You are on page 1of 19

The Gender Wage Gap and Poverty

in Colombia

Diego F. Angel-Urdinola — Quentin Wodon

Abstract. Using 19 years of cross-section wage data, we look at the trend of


the ratio of male to female wages in Colombia. First, we observe a long-term
trend towards an increase in the gender wage gap that may be related to new
labor regulations giving more protection to women and thereby raising the cost
of female employment for firms. Second, we measure the impact on poverty of
the increasing gap by producing estimates of the income distribution for the
subsequent years had the wage gap remained at the low level it used to be in the
early 1990s. As mainly low-skilled women suffered from the increase in the wage
gap, we find that the phenomenon contributed to an overall increase in poverty
during the decade.

1. Introduction

Despite significant advances over the last few decades, the gaps
between men and women in terms of labor force participation and
wages in Latin America remain substantial. According to a recent
World Bank (2002b) report, women’s labor force participation
remains much lower than for men. In Brazil, Chile, Colombia,
Mexico, and Peru, between 43 per cent and 56 per cent of
working-age women work, as compared with more than 88 per
cent among men. In all the countries reviewed, women also earn
less than men.
The ratio of female to male earnings in urban areas varies from 68
per cent in Uruguay to 91 per cent in Venezuela. As shown in
Figure 1 from World Bank (2002b), there is a weak negative

Diego F. Angel-Urdinola (author for correspondence) — Quentin Wodon, The


World Bank, ECSHD, Human Development, Europe and Central Asia, 1818 H
Street, Washington, DC 20433, USA. E-mail: dangelurdinola@worldbank.org.
The views expressed here are those of the authors and need not reflect those of
the World Bank, its Executive Directors, or the countries they represent.
LABOUR 20 (4) 721–739 (2006) JEL J31, J71, I32
© 2006 The Authors
Journal compilation © 2006 CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd., 9600
Garsington Rd., Oxford OX4 2DQ, UK and 350 Main St., Malden, MA 02148, USA.
722 Diego F. Angel-Urdinola — Quentin Wodon

Figure 1. Ratio of female to wage income in some Latin American


countries
95.00
Ratio of female to male wage income, 1999

Venezuela
90.00
El Salvador

85.00
Colombia
Ecuador Panama

80.00 Paraguay Argentina


Honduras Costa Rica
Nicaragua Chile
75.00
Bolivia Mexico
Guatemala Brazil
70.00
Uruguay

65.00

60.00
0.00 2,000.00 4,000.00 6,000.00 8,000.00 10,000.00 12,000.00 14,000.00
GDP per capita (average 1995–2002)

Source: Authors’ estimates using World Bank (2002b) estimates.

relationship in Latin America between the level of development of


a country (as measured by its per capita GDP) and the ratio of
female to male wages. Other factors, such as the size of the informal
sector, may influence this ratio. However, in this regard, there may
not be a lot of dispersion in Latin America as the informal sector
accounts for about half of the overall urban employment in many
countries. As for Colombia, it has been a country with a tradition-
ally low wage gap for Latin American standards. According to
Lopez Claros and Zahidi (2005), Colombia is indeed the Latin
American country with the lowest ‘gender gap’ as measured by a
composite index that takes into account economic participation,
economic opportunity, political empowerment, educational attain-
ment, and health well-being.
However, the gender wage gap in Colombia suffered from a
significant deterioration in the early 1990s. Velez et al. (2001) find
that during the first years of the 1990s in Colombia the wages for
male workers increased by approximately 15 per cent with respect
to those of their female counterparts. After 1995 the gender wage
gap narrowed by about 5 per cent (during a recessionary period)
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Gender Wage Gap and Poverty in Colombia 723

especially among less skilled wage earners (Correia, 2002). Esti-


mates using household survey data corroborate these findings. As
presented in Table 1, the ratio of male to female wages among wage
earners was close to one during the early and late 1980s (suggesting
no wage gap between men and women on average). The ratio
increased to 1.1527 between 1991 and 1996 and decreased back to
1.0725 between years 1997 and 2000. These ratios are however
based on simple means that do not control for the characteristics of
workers.
In this paper, we have two objectives. First, using repeated cross-
sections of household survey data from Colombia for 1982–2000,
we estimate the trend in the wage gap conditional on education and
other observable characteristics and assess whether the observed
increase in the wage gap can be associated to the adoption of the
Law 50 in 1990. Article 34 of the law specifies that all female
workers in a state of pregnancy have the right to a 12-week paid
leave while keeping their prevailing salary during the period. Article
35 states that a female worker cannot be fired due to pregnancy.
There is a common perception in Colombia that the law may have
provided disincentives for firms to hire young women as they are
associated to higher costs. As contracts were renewed and firms
progressively took into account the mandates of the law, employers
had the incentive to reduce female wages in order to compensate for
the rise in their benefit packages. If the adoption of the law indeed
contributed to the decrease in relative female wages, we would
expect that this decrease would be concentrated among younger
workers. This is not what we find, so that we cannot link Law 50 to
the increase in the wage gap.1
Our second objective is to assess the impact on poverty of the
observed increase in the gender wage gap in Colombia. For this, we
produce a counterfactual income distribution for the 1990s had the
gap remained at the low level observed in 1990. Poverty is a major
issue in Colombia. After a reduction in the share of the population
living in absolute poverty from 1978 to 1995, poverty increased
after 1995 due to a prolonged recession, so that by 1999 much of
that gain had been lost (World Bank, 2002a). Our analysis of the
impact on poverty of the changes in the wage gap focuses on the
1990s because it is during this period (as opposite to the 1980s) that
the increase in the wage gap was pronounced and significant. The
rest of the paper is structured as follows. Section 2 outlines our
methodology. Section 3 provides our key results. A conclusion
follows.
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
724
Table 1. Number of wage earners, relative labor supply, and relative wages (Colombia 1982–2000)
Mean hourly wage in constant 1999 pesos (all sample) Bootstrap estimates

