archivos de economía - DNP Departamento Nacional de Planeación

Anuncio
República de Colombia
Departamento Nacional de Planeación
Dirección de Estudios Económicos
ARCHIVOS DE ECONOMÍA
External Trade, Skill, Technology and the recent
Increase of Income Inequality in Colombia
Mauricio SANTA MARIA SALAMANCA
Documento 171
10 de Diciembre de 2001
La serie ARCHIVOS DE ECONOMIA es un medio de la Dirección de Estudios Económicos, no es un órgano
oficial del Departamento Nacional de Planeación. Sus documentos son de carácter provisional, de
responsabilidad exclusiva de sus autores y sus contenidos no comprometen a la institución.
EXTERNAL TRADE, SKILL, TECHNOLOGY
AND THE RECENT INCREASE OF INCOME
INEQUALITY IN COLOMBIA1
Mauricio SANTA MARÍA SALAMANCA.
INTRODUCTION
During the last decade there has been an increase in income inequality in Colombia, which has
been particularly notable in urban areas. This development marked a departure from the trend
observed from the beginning of the 60’s up to 1988, which showed consistent reductions in
inequality (inequality increased in rural areas in the 80s). Since the growth of inequality
coincided with an ambitious reform process, many analysts have concluded that the reforms are
to blame for this unfortunate development (see Santamaría, 1997, 1999 for detailed descriptions
of the reforms and the inequality developments). Despite the timing of these events, we will see
two important things in this paper. First, the increase in inequality has been a very complicated
process that has manifested itself very differently in the various parts of the income distribution, a
fact that has been obscured by the tradition of summarizing inequality with a single number. For
example, the left section of the distribution of earnings has not, in general, been negatively
affected, but rather the most significant developments have occurred at its middle. Second, I will
show that the connection between the increases in inequality and poverty and the trade reforms is
at best weak2. This is reinforced by the fact that the timing of the two phenomena is not so
coincident as it might seem at first. Instead, I will show that many of the most important changes
in relative earnings can be explained solely by the interaction of supply and demand for different
skill levels. The common view in Colombia that the rise in inequality is due mainly to the reform
needs to be revised.
This paper’s main objectives are thus to investigate the evolution of the distribution of labor
income in Colombia during the previous 20 years and to identify the factors that lie behind this
evolution. We attack the problem from two perspectives. First, we use a partial equilibrium
model of the labor market to study developments related to the distribution of labor earnings of a
homogeneous sample of individuals (homogeneous in terms of labor market attachement and
geographical location), using skills as our primary tool of analysis. That is, a simple supplydemand framework for different dimensions and levels of skill is utilized to evaluate the role
these variables played, and how they affected the distribution of labor earnings at different points
1
Very special thanks go to James W. Albrecht, Susan Vroman and Alejandro Mateus for their help,
comments and expert guidance. I also want to thank Carlos Eduardo Vélez and Ana María Ibáñez for
clever comments and other assistance. Finally, many thanks to the Banco de la República and the DNP for
data and financial assistance.
2
A complete review of the literature pertaining to income distribution, structural reform and their
relationship (not only for the Colombian case), can be found in Santamaría, 1999, and Cline, 1997.
1
in time. Second, we relate income to several demographical, skill and labor market factors, to
isolate the effect these have had on the evolution of that distribution along the entire income
range, and on other measures of overall dispersion and on skill differentials. We assess the merit
of different explanations and quantify the contribution of each factor to the total observed
changes. This second approach uses semi-parametric techniques that allow an analysis of the
entire distribution, without assuming a specific parametric form.
The difficulty of measuring the effects of trade on income distribution has been long recognized.
It has created much debate (see the various articles in Collins, 1998, Cline, 1997 and Kosters,
1991). For this reason we avoid making causal arguments and restrict ourselves to measuring the
degree of association between variables. This provides valuable information on the way that
foreign trade interacts with earnings and skills.
The paper is organized as follows. In the next section we study the changes in the distribution of
labor earnings using a skill demand-supply framework, with skill defined by education,
experience and gender. We begin by describing the data set and variables, and by discussing
adjustments made to the data. The formal aspects of the model are also explained. The third
section uses semi-parametric techniques to isolate the effects that several covariates had on the
changes of the distribution of labor income. Robustness checks are presented when necessary.
The fourth section concludes.
SUPPLY AND DEMAND FOR DIFFERENT SKILL LEVELS3
Data
The data used come from the ENH. Mostly, we worked with the national surveys of June 1978,
September 1988, December 1991, September 1994, 1996 and 1998 (the qualification “mostly”
refers to the fact that we carried out an exercise using additional surveys). My two main goals
were (i) to construct a stable sample over time in terms of labor market attachment, and (ii) to
extract information from the data that allows an investigation of the distribution of labor earnings
from a perspective emphasizing supply and demand for different levels and dimensions of skill.
Accordingly, we made several adjustments to the data and created new variables.
To guarantee full comparability over time we restricted the urban sample to the seven cities that
were present in all surveys. They accounted for 50% of the urban population in 1998 and are very
heterogeneous in terms of location and socio-economic characteristics (they are Barranquilla,
Bucaramanga, Bogotá, Manizales, Medellín, Cali and Pasto)4. In rural areas all four regions
defined in the survey were used. Additionally, I deleted observations according to the following
criteria. First, I deleted the inactive and unemployed. Second, those who stated that they were
employed but reported earnings of zero or did not report earnings were discarded. Third, we
deleted all observations for which the variables education, age, gender, sector of economic
activity and occupation were reported missing or outside the coding provided by DANE. Finally,
3
This section draws from Katz and Murphy (1992, KM hereafter), Bound and Johnson (92), and Murphy,
Riddell and Romer (1998).
4
In the 1978 survey the coding of the metropolitan areas changed and thus we resorted to other methods to
identify the seven cities. Although I am confident of the outcome of this exercise, some “leakage” may
have occurred and some observations from other cities may have been included or vice versa.
2
I discarded all workers that reported working less than 20 hours a week. After these adjustments,
the size of the weighted samples ranged from 3.400.000 workers in 1978 to 5.400.000 in 1998
(weighted means adding the sampling weights included in the ENHs). These final samples
include the self-employed. Although the purpose of exercises like this one is to measure the price
that the labor market assigns to different skills and that these prices are better captured by wages
than self-employment income, the latter depends on the skill of the recipient and including these
earners is especially important when trying to investigate the effects of international trade.
Several studies (e.g. Vélez, Bouillon and Kugler, 1999) have demonstrated that while selfemployment earnings display higher inequality than labor income, their dynamic behavior and
variation by skill level are very similar. The measure of earnings used is the monthly sum of labor
and self-employment incomes.
Finally, the trade, GDP and price data come from the annual statistical bulletin published by
DANE, while the exchange rate information was taken from the monthly magazine made
available by the Banco de la República (Central Bank). Exchange rate information is needed to
convert trade and GDP information to a common currency.
Methodology
Skill is proxied by education, potential experience and gender. Education was taken directly from
the ENHs and is measured in years. Experience was constructed as exp=min[(age-education-7),
(age-12)], where age is as of the time of the survey. I divided the population into 48 cells in the
urban case (2 genders times 6 education groups times 4 experience groups), and 24 cells for rural
areas (2 genders times 4 education groups times 3 experience groups). The urban educational
categories are uneducated, primary (1-5 years), high school dropouts (6-10), high school (11),
some college (12-15) and college or more (16 or more years), while the experience groups are 0-5
years, 6-10, 11-24 and 25 or more years of potential experience. For rural areas the corresponding
groups are no education, primary (same definition), high school (6-11 years) and college (12 or
more years) for education, and 0-10 years, 11-24 and 25 years or more for experience. Then, we
computed the relative labor supply provided by each cell (si) as the ratio of the number of
workers in the cell to the total number of observations for each year (weighted by the sampling
