low quantiles. In 2021, we can see a trend
toward wage equality, although the ’Glass Ceiling’ phenomenon persists due to discrimination.
In summary, gender wage inequality is significantly different for lower and higher-education
employees. The main factor preventing gender wage equality is discrimination against women.
Without it, the gender gap in the lower education level employees’ group would be indiscernible
for almost all earnings’ quantiles; furthermore, wage inequality would significantly drop for
men and women with higher education. Finally, Covid.19 seems to have dropped gender wage
inequality.
6. Conclusions
This research aimed to analyze the evolution of the gender wage gap and its components
(observed and non-observed) in Ecuador during the period between 2010 and 2021. We used
Melly’s process, which allows the decomposition of the gender wage gap per quantiles in
observed factors (characteristics effect) and non-observed (coefficients effect).
This research results showed that the gender wage gap is widely different over most quantiles,
which means that studies that only analyze the changes in the wage mean or in some points of
the distribution (quantiles 10, 50, and 90, for most cases), using Blinder-Oaxaca’s method or
methodologies based on quantile regressions present important limitations.
This study has proven that women in quantiles between 30 and 80 are better paid than men
and that during the analyzed period, this gap widened, even during the pandemic. However,
this is due to women having better school levels than men.
On the low part of the distribution, we found an evident pattern of ’Sticky Floor,’ which
rose between 2010 and 2015. Still, it slightly dropped in 2021, which leads to conclude that less
educated women engaged in low-productivity jobs face working more challenging than men
with similar characteristics. The problem with this gap is that it results from discrimination,
meaning the coefficients effect is the one that explains the wage gap benefitting men and not
the differences in the human capital (characteristics effect). Furthermore, although this research
results coincide with most of the investigation, many quantiles differ. Therefore, when we
consider the entire distribution, we can infer that the ’Sticky Floor’ mainly affects 20 percent of
the distribution (below quantile 20) and not only those in quantile 10, as it has usually been
studied.
Moreover, we found conclusive evidence showing ’Glass Ceilings,’ which is also the result
of discrimination. Nevertheless, during the Covid-19 pandemic period, there was an evident
downturn in this phenomenon, meaning there was more wage equality between men and
women with the highest income. This pattern is also more apparent after quantile 90. This last
quantile is the one that most studies consider analyzing the gap between employees with better
wages.
179
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
During the Covid-19 pandemic, we observed a change toward gender wage equality, especially
for women with low school levels; however, discrimination continues to be an essential obstacle
to wage parity for men and women with higher education.
This research provides a broad perspective on evaluating the gender wage in Ecuador during
the last decade. However, for further investigation, we recommend examining the gender wage
gap in more specific groups, such as employees in the public and private sectors, urban and rural
areas, and formal and informal sectors, amongst others. This will undoubtedly bring a broader
understanding of the domestic trends for wage gaps. Future research should also consider bias
correction over sample selection since this would lead to more precise estimations.
Concerning public policy, this research shows that policies should focus on fighting discrimi-
nation and women exclusion, mainly in jobs with low and high productivity. Results prove that
women have more than enough human capital to hold such positions and that granting access
to those jobs will significantly contribute to dropping the gender wage gap and to the seeking
of a more egalitarian society.
References
[1] World Economic Forum, Global gender gap report 2021, https://www3.weforum.org/docs/
WEF_GGGR_2021.pdf, 2021.
[2] A. S. Blinder, Wage discrimination: reduced form and structural estimates, Journal of
Human resources (1973) 436–455.
[3] R. Oaxaca, Male-female wage differentials in urban labor markets, International economic
review (1973) 693–709.
[4] B. Melly, Public-private sector wage differentials in germany: Evidence from quantile
regression, Empirical Economics 30 (2005) 505–520.
[5] H. Ñopo, L. Gallardo, Ethnic and Gender Wage Gaps in Ecuador, IDB Working Paper, 2009.
[6] H. Ñopo, Matching as a tool to decompose wage gaps, The review of economics and
statistics 90 (2008) 290–299.
[7] F. A. Pérez, D. I. Torresano, Etnia y género en el mercado laboral ecuatoriano: cuatro
aplicaciones empíricas para la descomposición salarial, Master’s thesis, Quito: EPN, 2015.,
2015.
[8] J. A. Machado, J. Mata, Counterfactual decomposition of changes in wage distributions
using quantile regression, Journal of applied Econometrics 20 (2005) 445–465.
[9] C. Juhn, K. M. Murphy, B. Pierce, Wage inequality and the rise in returns to skill, Journal
of political Economy 101 (1993) 410–442.
[10] D. E. Linthon-Delgado, L. B. Méndez-Heras, et al., Descomposición de la brecha salarial
de género en el ecuador, Remef-Revista Mexicana de Economía y Finanzas Nueva Época
REMEF (The Mexican Journal of Economics and Finance) 17 (2022) 1–25.
[11] J. Antón, J. V. Carpio, J. Rodríguez-Moreno, N. Lara, Brecha salarial de género en los
sectores público y privado del ecuador, X-pedientes Económicos 4 (2020) 47–67.
[12] R. Koenker, G. Bassett Jr, Regression quantiles, Econometrica: journal of the Econometric
Society (1978) 33–50.
180
Diego Linthon-Delgado et al. CEUR Workshop Proceedings 169–181
[13] J. Mincer, Schooling, Experience, and Earnings. Human Behavior & Social Institutions No.
2., ERIC, 1974.
[14] F. Bourguignon, F. Ferreira, N. Lustig, Decomposing changes in the distribution of house-
hold incomes: Methodological aspects, The microeconomics of income distribution
dynamics in East Asia and Latin America (2005) 17–46.
[15] P. Salardi, Wage disparities and occupational intensity by gender and race in brazil: an
empirical analysis using quantile decomposition techniques, in: World Bank Conference,
volume 7, 2012, pp. 1–12.
[16] D. Contreras, L. De Mello, E. Puentes, The determinants of labour force participation and
employment in chile, Applied Economics 43 (2011) 2765–2776.
[17] C. Piras, Mujeres y trabajo en América Latina: Desafíos para las políticas laborales, IDB,
2006.
[18] G. Psacharopoulos, Z. Tzannatos, Education and Female Labor Force Participation., ERIC,
1987.
181