© 2006 The Authors


Number of wage Ratio men 95 per cent
Year earners in sample to women Male Female Male/Female s.e. confidence interval

1982 13,152 1.49 2,295.86 2,281.21 1.0064 0.0212 0.9852 1.0277


1983 13,214 1.47 2,386.74 2,377.27 1.0040 0.0214 0.9826 1.0254
1984 11,858 1.36 2,374.22 2,497.72 0.9506 0.0203 0.9303 0.9708
1985 9,130 1.37 2,792.28 2,795.60 0.9988 0.0175 0.9813 1.0163
1986 9,792 1.37 2,384.66 2,458.66 0.9699 0.0235 0.9464 0.9934
1987 10,507 1.29 2,108.99 2,316.15 0.9106 0.0196 0.8910 0.9302
1988 14,234 1.68 2,648.90 2,312.02 1.1457 0.0151 1.1306 1.1608
1989 11,338 1.25 2,196.11 2,345.08 0.9365 0.0167 0.9198 0.9532
1990 9,505 1.27 2,162.66 2,305.83 0.9379 0.0197 0.9182 0.9576
1991 10,588 1.15 2,154.00 2,352.62 0.9156 0.0236 0.8920 0.9392
1992 11,945 1.46 2,641.10 2,303.66 1.1465 0.0207 1.1257 1.1672
1993 13,088 1.41 2,972.45 2,721.89 1.0921 0.0480 1.0441 1.1400
1994 13,638 1.46 3,100.44 2,692.30 1.1516 0.0338 1.1178 1.1854
1995 12,937 1.43 2,938.52 2,608.54 1.1265 0.0266 1.0999 1.1531
1996 13,226 1.35 2,564.05 2,224.34 1.1527 0.0194 1.1334 1.1721
1997 10,881 1.30 3,072.85 2,718.45 1.1304 0.0240 1.1064 1.1543
1998 10,757 1.29 3,341.04 3,008.61 1.1105 0.0244 1.0861 1.1349
1999 9,177 1.26 3,285.93 3,031.34 1.0840 0.0303 1.0537 1.1143
2000 9,048 1.20 3,125.14 2,913.78 1.0725 0.0280 1.0446 1.1005

Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Source: Authors’ estimates.
Note: Only individuals with positive wage are accounted for.
Diego F. Angel-Urdinola — Quentin Wodon
Gender Wage Gap and Poverty in Colombia 725

2. Methodology

Our first objective is to provide new and consistent estimates of


the gender wage gap in Colombia. This is done by estimating stan-
dard wage regressions for various education groups, namely indi-
viduals with 0–5 years of education, 6–10 years, 11 years, 12–15
years, and 16 years or more (these intervals correspond to the
education structure in Colombia for primary, incomplete secon-
dary, complete secondary, incomplete tertiary, and tertiary levels,
respectively). For each education group, we estimate separate
regressions for younger and older individuals (above and below 40
years of age).
The left-hand side variable is the hourly wage of the individual
in logarithm. The right-hand side variables include a constant, a
dummy variable for whether the individual is a male, age and age
squared, a dummy variable for whether the individual is single, two
dummy variables for the sector of activity of the individual (the
excluded category corresponds roughly to manufacturing), six
dummy variables for geographic location according to Colombia’s
main cities, the number of hours worked and its squared, and two
interaction effects between gender and, respectively, whether the
individual is single and the age of the individual.2
Denoting by ‘maleit’ the value of the male dummy for individual
i, by Iit the interaction terms with gender, by Xit the other indepen-
dent variables, and by wit the hourly wage, we estimate the following
regression for each year of data (repeated cross-sections for t =
1, . . . , T) and for various subsets of the sample defined by edu-
cation level and age group:

Ln (wit ) = φ + κ t′X it + θ t maleit + νt′I it + ε it t = 1, . . . T . [1]

From [1], our estimate of the average expected gender wage gap in
percentage terms, or more precisely our estimate of the ratio of the
logarithm of expected male to female earnings controlling for
observable characteristics, is