weights). Finally, the average supply for the 1978-98 period was computed for each cell. This
quantity is the fixed weight that we use throughout this section to compute earnings for more
aggregated groups, such as educational or gender groups. That is, earnings for those groups are
calculated using a fixed weight aggregation procedure, according to the following equations. The
fixed weight fi for cell i, i=1,…,n, n=48 or 24 is defined as
fi =
∑s
t
T
it
t = 1978, 79,..... with
∑
i
fi = 1
where T is the total number of years in the sample. Earnings (w) for the more aggregated group j
in year t in turn are given by
3
w jt =
∑
∑
i∈ j
f i wit
i∈ j
fi
Thus, in effect we are comparing earnings for a fixed skill composition of the population. In this
way we avoid confounding pure earnings movements with changes in the composition of the
labor force that may influence earnings behavior5.
Finally, a word related to trade, GDP and price deflation. The trade variable was created as the
ratio of net imports to GDP for the 23 two-digit sectors included in the ENHs. Then, each
observation was assigned an import coefficient based on its sector of employment. The trade and
GDP data were converted to a common currency using the average nominal exchange rate
(pesos/US$) for the year in question. Income was deflated to pesos of 1978 using the CPI for
each of the seven cities to cope with differential price movements across regions.
Evidence
We begin with the evolution of the distribution of labor earnings across (between) skill
categories. Table 4.1 and figures 4.1 and 4.1a summarize data on changes in urban earnings for
different skill groups of the population, distinguishing absolute (percentage change of mean
income of each group) and fixed weight changes6. The differences between these two measures
show what part of the changes was caused by shifts of the skill composition of the labor force.
Thus, the fixed weight measure is more accurate in reflecting pure earnings changes. For ease of
presentation, we further aggregated the educational groups into low education (uneducated plus
primary), medium (high school dropouts, high school and college dropouts) and high education
(college or more). The criterion used to choose these coarser aggregates was the observation of
how close the earnings of the finer groups moved through time. In the graphs every group’s
earnings are normalized to one in 19787. The tables and figures referenced above show that all
education groups improved their situation along the entire period. The gains for the low and
medium educated came mainly before the 1990s, while the highly educated lost over that period
but gained in the final decade. During the 1990s all educational categories enjoyed rises in
absolute earnings, the most for the highly educated group (38.7%), and the least for the low
5
We also computed average earnings for each cell and a relative earnings index for it based on the ratio of its average
earnings to the overall mean earnings. Then, we multiplied the average supply of each cell by this index, obtaining
what KM call supply in efficiency units, which amounts to a fixed weight aggregating scheme increasing the weight
of the richer cells. Results for both procedures will be shown when we feel that it is worthwhile. For the
most part, however, they will be reported for the first methodology
6
It is important to clarify that the graphs in this section were computed obviously using the 6 years of the
data (1978-88-91-94-96-98), but they were drawn for the entire period 1978-98 for presentation purposes.
7
We computed also a predicted change from a regression of the log of earnings on several covariates (47
dummies for the skill cells, education and experience within them, experience squared, sectoral affiliation,
occupation, 6 dummies for the cities, and profession). These predicted changes are quite close to the actual
ones for the 1978-88 and 88-91 periods, but then during the 1990s the regression lost power (the
predictions differ by a larger amount from the actual changes, and in many instances even the signs differ).
That is, starting in 1991 unobservables began to play an important role in earnings formation. Moreover,
the discrepancies are consistently larger for higher levels of skill, indicating that this phenomenon is
affecting the high skilled disproportionately.
4
education group (8.3%). However, this change in the 1990s occurred completely from 1991 to 94
for the two lower groups. Fixed weight changes were especially large among the low and high
education groups for the entire period (17.5 and 14.4%), and negative for the intermediate
category (-3.9%), while in the 1990s college individuals’ fixed weight earnings increased by
37.6%, and the low and intermediate groups individuals’ by 8.5%. Hence, we conclude that
earnings for the intermediate education group have eroded significantly, while the two remaining
categories have experienced growing earnings, but over different periods. One important point is
that the absolute and fixed weight changes are almost identical for the low education group, and
very different for the other two categories. In the case of the intermediate one, the fixed weight
measure is significantly lower than the absolute one, while the opposite is true for the highly
educated. This is the first indication of a point that will be central throughout the paper: supply
of intermediate education has been growing “too fast”, while that of high education is not
augmenting fast enough.
Table 4.1 reveals that the gains for the highly educated in the 1990s were larger among women.
While college educated women experienced fixed weight earnings growth of 55.7%, among men
this figure reached only 30.8%. In fact, for the whole period (1978-98), highly educated women
enjoyed the largest positive change in absolute earnings (38.5%). For the two remaining
categories the growth in fixed weight earnings was also larger among women. Absolute changes
for women between 1991 and 98 ranged from 19.5% (intermediate) to 59% (college). The female
group that benefited the least during the 1990s was the intermediate one. Among men the three
categories displayed almost identical absolute earnings changes for the entire period, while the
fixed weight measure shows the intermediate category as the clear loser, with a decrease of 5.7%.
Replicating the pattern observed in the full sample, highly educated women experienced all their
gains in the 1990s. Highly educated males began to experience larger earnings growth than the
rest of the population since 1988, but females’ increases during the 1990s were more than enough
to outpace them. Finally, the intermediate category among men and women endured falling fixed
weight and absolute earnings in most of the sub-periods. Thus, educational earnings differentials
followed similar qualitative patterns, but the magnitude of the changes was different for the
various categories. Figure 4.1b depicts the evolution of the fixed weight log differentials of each
category against college graduates for the period 1978-98, normalizing them to one in 1978. The
differentials were reduced between 1978-88, and then a monotonic increase started in the latter
year and was maintained until the end of the data, especially for the college-intermediate
differential. This graph reinforces the message expressed in the previous paragraphs: high-school
dropouts and completers were the groups that lost the most, while low educated individuals
actually gained on college ones (the differential in 1998 was slightly less than in 1978). Women’s
earnings grew more than men’s during the 20 years. Indeed, while men earned on average 63.3%
more in 1998 than they did in 1978, this figure was 85% for women. These quantities translate
into fixed weight changes of 1.6% for men and 15% for women. This also applies for every subperiod but 1988-91. Gender differentials by education level are depicted in Figure 4.1c. In the
educational category in which the gender differential fell the least was the intermediate one,
which had the lowest differential in the initial year. In the last year of the data, the gender
differential was practically indistinguishable between low and intermediate educated individuals,
and it had decayed by more than 0.2 log points among highly educated ones. This fall, however,
occurred almost completely during the period 1994-98.
Looking at the changes by experience groups, one notices that this variable did not play a
5
significant role in the evolution of labor income inequality. That is, changes in earnings for the
three groups presented in Table 4.1 do not show a detectable pattern, especially during the 1990s.
The only salient feature is a trend towards a deterioration of the relative position of the older
cohorts. Younger cohorts of women with a high school degree or less benefited the most during
the 90s. At higher levels of education the situation is the opposite: older cohorts enjoyed the
highest earnings increases in the 1990s. Among men the pattern was similar but the differences
between experience groups are not so easily inferred. Notwithstanding this, a generalized
earnings increase for younger males can be detected for the two lower educational groups, while
older workers tend to do better in the more educated categories.
The changes in absolute mean income for the entire sample were 71.7% along 1978-98 and
35.5% for 1991-98. This 71.7% breaks down to 31.9% in 1978-88, -4% for 1988-91, 31.2%
during 1991-94, -8% for 1994-96 and 12.2% for 1996-98. Looking at fixed weight changes and