Gˆ t ≡ E [Ln ( wmale ) X t , I t ] − E [Ln ( wfemale ) X t , I t ] = θˆt + νˆ t′I t . [2]

Next in order to assess the impact on poverty of changes in the


gender gap, we define a counterfactual distribution of wages in the
absence of changes in the wage gap by rescaling the wages for
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
726 Diego F. Angel-Urdinola — Quentin Wodon

women in each year of data and for each education group (recall
that although we have not indicated this in the notation of the
variables to keep the equations easy to read, the above is done for
various education and age groups). For each woman k with earn-
ings, we calculate:

Gˆ baseline
kt ≡ wfemale,kt
C
wfemale, . [3]
Gˆ t

The counterfactual wage density for women (wC) at time t there-


fore corresponds to the wage level that would have been observed
if the wage trend for women had followed in a proportionate
way the trend for males (i.e. if the wage gap had remained con-
stant at the value observed in the base year). The final step for
estimating the impact on poverty of changes in the gender gap
consists in estimating the poverty measures. A first set of poverty
measures is obtained using the actual incomes observed for each
household in the various surveys. The second (counterfactual) set
of measures is obtained by replacing the wages of working women
wages by their counterfactual value as expressed in [3]. Thus, only
those households who have at least one woman working may see
their wage income adjusted and poverty status altered.
For poverty measurement, we use total household per capita
income as our indicator of well-being, and half the median per
capita income as our relative poverty line. The poverty measures
themselves are the standard measures of the FGT class (Foster
et al., 1984). Denoting household income for household i from
cross-section t by yit, the number of poor in that year by qt, the
population by nt, and the poverty line by zt, the FGT poverty
measures are defined as
α
qt
z −y
∑ ⎛⎜⎝ t zt it ⎞⎟⎠ .
1
Pαt = [4]
nt i =1

When a = 0, we obtain the headcount index of poverty, which is


simply the share of the population below the poverty line. When
a = 1, we obtain the poverty gap that takes into account the dis-
tance separating each household from the poverty line. When a = 2,
we get the squared poverty gap that takes into account the square of
the distance separating each household from the poverty line. As to
compute the change in poverty due to the change in the wage gap we
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Gender Wage Gap and Poverty in Colombia 727

simply replace in the income aggregate of each household the earn-


ings of working women, the change in poverty due to the change in
the wage gap is defined as
α
qt
z −y
∑ ⎛⎜⎝ t zt it ⎞⎟⎠
1
Pαt − Pαt,1990 =
nt i =1
α
1 qt ,1990
⎡ zt
1990
− ( yit + wfemal
1990
e,kt − wfemale,kt ) ⎤

nt
∑ ⎢⎣ zt1990 ⎥⎦ . [5]
i =1

We actually compute the counterfactual poverty measures only


when estimates of the gender wage gap (Ĝt) are statistically signifi-
cant. The procedure is implemented for all years after 1990. Note
that when we change the earnings of women according to [3], the
poverty line also changes, because it is defined in relative terms, as
half the median per capita income (obviously, the number of the
poor changes as well). The reason for using a relative rather than
absolute poverty line is that although we do have income data for
all years, we do not have the necessary information for properly
estimating absolute poverty lines over time in a cost of basic needs
framework (this is because household consumption surveys are
implemented only every 5 years or so in Colombia). In the case of a
wage gap increase over time, women’s wages will be lower than
what they would have been if the male to female wage ratio had not
changed. This would increase absolute poverty, as we would have a
decrease in earnings in the simulations. But the impact on relative
poverty is uncertain: it depends on where the working women
affected by the change in the wage gap are located in the distri-
bution of household income.
In the estimates to be discussed in the next section, we will see that
changes in the gender wage gap mostly affect women with low levels
of education. As these women tend to belong to households in the
lower part of the distribution of income, we would expect to see an
increase in relative poverty associated with the increase in the wage
gap. The decrease in absolute poverty, which we cannot measure
rigorously here for data reasons, would likely be even larger.

3. Data and results

Data are from the Encuesta Nacional de Hogares (National


Household Survey) conducted by the National Department of
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
728 Diego F. Angel-Urdinola — Quentin Wodon

Statistics and the National Department of Planning. Labor force


participation, wages, and information on individual attributes are
available for all individuals of age 12 or older. We use 19 surveys,
one for each year between 1982 and 2000. For comparability pur-
poses over the years, we restrict the samples to workers in seven
metropolitan areas (Barranquilla, Bucaramanga, Bogotá, Mani-
zales, Medellín, Cali, and Pasto). These areas account for 71 per
cent of the urban Colombian population on average over the
period. We include workers between 12 and 60 years of age with a
positive wage, but we exclude domestic employees, some among the
self-employed, unpaid family workers, and firm-owners from the
sample. The workers in the sample account for 63 per cent of total
urban employment in Colombia on average over the period under
review (this statistic takes into account the individual survey
weights). Poverty measures are obtained for the urban sample as a
whole in the seven metropolitan areas for the surveys considered
here. Wage and income variables are deflated using the country’s
urban consumer price index.
The data covers a total of 1,224,856 individuals; 218,015 wage
earners; and a total of 312,748 households from 1982 to 2000.
Table 2 gives descriptive statistics on key variables over the 19 years
of data. For example, 80 per cent of workers in the sample have less
than complete high school (i.e. less than 12 years of education), and
most work in the manufacturing, retail, and service industries.
Some 70 per cent of workers in the sample are residents of the cities
of Bogotá, Medellín, and Cali (the three largest Colombian metro-
politan areas).
The wage regressions are given for 2 years of data and are pre-
sented in Table A1 in the Appendix. The regressions for the other
years are similar and available upon request. As regressions are run
for each education group separately, we do not have education on
the right-hand side, but the other variables behave as expected. Older
individuals, who presumably have more experience, have higher
hourly earnings. There are some geographic differences in earnings;
for example, workers in the capital of Bogotá tend to have higher
earnings than workers living elsewhere. Workers working more
hours also have higher hourly earnings, presumably because they
have better and more stable jobs, and because workers working more
than 48 hours per week get overtime wage premiums of up to 20 per
cent. The gender variables tend to be statistically significant.
Our estimates of the gender wage gap are given in Table 3.
Among unskilled workers (workers with less than tertiary
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Table 2. Sample means of key variables, Colombia 1982–2000
0–5 years of 6–10 years of 11 years of 12–15 years of 16+ years of
education education education education education All

No. of observations 53,499 58,241 58,180 20,528 27,567 218,015


Group as share of all wage 0.25 0.27 0.27 0.09 0.13 1.00
earners
1. Mean household income 995,541 1,147,290 1,406,335 1,914,673 2,749,534 1,452,908
2. Mean hourly wage (in 2000 1,615 1,912 2,372 3,314 5,759 2,579
pesos)
3. Mean log hourly wage 7.24 7.39 7.60 7.90 8.35 7.58
4. Mean hours of work 47.15 46.91 46.25 45.11 44.81 46.36
5. Mean age 34.17 29.75 30.19 30.45 35.33 31.73
6. Per cent married 0.51 0.55 0.57 0.59 0.48 0.54
7. City
Barranquilla 0.10 0.11 0.13 0.14 0.16 0.12
Bucaramanga 0.12 0.09 0.09 0.09 0.09 0.10
Gender Wage Gap and Poverty in Colombia

Bogotá 0.26 0.28 0.27 0.29 0.32 0.28


Manizales 0.06 0.05 0.07 0.07 0.07 0.06
Medellín 0.26 0.26 0.24 0.21 0.16 0.24
Cali 0.14 0.16 0.13 0.13 0.12 0.14
Pasto 0.06 0.04 0.06 0.07 0.09 0.06
8. Sector of employment
Agriculture and mining 0.03 0.01 0.01 0.01 0.02 0.02
Manufacturing 0.37 0.39 0.27 0.19 0.14 0.30
Construction and electricity 0.13 0.07 0.04 0.04 0.06 0.07
Commerce 0.20 0.23 0.26 0.18 0.10 0.21
Services 0.28 0.30 0.43 0.58 0.67 0.40

Source: Authors’ estimates.


729

© 2006 The Authors


Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
730
Table 3. Gender wage differential (Ĝt) by years of education, Colombia 1982–2000
0–5 years of 6–10 years 11 years 12–15 years 16+ years
education of education of education of education of education All

© 2006 The Authors


Year Ĝt s.e. Ĝt s.e. Ĝt s.e. Ĝt s.e. Ĝt s.e. Ĝt s.e.