thus removing the effect of shifts of the skill distribution, we find that it reached 5.7% for the
whole period, which occurred entirely during 1991-94. Indeed, from 1978 to 1988 fixed weight
earnings fell by 0.8% and a further decrease was observed in 1988-91 (-8.7%). During 1991-94,
the change reached 28.6% to fall by 9.2% in 1994-96 and by 0.2% in the final period8.
We summarize the evolution of labor income inequality in Figure 4.2, which shows the log
differentials between the 90th-10th, 75th-25th, 90th-50th and 50th-10th percentiles of its distribution
for the six years analyzed (normalized to one in 1978). Several important facts can be extracted
from this graph. First, inequality began to rise somewhere between 1988 and 1991. Moreover,
this pattern became steeper and affected the entire distribution between 1996 and 1998. Second,
the increase in inequality has been driven by a rising dispersion above the median and not by
relative income decreases in the left tail of the distribution (the first decile). Finally, reflecting the
fact that the high-school categories experienced large earnings reductions, the most substantial
increase in inequality occurred between the 25th and 75th percentiles. Table 4.2 displays
information about average earnings by decile, reinforcing the message that the highest and lowest
deciles have been the most favored (deciles 8, 9,10 and 1).
To end the discussion of the urban data, we proceed to study the evolution of within group
earnings dispersion. Up to this point we have been concerned with inequality between groups. To
study this component we ran a regression in each year of the log of earnings on 47 dummies
corresponding to the skill cells (47=48-1 reference cell), experience squared, its interaction with
gender and linear terms for years of education and experience within each cell. The residuals
from this regression capture inequality within each group, i.e. they capture the effect on labor
income of variables unrelated to the three skill dimensions used. Results are shown in Figure
4.2a, where the same log percentile differentials of the predicted earnings are depicted. Residual
inequality began to increase between 1988 and 1991 everywhere in the distribution, and this rise
continued until 1998. Furthermore, starting in 1996 the growth in residual inequality increased.
This indicates that not only has the relative situation of high school individuals worsened, but
also within this group differences have been growing, even more than the ones observed between
groups (measured by the evolution of mean earnings of each skill cell). To put these changes in
perspective, the rise of residual inequality has been larger than the changes that can be explained
by gender, education and experience combined.
8
The changes reported are very similar to the ones computed using efficiency-units weighting.
6
To conclude, we observe a mixture of equalizing and unequalizing developments in urban areas,
and two sub-periods that are clearly differentiated. We found a reduction in the differential
between the least educated and the college educated workers, an improvement of the relative
position of younger entrants to the market, and earnings gains for women. On the other hand, we
found a worsening of the relative position of intermediately educated individuals with respect to
the rest of the population (especially to college workers), and relative income gains for the more
experienced workers during the 1990s. Over time, 1978-88 was a period of reduced inequality
between groups, while after 1988 the opposite happened due to the rise in earnings of the highly
educated. Also, during the latter period experience differentials grew among the highly educated,
meaning that the unequalizing effect of this group was reinforced by this experience effect.
Now we turn our attention to rural areas, which we divided into 24 cells as explained above. In
contrast to urban areas, the predictive power of the regression used is poor (see footnote 6). This
indicates that the rural labor market is not as strongly governed by observable skill measures as
the urban one is. It seems that the price of labor in this market responds to other variables, such as
the conditions of the product markets, which are not captured by traditional skill analysis.
Table 4.3 shows the between component of earnings changes. The group with the lowest
education (0 years) fared better than the rest of the population, both for the entire period and
during the 1990s. The fixed weight change in earnings for this group reached 33% for the full
period and 4.5% for the last decade, in which they were the only ones who enjoyed positive
earnings changes. For the high school dropouts and graduates the reductions were close to 8 and
46%, and the highly educated (more than high school) experienced an increase in the whole
period of 33%, but a sharp reduction over the 1990s of 22%. Looking more closely, we find that
the uneducated individuals only suffered losses between 1988 and 91, while the highly educated
ones reduced their earnings in the period 1991-96. These fixed weight shifts translated into
absolute earnings gains over the entire period for the uneducated (36%), primary (8.5%) and
highly educated (6%), while the high school group ended up earning 4.2% less in 1998 than they
did in 1978. Within the uneducated group, the younger cohorts’ (0-25 years of experience)
enjoyed fixed weight earnings changes greater than those observed among older workers (23 vs.
15.5%). This constitutes a generalized pattern across educational groups of both sexes. Thus, the
older high school workers endured the greatest losses in earnings (almost 65%). It is not difficult
to infer that women did better than men all along the 20 years. Women experienced fixed weight
earnings changes of 37% for the entire period and of –5.1% during the 90s, while these figures
were 4.5 and –34% for men. In absolute terms men were earning 20% more in 1998 than in 1978,
and 36% less than in 1991. Women’s earnings were 50% higher in 1998 than in 1978 and 4%
smaller than in 1991. The developments related to education and experience are homogeneous
across genders. As a consequence, the groups who benefited the most during the 90s were young
women of any educational category (especially uneducated), older women with college
education, and young uneducated men. The left-behind groups were all the older men but the
uneducated, and within these, the high school ones endured absolute losses of more than 42%.
The absolute changes in mean income for the entire sample were 22% for the whole period and 31% for 1991-98. This 22% breaks down to 57% in 1978-88, 12% for 1988-91, -14% for 199194, -30% for 1994-96 and 14.3% for 1996-98. Fixed weight changes were of 9% for the entire
period and close to –30% in the last decade. Periods of positive growth were 1978-88 (43%),
7
1988-91 (10%) and 1996-98 (15%). From 1991 to 1996, the negative change reached 39%.
We summarize the changes of rural labor earnings with the evolution of the same percentile
differentials used in the urban case, presented in Figure 4.3. The period 1988-94 was one of
negligible changes in inequality as described in Santamaría (1999). Also, from 1978 to 1988 we
observe a deterioration of rural earnings, especially for the poorest segments of the population.
There was remarkable stability in the distribution of earnings above the median and below the top
10%, for the 90th-50th differential displayed virtually no change. Finally, starting in 1994
inequality increased significantly. It is evident that the lowest 10% of the distribution suffered
huge losses between 1994 and 96 (evidenced by the jumps of the 90th-10th and 50th-10th
differentials), and then they somewhat improved their situation during 1996-98. The 75th-25th
differential displays continued growth since 1994, reflecting the losses endured by the high
school educated individuals.
We wrap up this section with a brief comment related to the within component of inequality
presented in Figure 4.3a. It is evident that residual inequality followed the same pattern as the
observed one, i.e. the least lucky within the lowest deciles of the overall distribution (who are
among the older cohorts of high-school individuals), endured great losses during the period 199496, while the luckiest individuals in low deciles were able to at least maintain their situation.
Finally, what is clear from the discussion above is that the increase in inequality in rural areas is
not linked to rising prices of higher skill levels, as defined by education, experience and gender.
Possible Explanations
In this section we try to explain the changes in the distribution of labor income described above
using a simple model that highlights the interaction of supply and demand for different levels of
skill. The model I develop is in the spirit of the one proposed by KM. It assumes a concave
aggregate production function for each of J sectors that comprise the economy. The inputs are the
different skill types of labor, indexed by i (i=1,…, I). The production function is assumed to be
CES. Thus,
Yj =
(∑ a N )
i
1
γ γ
ij
ij
(1)
Where Yj is output in sector j, Nij is the number of individuals of skill category i employed in
sector j, aij is an unobservable technological parameter for skill group i in sector j, and γ is the
elasticity of substitution among skill groups. The marginal conditions for each sector are
p (∑ a N )
j
i
ij
γ
ij
1−γ
γ
aij N ijγ −1 − wi = 0
where pj and wi stand for the price of sector j output and the competitive wage for skill i,
respectively. The demand for skill group i in sector j is thus given by
8
 aij p j
N ij = 
 wi
1
 1−γ
 Y j

(2)
Differencing and assuming that real wages are fixed we get
 aij p j
dN ij = 
 wi
1
N
 1−γ
 dY j = ij dY j
Yj

(3)
Summing over j on both sides of (3) we obtain
∆Di = ∑ j N ij
dY j
(4)
Yj
The quantity ∆Di gives the weighted sum over sectors of changes in employments for skill i with
weights given by percentage changes in sectoral output. Thus, it measures the change in demand
for skill group i that would take place at fixed factor and output prices as a result of shifts in the
industrial composition of the economy, caused, for example, by technical change.
We follow the tradition in the literature (see for example Freeman, 1975, 1980; KM and Juhn,
Murphy and Pierce, 1992) and implement this index proxying the sectoral change in output by the
change in total factor inputs. The reason for doing this is twofold: first, I want to avoid
complications related to the correct value added measurement of sectoral outputs, and second it is
not possible to get information on sectoral outputs that matches the support of the index j (see the
paragraph after the next). Thus, the demand shift index is computed as
∆Di
=
Ni
∑
j
µ ij ∆N j
Ni
,
µ ij =
N ij
Nj
(5)
We measure µ ij as the average share of workers of skill group i in sector j over the entire period.
Additionally, we divide by Ni to express the change in demand relative to the fixed weight
employment share of skill group i in the base period (Ni is the average share of skill i in total
employment). Note that all quantities involved in (5) are equilibrium quantities and directly
measurable from the data at hand.
Initially j indexes 165 industry-occupation “sectors”, made up of 33 industrial divisions times 5
occupations (blue and white collar, domestic employee, self-employed and employer)9. The
9
The 33 sectors are: agriculture and hunting; lumber extraction; fishing; coal extraction; oil and natural gas
extraction; metallic minerals extraction; other minerals; manufactured food products, beverages and
tobacco; textile and leather industry; lumber industry; paper products and publishing; chemical products;
production of non-metallic goods; basic metallic products; other metallic products; other manufactures;
9
quantity in (5) corresponds to the fixed weight manpower requirements index proposed and
further developed by Freeman (1975, 1980). Intuitively, this index tells us that inputs employed
mostly in expanding sectors will experience increased demand, while the demand for factors used
in contracting sectors will fall. As can be seen in equation (3), when ∆Di N i is not computed
using fixed input prices, it gives a biased measure of true demand shifts. This is because under
the fixed input prices assumption skill groups that experienced real wage increases would show
higher growth in their relative demand than the one given by applying (5) directly. That is,
changes in wages affect the distribution of sectoral outputs. Since we work with changes in
sectoral output (or its proxy) that already occurred in the presence of non-fixed wages,
∆Di N i is biased. True demand shifts are generally higher for skill groups with increasing
wages, and lower for those that endured wage losses (for an alternative derivation of the index
measured in efficiency units see KM). Within the confines of our model we try to construct a
more exact measure of the true demand shifts by noting that with constant input prices (5) is the
correct measure, but since we are not controlling for changes in wages when we apply it, we are
really computing the following quantity (still assuming fixed output prices),
∆Di* = ∆Di −
1
1−γ
 dwi

 wi

 N i

(5a)
This comes from differencing (2) letting w vary, making some substitutions, and summing over j
on the final expression. Thus, for elasticities smaller than one, we underestimate the true demand
shifts for skill groups with rising wages. We implement this corrected measure when necessary,
making it clear in the text, using an estimated value for the elasticity of substitution that comes
from a regression specified below. If we want the true shift, for example, for a skill category
whose wages grew, we should add to (5) the second term in the r.h.s. of (5a). Total demand for
labor in skill category i is obtained by summing over j in both sides of equation (2)
Di = ∑ j
∂Di
=
∂wi
 aij p j