1982 0.018 0.024 -0.034 0.024 -0.049* 0.030 -0.119** 0.052 0.160** 0.058 -0.022 0.025
1983 -0.027 0.025 0.018 0.023 -0.053* 0.029 -0.006 0.049 0.067 0.058 -0.009 0.025
1984 -0.007 0.028 0.004 0.026 0.062** 0.031 0.008 0.051 -0.108* 0.064 -0.004 0.029
1985 0.010 0.032 -0.014 0.028 0.045 0.034 -0.110* 0.062 0.149** 0.068 0.027 0.031
1986 0.097** 0.029 0.033 0.026 0.001 0.029 0.045 0.052 -0.085 0.058 0.021 0.029
1987 -0.020 0.029 -0.044* 0.024 0.051* 0.028 -0.017 0.049 0.124** 0.054 0.006 0.028
1988 0.196** 0.015 0.101** 0.013 0.050** 0.014 -0.081** 0.025 0.210** 0.029 0.137** 0.013
1989 0.103** 0.027 0.003 0.024 -0.052** 0.024 -0.047 0.052 0.220 0.047 0.034 0.027
1990 -0.001 0.031 -0.135** 0.025 0.015 0.027 -0.018 0.050 0.063 0.053 -0.027 0.030
1991 0.003 0.033 -0.024 0.028 0.029 0.027 -0.089* 0.048 -0.049 0.053 0.003 0.032
1992 0.221** 0.020 0.128** 0.017 0.078** 0.016 0.012 0.031 0.250** 0.032 0.167** 0.018
1993 0.205** 0.019 0.115** 0.017 0.067** 0.015 0.171** 0.031 0.132** 0.034 0.140** 0.018
1994 0.224** 0.021 0.199** 0.019 0.118** 0.017 -0.072** 0.032 0.076** 0.034 0.169** 0.019
1995 0.213** 0.018 0.144** 0.016 0.122** 0.015 -0.023 0.029 0.264** 0.033 0.161** 0.018
1996 0.156** 0.018 0.197** 0.016 0.122** 0.014 -0.061** 0.027 0.058* 0.030 0.175** 0.017
1997 0.166** 0.023 0.188** 0.022 0.068** 0.015 0.122** 0.028 0.134** 0.030 0.1500** 0.020
1998 0.106** 0.024 0.116** 0.020 0.036** 0.017 -0.029 0.031 0.041 0.032 0.097** 0.022
1999 0.060* 0.031 0.065** 0.024 0.034** 0.019 0.074** 0.033 -0.003 0.035 0.086** 0.025
2000 0.060* 0.032 0.023 0.026 0.011 0.019 -0.001 0.037 0.067* 0.036 0.073** 0.027

Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Source: Author’s estimates.
Note: * Significant at 10 per cent level; ** significant at 5 per cent level. These are the joint significance tests (taking into account several estimated
parameters and the value at the mean of the independent variables).
Diego F. Angel-Urdinola — Quentin Wodon
Gender Wage Gap and Poverty in Colombia 731

Figure 2. Four-year moving average wage gap (Ĝt), Colombia, 1985–


2000, unskilled workers
0.25

0.20
Wage gap in percentage

0.15

0.10

0.05

0.00

–0.05
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

0–5 6–10 11

Source: Authors’ estimates.

education), who constitute 80 per cent of the sample, the wage


gap in favor of men averaged 8 per cent over the period of study.
Within this group, the gap was relatively low during the 1980s,
increased during the first half of the 1990s, and decreased back
after the recession in 1995. Among workers with incomplete uni-
versity or technical degrees (12–15 years of education), the wage
gap was close to zero. Among workers with complete tertiary edu-
cation (16 and more years of schooling), the gap averaged 10 per
cent. Figures 2 and 3 provide the trends in the gender wage gaps
for unskilled and skilled workers. To smooth the data in the
graphs and to show the trends more clearly, four-period moving
average values were used, so that the first year on the graphs cor-
responds to 1985.
Figures 4 and 5 present the estimates of the wage gap for
unskilled workers by age group (workers above and below 40
years of age). We do so to identify whether the upward trend in
the ratio of male to male wages is steeper among younger workers
as they are presumably the more affected by the new labor law.
Although we find that the ratio of male to female wages increases
after year 1990 for both younger and older workers, contrary to
what we expected, we find that such increase is more pronounced
(and volatile) among older workers. This gives up prima facie
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
732 Diego F. Angel-Urdinola — Quentin Wodon

Figure 3. Four-year moving average wage gap (Ĝt), Colombia, 1985–


2000, skilled workers
0.20

0.15
Wage gap in percentage

0.10

0.05

0.00

–0.05

–0.10
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

12–15 16+

Source: Authors’ estimates.

Figure 4. Four-year moving average of the wage gap by age group,


Colombia, 1985–2000 (workers with 0–5 years of education)
0.40

0.30
Wage gap in percentage

0.20

0.10

0.00
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999

–0.10
40 years and older
Less than 40 years old
–0.20

–0.30

Source: Authors’ estimates.

© 2006 The Authors


Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Gender Wage Gap and Poverty in Colombia 733

Figure 5. Four-year moving average of the wage gap by age group,


Colombia, 1985–2000 (workers with 6–10 years of education)
0.25

0.20
Wage gap in percentage

0.15

0.10

0.05

0.00
1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999
–0.05
40 years and older
Less than 40 years old
–0.10

–0.15

–0.20

Source: Authors’ estimates.

evidence that although the increase in the gap after the 1990s
coincided with the establishment of the new law, the increase may
have been motivated by other circumstances different from legis-
lation (and which affected mainly older workers) that need further
understanding.
We now turn to the impact of the increasing gender wage gap on
relative poverty. As mentioned before, there was a 5–10 per cent
decrease in relative wages for female workers from low education
groups due to an increase in the gender wage gap between 1990 and
2000. Given such information, and as described in the last section,
we compute the counterfactual relative poverty at time t had the
wage gap remained as it used to be in 1990 (the baseline year).
The results are presented in Table 4. Had the gender wage gap
remained at the lower level it used to be in 1990, relative poverty as
measured by the headcount, the poverty gap, and the squared of the
poverty gap would have decreased by 15.13, 10.7, and 13.9 per cent
(equivalent to 1.38, 0.15, and 0.04 percentage points; these are
averages for the period 1991–99) among households having at least
one affected female worker. As expected, as the gender wage gap
increased, so did relative poverty, as the decrease in relative wages
among unskilled females hit poor households the most. The
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
734 Diego F. Angel-Urdinola — Quentin Wodon