 wi
1



2 −γ
1−γ
i
(γ − 1) w
1
1−γ
Yj
∑ (a
j
(6)
p j )1−γ Y j
1
ij
(7)
Equation (7) tells us how the demand for a skill type varies with its own wage. We assume that
the supply of workers of each skill is predetermined at each point in time by past schooling, etc,
so the demand schedule becomes a wage determination equation, and shifts of the supply curve
move wages accordingly. For this system to be stable we require the quantity given by (7) to be
negative, which amounts to the requirement that the elasticity of substitution be less than one.
electricity and gas; water; construction; retail commerce; wholesale commerce; restaurants and hotels;
transportation and storage; communications; banking; insurance; real estate; public administration and
defense; health services; other social services; enjoyment services; domestic services, and international
organizations.
10
Finally, we assume full employment for each skill group
N i = Di , so
 ∆D dN 
dwi
= (1 − γ ) i − i 
wi
Ni 
 Ni
(9)
According to (9) percentage changes in own wages should be negatively related to percentage
changes in net supply. Thus, to test the hypothesis of whether the observed changes in earnings
are fully explained by skill supply and demand interaction I estimate equation (9) and consider
that the changes are explained by the model if the elasticity of substitution is less than one (i.e. if
the regression coefficient is statistically greater than zero)10. If that is not the case, we conclude
that the data are not consistent with such a model and some other factors (technical change, for
example) are playing a major role in the determination of earnings. We estimate (9) for each
period using three different demand shifts indices. First, the one given by (5) (the biased one),
and then we use equation (5a) with two possible values for γ chosen in an iterative manner. First
we recover the estimate of the elasticity from the previous estimation and re-estimate (9) using
the corrected demand measure computed with this recovered value. Then we repeat this exercise
once more utilizing the second estimation value of γ . Since testing the hypothesis of the
regression coefficient being greater than zero is not possible in practice (because the t statistic
really informs us as to whether the coefficient is statistically different from zero), the main
statistic we use is the 95% confidence interval for the estimated coefficient, which allows us to
infer with 95% of probability whether the elasticity of substitution is less than one or not.
Table 4.4 presents the results of estimating (9) for the three proposed scenarios. Equation (9) is
estimated using weights equal to the average share of each skill group over the 20-year period
The Table tells us that supply and demand shifts are able to explain the changes in wages for the
periods 1978-88 and 88-91, while they fall short during 1990s, especially 1996-98. During the
first 6 years of the decade the results are ambiguous. Additionally, the periods for which the
differences among estimation results using the alternative demand indices is greatest are 1978-88
and 1994-96, which correspond to periods in which the economy was subject to important
exogenous shocks, as explained below.
We complement Table 4.4 with Figure 4.4, which we borrow from KM with some modifications.
Each panel in this figure plots the percentage changes in earnings versus net demands for our 48
skill cells and for each sub period, and additionally we fit the regression line coming from
estimating (9) using the biased demand measure. Figure 4.4 confirms the results provided by
Table 4.4, but offers some additional information. First, in the period 1978-88 the supply-demand
framework does not explain the observed changes in earnings well. During that period very large
changes in the skill composition of the Colombian population took place, especially in
educational attainment and female participation rates. In addition, the economy went through a
10
More specifically, this test will indicate whether the observed changes in earnings are consistent with a
model of earnings determination in which demand is negatively sloped (gamma<1) and constitutes a wage
determination schedule, and supply is a vertical line predetermined by exogenous factors.
11
recession from 1981 to 84, followed by a severe adjustment program and an important recovery
during the period 1986-88, helped by the coffee boom that started in 1986. The second comment
related to Figure 4.4 refers to the 1991-94 period. The poor fit of the regression seems to be
driven by outliers, calling for a more detailed look. We finish with an additional exercise.
The estimations may be contaminated by spurious components coming for example from
technological progress or changes in the size of the economy (which most likely create linear
trends within cells). To cope with this problem, we re-estimate (9) in second differences. The
results are presented in Table 4.4a. Note the switch in the sign of the regression coefficients for
the periods 1991-94 and 94-96. Once the cell-specific trend is removed from the regression,
changes in net demand for the various skill levels are able to explain shifts in earnings for the
interval 1978-94. For the second part of the 1990s they are not able to do this. In summary,
changes in supply and demand for different skill levels are enough to explain shifts in earnings of
finely defined skill groups all along 1978-91, they do so in a weaker manner during 1991-94, and
they fail to explain earnings changes in the second half of the 1990s. Earnings changes in this
period are more likely explained by macroeconomic shocks and technical change. The latter is
consistent with the difference in the estimation results between the raw and detrended series.
Several robustness checks were carried out for this exercise. First, we estimated equation (3) of
the KM paper, which shows that changes in wages and net supply must negatively covary. We
estimated this equation using both efficiency units weighting (as KM do in their paper) and fixed
weights. The results confirm that earnings changes are explained by shifts in net demand over the
period 1978-91, while for the latest period of our data (1996-98), exogenous factors altered the
interaction among prices and quantities in the skill market significantly. Additionally, for the two
intermediate intervals, 91-94 and 94-96, the results weakly accord with the coefficients from the
first regression above. This observation, combined with the fact that for these two periods we
obtained different signs depending on the specification used in the previous estimation, lead me
to conclude that changes in skill composition and demand are not enough to explain earnings
changes in this period. Second, I carried the same three estimations for samples including only
people that worked more than 30, 40 and 48 hours a week obtaining exactly the same results (48
hours a week is the legal definition of full time in Colombia). Finally, I repeated the exercises for
hourly wages, exercise that marginally reinforced the results11.
We next look in more detail at the changes in earnings, supply and demand for the finer skill
groups. In Table 4.5 the first column shows the change in supply ( ∆S ), the second the demand
shift (5), the third the change in net supply ( ∆NS ), and the final column shows the change in the
log of fixed weight earnings. The variables are computed as
11
The reason why I do not use hourly wages as my variable of analysis is simply that there are too many
observations for which this variable is missing in the ENHs. Additionally and in contrast to the US, in
Colombia wages are set monthly and not hourly. I believe that monthly earnings are a better measure than
hourly wages of the price of labor in Colombia. Notwithstanding this, I repeated the majority of exercises
presented in the paper for hourly wages and the results are almost identical. I notice that they are somewhat
reinforced, especially the one related to the increases of wages of highly educated individuals in the 1990s.
12
N
∆S i = Ln i ,t
 Nt

N 
 − Ln i ,t −1 ,

 N t −1 
∆NS i = ∆S i − Ln(1 + ∆Di )
Table 4.5 shows that the supply of college-educated workers grew over the whole period,
especially among women, while that of uneducated and primary individuals shrank. One point
that will be crucial later is that the bulk of these shifts occurred between 1978 and 88. Looking at
the relationship between changes in earnings and in net supply for educational groups, we find
that they agree with the model discussed above, except for college and high school dropouts
(weakly for the latter). Demand for uneducated individuals fell over the entire period, but the
reduction in supply was so sharp that their net supply was reduced by 1.34 log points, allowing
their earnings to go up almost 30%. A similar pattern can be observed for primary educated
individuals. These developments were essentially the same for all sub periods. Demand growth
for high school workers can be characterized by sluggish (at best) all along 1978-98. At the same
time we observe a huge increase in their supply. Thus, the fall in high school graduates’ earnings
can be completely explained by the increase in their net supply. High school dropouts, on the
other hand, displayed a fixed weight earnings reduction of 5% in the presence of no change in net
supply. Finally, college individuals enjoyed growing demand in every period that accumulated to
more than 31% from 1978 to 1998. However, this was not enough to offset the rise in supply
commented above. Their earnings increased by 0.13 log points despite growth in net supply of
more than 130%. For the 1990s we can reproduce exactly the same comments, except for the fact
that net supply of high school dropouts fell abruptly. The negative association between net supply
and earnings for educational-gender groups for the entire period is exact for men (except college),
while among women it does not hold for high school, college dropouts, and college. Earnings of
female workers increased all along the 20 years, despite growth in their net supply. Hence, we
have two unexplained phenomena: the growth in earnings of college educated individuals and
women, which took place even when net supply was growing fast.
Next, we look at the changes in earnings and quantities for each sub period. For the period 197888, the negative correlation between changes in earnings and in net supplies is almost exact, even
for college graduates. The exceptions are men and women in the aggregate. For these groups,
however, relative changes in earnings (relative to the fixed weight mean) are in accordance with
the movements in quantities. The period 1988-91, as we saw, was one in which the tests carried
out above showed that the movements in earnings could be explained by changes in net supply.
The information contained in Table 4.5 displays a generalized fall in earnings for every group.
Notwithstanding this, relative earnings changes are in the right direction. Similar patterns can be
detected for the next two periods, 1991-94 and 94-96. In the first, the changes in earnings were
positive for all splits, but the relative changes show a general negative correlation with shifts in
supplies. The caveat is that the growth in net supplies for the highly educated were large, and
their positive changes in earnings were even larger, especially for women. This is the reason why
we found that net supply movements were not able to explain earnings changes in this period. In
the latest years all the groups show the changes in fixed weight and relative earnings going in the
same direction as changes in net supply.
Up to now the demand index (5) has been computed with j indexing 165 industry-occupation
cells. Thus, it reflects the shift in demand for different levels of skill that occurs between those
partitions due to changes in the cell composition of the economy. If we instead let j index only
13
the 33 industrial sectors and calculate (5) again, we will obtain the shift in demand that takes
place between them. The difference between these two measures gives us an estimate of the
demand shift that occurs within industrial sectors and between occupations, the “within” demand
shift index. These computations are presented in Table 4.6. The total measure shows that the rate
of growth of demand for the more educated workers decreased from 1988 to 1994 and
accelerated again during the second half of the 1990s. However, this increase in demand can be
best characterized as growing at a quasi-constant yearly rate. The yearly average of demand
growth for this type of individuals was close to 1.6% between 1978-88, and then decreased to
0.9% during the next six years. Starting in 1994 it went back to levels close to those observed in
the initial period. Thus, demand for college educated workers increased by 30% during the entire
period, and 12.5% during the 1990s. This increase in demand came mainly from the within
component, especially during the periods 1978-88, 91-94 and 96-98. This element accounted for
18 of the 31 points of the overall increase in demand for the entire period, and for 4 points out of
the 12 during the 1990s, due its negative growth of 1994-96. Thus, for the this three periods
demand for highly educated individuals came from within the sectors in which they were already
employed, shifting them between occupations, fact that constitutes evidence supporting the
presence of factor biased technological change favoring high levels of education. High school
individuals endured sluggish growth in demand from both the between and within components,
reinforcing the statement made before that it seems that their poor earnings performance is
explained completely by the behavior of net supply. Even demand for uneducated individuals
grew more during the 90s than the one for high school workers.
Primary educated workers experienced reduced demand mainly from the between component.
Uneducated individuals endured large reductions in demand for their services during the earlier
period, with the bulk of the change arising from the within component. This comes entirely from
female workers, reflecting the decrease of the relative importance of the domestic service
occupation. In effect, 18% of women were employed in this occupation in 1978 (31% of the
uneducated), while this figure was 9% in 1998, (23% of the uneducated). The same comments
can be extended to the primary educated women. Finally, demand favored women over men by 5
percentage points approximately over the entire period, but the shift in this variable was larger for
men during the 90s (0.1 vs. –0.6%). This last development took place entirely between 1991 and
1994, when demand grew 4% more for male than for female workers. The between component
was invariably negative for men all along the 20 years (except during 1991-94), but it was
compensated by positive within shifts in some periods, due to the movements observed among
highly educated men. For women, the situation was the opposite: positive between changes to
some extent offset by negative within shifts. Thus, demand for female workers increased because
expanding sectors used women more heavily than shrinking ones, but declining female intensive
occupations like domestic service partially offset this increase. The expanding sectors were
banking, insurance and commerce. The average share of female labor in the first two was 55%,
while in the latter it was 51%. The abnormal behavior of the period 1991-94 is explained by the
high growth of the male intensive construction sector.
Changes in Relative Demand Arising from International Trade Flows: In the last section we
learned that the growth in earnings of college educated individuals and women cannot be
explained by changes in relative supplies combined with shifts in demand arising from changes in
the industrial structure of the economy. Thus, in this part we examine whether flows of external
trade of goods and services can help us explain those developments. For this purpose we rely on
14
the factor content (FC) approach to the subject which, simply put, tells us that imports bring into
the recipient country the workers used to produce the goods, effectively adding them to the native
supply. Exports, of course, are thought to do the opposite (for detailed treatments of this
statement see for example Murphy and Welch, 1991; Cline, 97; Freeman, 98; Santamaría, 99;
Deardorff and Staiger, 88 and Krugman, 95). In this sense, this model constitutes a natural
extension of the one used above. Given the postulates of the FC model, the change in
endowments for different levels of skill arising from trade, can be computed using the same logic
used to calculate changes in relative demand due to shifts in industrial composition. Hence, we
need to multiply skill coefficients for each sector by net imports in that sector to obtain the
change in labor demand for the particular level of skill that occurs due to actual trade flows (note
that this is done at fixed wages and thus we need not to worry about changes in earnings affecting
demand). Keeping the notation used above, we measure the net supply of workers of skill level i
embodied in trade in any period by the following quantity (relative to the entire labor force with
the latter normalized in each year to sum to one). NM stands for net imports measured in dollars
 NM j
STi = ∑ j µ ij N j 
 Y
 j