Table 4. Changes in poverty, households with female worker(s),


Colombia 1982–2000
All measures multiplied Factual minus
by 100 Factual Counterfactual counterfactual

Headcount ratio 1990 9.14 9.14 0.00


Poverty gap ratio 1.37 1.37 0.00
Squared poverty gap 0.31 0.31 0.00
Headcount ratio 1991 9.70 9.60 0.10
Poverty gap ratio 1.87 1.83 0.04
Squared poverty gap 0.54 0.52 0.01
Headcount ratio 1992 3.20 2.57 0.63
Poverty gap ratio 0.49 0.41 0.08
Squared poverty gap 0.16 0.12 0.04
Headcount ratio 1993 3.67 3.14 0.53
Poverty gap ratio 0.60 0.49 0.12
Squared poverty gap 0.20 0.15 0.04
Headcount ratio 1994 6.51 5.12 1.39
Poverty gap ratio 0.93 0.82 0.11
Squared poverty gap 0.33 0.29 0.04
Headcount ratio 1995 3.10 2.45 0.65
Poverty gap ratio 0.61 0.47 0.14
Squared poverty gap 0.21 0.15 0.06
Headcount ratio 1996 5.55 3.56 1.99
Poverty gap ratio 0.53 0.39 0.14
Squared poverty gap 0.14 0.10 0.04
Headcount ratio 1997 13.37 9.69 3.68
Poverty gap ratio 1.73 1.29 0.44
Squared poverty gap 0.40 0.31 0.09
Headcount ratio 1998 13.10 11.18 1.92
Poverty gap ratio 1.26 1.05 0.20
Squared poverty gap 0.23 0.19 0.04
Headcount ratio 1999 6.10 4.43 1.67
Poverty gap ratio 0.90 0.80 0.10
Squared poverty gap 0.27 0.25 0.02
Headcount ratio 2000 8.44 7.15 1.29
Poverty gap ratio 1.56 1.45 0.11
Squared poverty gap 0.67 0.60 0.06

Source: Author’s estimates.


Note: Poverty line: one half of the median wage.

percentage changes in relative poverty throughout the period due to


changes in the gender wage gap for the headcount index and the
poverty gap are shown in Figure 6 (the graph for the squared
poverty gap is very similar).3
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Gender Wage Gap and Poverty in Colombia 735

Figure 6. Change in relative poverty due to changes in the gender wage


gap, Colombia, 1990–2000 (headcount and poverty gap)
4.00 0.50

0.45
3.50
0.40
3.00

Change in the poverty gap


Change in the headcount

0.35
2.50
Change in the headcount 0.30
Change in the poverty gap
2.00 0.25

0.20
1.50
0.15
1.00
0.10
0.50
0.05

0.00 0.00
1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000

Source: Authors’ estimates.

4. Conclusion

Using repeated cross-sections of data, we have assessed empiri-


cally the long-term trend in the gender wage gap in Colombia. Our
findings suggest that there has been an increase in the gender wage
gap, which may however not be related to new labor regulations
introduced in the 1990s giving more protection to women. Indeed,
although the decrease in the ratio of female to male wages coincided
with the establishment of the new law, we found (somewhat unex-
pectedly) that the wage gap increased faster among older workers
(who were presumably less affected by the new law) than among
younger ones.
We have also simulated the impact of changes in the wage gap on
relative poverty measures. Because less educated women were
exposed the most to the increase in the gender wage gap, we found
that relative poverty would have been lower in the period 1991–
2000 had the wage gap remained at its level of 1990. The estimated
impact on poverty of the changes in the gender wage gap was
important showing that the effects on welfare were not trivial.
© 2006 The Authors
Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Appendix
736