(11)
The results of applying (11) are shown in Table 4.7. The implicit supply of workers of all types
has always been positive, i.e. Colombia has been a net importer of workers in the last 20 years.
As expected, implicit supply increased for every skill group between 1991 and 94, with the
aggregate increment larger than 200%. We also compute relative demand shifts due to
international trade. The measure used was proposed by Murphy and Welch (1991) and developed
by KM.
∆DTi = −
1
Ni
∑
 NM j

µ
N
ij
j
j
 Y
 j

 NM j
+∑ Nj
j

 Y

 j




(12)
There are three differences between (11) and (12). First, the supply measure is divided by the
average share of skill level i over the entire period to make the index reflect the change in
demand relative to the fixed weight distribution of employment. Second, (11) is multiplied by
minus one to turn it into a demand measure. Finally we add the second term, which corresponds
to the weighted overall trade deficit (the weights being sectoral employment distributions),
turning the index into a relative one (i.e. it subtracts from the change in demand for each group
the overall change in demand arising from the trade deficit). The calculations for equation (12)
are presented in Table 4.7a. That Table shows that the demand shifts are in the right direction to
explain earnings changes, but these shifts are too small to counteract the changes in supplies,
even in 1994, when the demand shift measure increased substantially for all groups12. We
12
Since trade flows in services suffer from measurement error, we computed the demand index restricting
the sample to the agricultural and manufacturing sectors to see if the measures computed on the entire
sample were biased downwards. These calculations show that indeed the demand indices increased for all
skill categories, but they still were very small and the order of magnitude of the numbers was not greatly
altered
15
conclude that demand shifts induced by external trade flows have not exerted an important effect
on earnings for the different levels of skill, although they have contributed to a small extent to the
widening of the educational wage differentials between highly and intermediately educated
workers. However, the additions to the native supply made by the implicit number of workers
embodied in such flows seem to be more important quantitatively and also display a qualitative
pattern compatible with the observed earnings changes. Hence, our next exercise will explore this
issue more deeply.
- Trade, Technology and Education: Here we explore the relationship between education,
technology and relative earnings on one hand, and trade, education and relative earnings on the
other. Thus, we need a simple method to test the presence of factor biased technical change,
meaning a flexible model that permits different factors to be affected in diverse ways by technical
change. For the second part, the model should incorporate the interaction between relative
supplies and earnings. I use the model proposed by Murphy, Riddell and Romer (1998, MRR
hereafter). Production is assumed to be Cobb-Douglas with two inputs, capital (K) and labor (B).
The latter is partitioned into skilled (H) and unskilled (L) with a CES aggregator. Thus,
[
]
1
γ γ
B = λ ( A(t ) H ) + (1 − λ )(B(t )L )
γ
[
(13)
Y = K α B (1−α ) ⇒ Y = K α λ ( A(t ) H ) + (1 − λ )(B(t )L )
γ
γ
(1−α )
]
γ
(14)
Through the functions A(.) and B(.) the productivity of both types of labor is affected differently
by technological progress. Using the first order conditions for profit maximization and letting
gi(t)=ln[A(t)] for i=A,B we get
W

H
ln  H (t )  = a + γ [g A (t ) − g B (t )] − (γ − 1) ln  
L
 WL 
(15)
Following MRR, we impose the restriction on (15) that the growth of the difference in
productivity changes for both inputs is linear. Thus,
W

H 
ln  H (t )  = a + γgt − (γ − 1) ln  
L
 WL 
(16)
for some constant g. If the second term on the r.h.s is not zero, then technical change affects each
labor type differently, i.e. their technological efficiencies grow at different rates over time. That
is, relative marginal productivities change over time, causing variations in relative returns.
Second, when the restriction implied by (16) is imposed we are testing the hypothesis that the rate
of growth of g(.) is linear, i.e. that this progress increases the productivity of one type of labor
more than the other’s in a constant fashion overtime, which is fully compatible with the evidence
found when analyzing demand shifts. For our technological inquiry, we follow MRR’s procedure
almost exactly, while for the trade part of the analysis we devise new ways to estimate (16). For
now, we turn our attention to the technology aspect.
16
To estimate (16) the fixed weight log earnings differential between highly and low educated
workers is used as the dependent variable. We define the highly educated workers as those
individuals who have a college degree or more. Unskilled workers are those having between 6
and 11 completed years13. We chose these groups for two reasons. First, we saw above that the
growth in earnings of highly educated individuals cannot be explained by shifts in net supply
alone, so our objective is to examine in more detail their earnings. Second, this framework is
especially useful for examining relative earnings, and high school workers are a good comparison
group due to its relative size and earnings beahvior.
The data come from of a series of urban ENHs covering 22 years between 1977 and 1998 mainly
for the month of September. We use the same criteria above to drop observations, divide the
sample into skill groups, and to compute fixed weight earnings for aggregated groups.
The results from estimating (16) are (small numbers below the coefficients are standard errors)
W

H
ln  H (t )  = 0.01 + 0.029 t − 0.476 ln  
0.007
0.201
L
 WL 
(16a)
Predicted and actual values are plotted in Figure 4.5. By inspecting Figure 4.5 we can see that
equation (16) does a good job of explaining the variation of the earnings differential between
college and high school workers, except for the periods 1979-82, 88-90 and the very last year
(1997-98). We have found important evidence telling us that changes in relative supplies
combined to constant growth in productivity differentials between these two groups are able to
explain the rise in relative earnings for most of the period under study. Figure 4.5a shows the
trend of the relative earnings series against the observed differentials. Two things are apparent.
First, the periods that require more attention are 1979-82 and 1991-93, when the actual
differential deviated from the trend downwards and upwards, respectively. Second, from 1983 to
1990 the slopes of these curves are very close. Even for the years 1993-98 if we plot a 3-year
moving average of the differential against the trend, the slopes are surprisingly similar. These
facts call for a closer examination of the time path of g(t). Note that (15) implies that g(t)=
gA(t)- gB(t) is
equal to
g (t ) =
1   WH
 Ln
γ   WL