Table A1. Wage regressions, selected years, Colombia


Years of education: 1982 Years of education: 1997

© 2006 The Authors


0–5 6–10 11 12–15 16+ 0–5 6–10 11 12–15 16+

Male -0.13 0.05 -0.03 -0.16 0.61 0.23 0.14 0.07 0.11 0.10
[1.78]* [0.57] [0.25] [0.70] [2.46]** [2.94]*** [1.77]* [1.08] [0.80] [0.70]
Single ¥ male -0.03 0.08 -0.02 0.00 -0.13 -0.13 -0.16 -0.12 -0.06 -0.07
[0.78] [2.04]** [0.31] [0.05] [1.38] [2.93]*** [3.82]*** [3.84]*** [1.05] [1.20]
Age ¥ male 0.00 -0.01 0.00 0.01 -0.02 0.00 0.00 0.00 0.00 0.00
[1.52] [2.12]** [0.02] [0.75] [2.48]** [0.11] [1.67]* [1.55] [0.04] [1.11]
Age 0.02 0.03 -0.01 -0.01 -0.14 0.05 0.04 0.03 0.04 0.04
[3.33]*** [4.53]*** [0.62] [0.71] [6.42]*** [8.97]*** [6.23]*** [6.18]*** [3.48]*** [3.42]***
Squared 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
of age [2.76]*** [2.45]** [1.43] [0.65] [6.57]*** [7.64]*** [4.29]*** [3.47]*** [1.77]* [2.08]**
Single -0.09 -0.15 -0.13 -0.16 -0.18 0.00 0.02 -0.10 -0.02 -0.05
[2.67]*** [4.74]*** [3.45]*** [2.26]** [2.43]** [0.11] [0.62] [4.75]*** [0.48] [1.22]
Agriculture, 0.03 0.08 0.03 -0.02 -0.18 0.00 -0.01 -0.01 -0.15 -0.24
manuf., [1.31] [3.30]*** [0.74] [0.27] [2.00]** [0.18] [0.49] [0.70] [3.44]*** [4.42]***
and
construction
Wholesale 0.06 0.07 0.06 0.02 -0.04 0.05 0.12 0.11 0.02 -0.10
and retail [2.73]*** [3.30]*** [1.99]** [0.40] [0.76] [2.31]** [5.54]*** [6.26]*** [0.48] [2.71]***
Barranquilla 0.72 0.52 0.51 0.21 0.32 0.36 0.21 0.17 0.00 0.05
[14.66]*** [8.87]*** [7.12]*** [1.69]* [2.42]** [8.45]*** [5.00]*** [5.61]*** [0.05] [1.00]
Bucaramanga 0.44 0.42 0.45 0.24 0.33 0.41 0.28 0.21 0.15 0.15

Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
[8.32]*** [6.83]*** [5.73]*** [1.58] [2.25]** [10.99]*** [7.12]*** [6.57]*** [2.81]*** [2.60]***
Diego F. Angel-Urdinola — Quentin Wodon
Bogotá 0.57 0.46 0.56 0.31 0.54 0.47 0.36 0.34 0.31 0.35
[13.25]*** [8.88]*** [8.95]*** [2.74]*** [4.57]*** [11.43]*** [9.14]*** [11.24]*** [5.97]*** [6.92]***
Manizales 0.00 0.05 -0.06 0.00 0.06 0.11 0.03 -0.03 -0.02 0.05
[0.18] [1.45] [1.28] [0.06] [0.58] [3.69]*** [1.26] [1.11] [0.55] [0.97]
Medellín 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00 0.00
[0.71] [2.26]** [0.61] [0.21] [0.82] [4.20]*** [1.97]** [0.03] [0.21] [1.45]
Cali 0.58 0.31 0.51 0.25 0.62 0.24 0.20 0.11 0.14 0.17
[8.33]*** [3.72]*** [5.95]*** [1.55] [3.50]*** [6.26]*** [4.94]*** [3.52]*** [2.60]*** [2.86]***
Hours of 0.63 0.54 0.53 0.39 0.50 0.39 0.28 0.28 0.30 0.34
work [13.97]*** [9.99]*** [8.07]*** [3.18]*** [3.78]*** [10.84]*** [7.57]*** [9.40]*** [5.97]*** [6.28]***
Squared of 0.57 0.46 0.46 0.22 0.32 0.38 0.29 0.23 0.17 0.26
hours [12.30]*** [8.36]*** [6.85]*** [1.86]* [2.50]** [9.74]*** [7.59]*** [7.22]*** [3.25]*** [4.29]***
of work
Constant 7.06 5.96 9.18 8.58 9.50 3.81 5.93 7.90 7.65 6.71
Gender Wage Gap and Poverty in Colombia

[11.50]*** [7.45]*** [9.07]*** [4.30]*** [4.20]*** [5.79]*** [9.83]*** [14.00]*** [8.02]*** [5.83]***
2
R 0.11 0.09 0.10 0.05 0.10 0.23 0.23 0.27 0.22 0.22
Observations 4,453 4,006 2,461 904 1,333 1,976 2,389 3,571 1,217 1,719

Source: Authors’ estimates.


Notes: Absolute value of t-statistics in brackets.
* Significant at 10 per cent; ** significant at 5 per cent; *** significant at 1 per cent.
Omitted categories are: gender — female; industry — services; city – Pasto.
737

© 2006 The Authors


Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
738 Diego F. Angel-Urdinola — Quentin Wodon

Table A2. F-test (Null hypothesis (Ho): interaction of age and marital
status with male dummy is equal to zero)
F-statistic (Prob > F)

0–5 years of 6–10 years 11 years


All education of education of education

1982 0.8343 0.1410 0.0036 0.9525


1983 0.0000 0.0546 0.0129 0.0001
1984 0.0000 0.2364 0.0042 0.0002
1985 0.8508 0.8164 0.1541 0.9176
1986 0.0054 0.1227 0.6662 0.0001
1987 0.0103 0.4228 0.6124 0.0008
1988 0.0000 0.0000 0.0000 0.0817
1989 0.0000 0.0809 0.0849 0.1335
1990 0.0197 0.0016 0.1838 0.3990
1991 0.0097 0.5067 0.2737 0.0026
1992 0.0000 0.0000 0.0001 0.0354
1993 0.0000 0.0016 0.0000 0.0000
1994 0.0000 0.0002 0.0077 0.0135
1995 0.0000 0.0000 0.0001 0.0018
1996 0.0001 0.0106 0.0230 0.0010
1997 0.0000 0.0129 0.0000 0.0000
1998 0.0020 0.0220 0.0256 0.0003
1999 0.0000 0.0018 0.0038 0.0025

Source: Authors’ estimates.


Note: Fail to reject Ho (at a 10 per cent confidence level) results are underlined.

Notes
1
Changes in the wage gap may also occur as a result of shifts in relative supply
and demand. As presented in Table 1, the relative supply of male workers has
decreased relative to the supply of women in Colombia. An increase in relative
supply of women may affect female wages negatively if workers from opposite
sexes are not perfect substitutes. Employers may not substitute female with male
employment if they perceive that they cannot do equivalent jobs and thus males
would then enjoy a relative wage premium that would increase as their relative
supply decreases. This argument has been made by Freeman (1976), Katz and
Murphy (1992), and Card and Lemieux (2002) in the context of imperfect substi-
tution between high- and low-skilled labor. Blau and Kahn (1997) find that the net
effect of shifts of men to women supply and demand in the USA between 1971 and
1988 was unfavorable for women as a group (more specifically, the authors docu-
ment that the higher demand for female workers was more than offset by the rapid
increase in female supply).
2
Hours of work are often excluded from wage regressions. However, they are an
important determinant of hourly wages in Colombia. According to Colombian

© 2006 The Authors


Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006
Gender Wage Gap and Poverty in Colombia 739

labor law, full-time wage earners working more that 48 hours per week are entitled
to a wage premium of up to 20 per cent for overtime. As presented in Table A1,
hours of work and their squared value are positive and significant covariates in all
specifications. Estimates of [1] include interaction effects of selected variables
(evaluated at the mean) with the sex dummy. Table A2 presents F-statistics
showing that (in most cases) the inclusion of interaction terms between age and
marital status and the male dummy variable is indeed relevant to estimate the
wage gap.
3
We do not discuss here the causes for the change in the gender wage gap. As
pointed out by an anonymous referee, although this change could be due to
changes in how women are treated on the labor market, it could also be due to
other factors, such as a potential change in the unobserved aspects of the quality of
female workers employed in the economy.

References

Blau F. D. and Kahn L. M. (1997) ‘The Impact of Wage Structure on Trends in


U.S. Gender Wage Differentials 1975–1987’, National Bureau of Economic
Research, NBER Working Papers 4748.
Card D. and Lemieux T. (2002) ‘Can Fall in Supply Explain the Rising Return to
College for Younger Men? A Cohort-based Analysis’, Quarterly Journal of
Economics 116(2): 705–746.
Correia M. (2002) ‘Gender’ in Guigale M., Lafourcade D. and Luff C. (eds.)
Colombia: The Economic Foundation of Peace, Washington, DC: The
World Bank: 767–783.
Foster J. E., Greer J. and Thorbecke E. (1984) ‘A Class of Decomposable Poverty
Indices’, Econometrica 52: 761–766.
Freeman R. (1976) The Overeducated American. San Diego, San Diego, CA:
Academic Press.
Katz L. and Murphy K. (1992) ‘Changes in Relative Wages, 1963–1987: Supply
and Demand Factors’, Quarterly Journal of Economics 107(1): 35–78.
Lopez Claros A. and Zahidi S. (2005) ‘Women’s Empowerment: Measuring the
Global Gender Gap’, World Economic Forum, Geneva, Switzerland.
Velez C., Leibovich J., Kugler A., Bouillón C. and Núñez J. (2001). The Re-
versal of Inequality Trends in Colombia, 1978–1995: A Combination of Per-
sistent and Fluctuating Forces. Mimeo. Washington, D.C. The World
Bank, Available at http://wbln0018.worldbank.org/lac/lacinfoclient.nsf/0/
fc5c8989b7b2d32285256b8a00733137/$FILE/MidiColombia.pdf.
World Bank (2002a) Colombia Poverty Report, No. 24524-CO, Washington, DC:
World Bank.
World Bank (2002b) Challenges and Opportunities for Gender Equality in Latin
America and the Caribbean, Washington, DC: Latin America and
Caribbean Region.

© 2006 The Authors


Journal compilation © CEIS, Fondazione Giacomo Brodolini and Blackwell Publishing Ltd. 2006

You might also like