H 
 + (γ − 1) Ln  − a 
L 

(17)
The results given in expression (16a) imply a value for the elasticity of substitution between
college and high school workers of 1.476, which we use to compute the implied time path of g(t)
according to equation (17), along with hypothetical values for that elasticity of 1.3, 1.65 and 3. In
Figure 4.5b we plot these paths accompanied by trend lines coming from regressions of each g(t)
on time. For all values of the elasticity the slopes of the actual g(.) and the trend are very close.
13
We do not follow the aggregation scheme used by MRR to differentiate H and L. See their paper.
17
Also, the implied paths of g(t) fit a straight line almost perfectly, which constitutes one of the
main results extracted from this exercise, because it provides evidence pointing to the existence
of a technological process that is augmenting the productivity of college labor more rapidly than
that of high school individuals, and that this increase in the differential is well captured by a
linear growth over the last 22 years, with small deviation from the trend from time to time. The
computed trends imply a yearly simple average growth of g(t) of about 2%. The second piece of
evidence that we extract from the figure is that there seems to be a period of decrease in the rate
of growth of g(t) from about 1984 to 1987/88, followed by catching up to trend level and then a
new short period of slowdown between 1989-91. The intervals in which g(t) grows faster than the
trend are 1978-80, 1992-94 and the last two years14.
Note the similarity of these results and those of MRR for the US over the years 1960-1995. They
found an elasticity of 1.42. Taking into account that the vast majority of Colombia’s external
trade takes place with the US, that foreign investment comes mainly from the US, and that the
dollar is the dominant currency for foreign exchange policy, this may tell us that both countries
have been subject to similar technical change and trade shocks.
We finish the technology part of the analysis by inspecting Figure 4.5c, in which we depict actual
and detrended changes in relative supplies of college versus high school workers. This variable
has grown at an average rate of about 3.8% per year over the entire period. The relative supply of
college individuals grew faster than average only during the periods 1978-79 and 1996-98. For
the years 1989-95 the rise in this aggregate was substantially below average, while the interval
1980-88 was characterized by a small downward deviation form the average pattern. These facts,
combined with the evidence found before regarding the behavior over time of the function g(t)
provide a compact explanation of the behavior of the college-high school earnings differential.
That is, if we turn back our attention to Figure 4.5, we notice that by estimating equation (16) we
are under-predicting the growth of the differential during 1979-80 and 1991-1993, which are
precisely the periods for which we observe the function g(t) growing faster, combined with the
relative supplies growing slower than average. The period 1982-1988, when the changes in the
differential are perfectly predicted, corresponds to an interval of close-to-average movements in
both g(t) and relative supply. Therefore, we can conclude that changes in relative supplies
combined with an almost constant rate of increase in the skill bias of technology towards highly
educated workers fit data well, and this allows us to explain the rise in relative earnings of that
group of workers. As MRR and KM note for the US, this can be also interpreted as technological
change increasing demand disproportionately for more educated workers in a constant fashion
over time. The results found here are fully compatible with the ones of the previous section
regarding relative shifts in demand (especially the “within” component) and the inability of a
simple supply-demand framework to explain the increase in college workers’ earnings.
14
We carried out a statistical test for changes in the growth rate of g (see MRR’s paper, pp 30-31 for the
exact specification of the test) It yielded significant changes for the growth rate in the years 1981, 92, 94
and 98. For 1981 and 94 the sign indicates downward deviations from the trend, while the opposite is true
for the other two periods. Thus, we conclude that the rate of growth of g(t) slowed down during the 80s,
accelerated between 1992-94 and then decayed in the latter part of the 90s until 1998, when it sped up.
18
Labor demand shifts induced by external trade flows also contributed, but to a lesser extent. The
trade part of the exercise is very simple. It consists of computing the supplies of college and high
school workers implicit in trade flows according to equation (11), and then adding these implicit
supplies to the actual supply series. We then calculate new relative quantity series and re-estimate
equation (16) to assess the effect of trade flows on changes in relative earnings from the
perspective of factor content theory. We start by analyzing Table 4.8 where we summarize the
changes in total and relative supplies induced by trade flows. In the first column we find the
adjusted total supply of workers with the domestic labor force normalized to one. Two comments
are in order. First, the effects of the opening of the economy can be detected only starting in 1993
when the adjusted supply went from 1.086 in 1992 to 1.132, and reached a maximum value of
1.167 in 1995. In the last part of the decade it stabilized around 1.13. Second, from the point of
view of labor quantities this liberalizing event has not been as important as believed. In fact, the
opening that occurred at the beginning of the 80s (1980-84), which is regarded in Colombia as a
very modest liberalization of import restrictions, appears to be more important quantitatively
because during those years the measure of total labor supply was consistently around 1.15. The
second and third columns of the Table show that the effect of trade flows on the relative supply of
college vs. high school workers has been very modest, although there is a change in the pattern
starting in 1992. The changes seem too small to explain any significant part of the changes in
relative earnings. The results of estimating equation (16) with the adjusted supply series are the
following
W

H
ln  H (t )  = −0.072 + 0.028 t − 0.527 ln  
0.007
0.217
L
 WL 
(16b)
Instead of analyzing the effect of trade by means of a graph comparing the time path of the two
series of predicted values, we present them in Table 4.9 along with the percentage changes in the
college-high school earnings differential. There we can see how the adjusted supply series moves
the predicted values closer to the actual earnings differentials, especially for the period 19911993. The accumulated growth in the actual differential during this two-year period was 24.9%,
while equation (16) estimated with the unadjusted supply series predicted an increase of 7.6%. If
we instead estimate (16) with the adjusted series the predicted rise of the differential is of 9%,
which leads us to conclude that the growth of net imports is responsible for approximately 5.6%
of the increase in the college-high school differential (5.6=(9-7.6)/24.9), figure which is in line
with the conservative estimates made for the US (see for example Borjas, Freeman and Katz,
1992 and 1998). In conclusion, in this section we used a simple framework to account for the
growth of relative earnings of the college educated, based on factor biased technological change,
relative supply measures and external trade. The contribution of this last factor was very modest.
Our last task is to provide at least a partial explanation of the rise in female earnings.
Gender: I argue that the bulk of the gains experienced by women are due to a reduction in pure
labor market discrimination. We found above that relative demand shifts coming both from
changes in the industrial composition of the economy, and from external trade flows display the
right signs, but once female supply is taken into account these are not enough to explain the
observed changes in the male-female earnings differential. Technical change cannot be
considered as a main factor to explain such changes due to the evidence offered by the within
component of demand shifts examined above. Thus, I estimate a simple reduced form equation to
19
test the hypothesis of reduction in discrimination as follows:
y i = β i xi + ε i
for i = male(m), female( f )
(19)
where y is the log earnings, x is a matrix of personal characteristics including education,
experience, hours worked, sectoral affiliation, dummies for age groups and city of residence and
profession (the “endowments”), and β is a vector of parameters that measures the “prices” paid
for those endowments to men and women. After estimating (19) for both genders separately, I
proceed to predict two wages for women according to the following equations
yˆ f = βˆ f x f
,
yˆ m f = βˆ m x f
(19a, 19b)
The first of these quantities (yf) simply measures the earnings for women as if the variables
included in x were able to account for the entire variation of female earnings, while the second
(ymf) constitutes a counterfactual wage measure for women, displaying how much they would
earn if they were paid at men’s prices. The difference between these two quantities can be
interpreted as discrimination in the sense that for the same observables we detect different prices
that depend solely on the sex of the worker. However, a word of caution is in order. Since the
measure of experience used here is potential and not actual, and taking into account that actual
experience is lower among women than men, the difference between these two measures may be
picking some of this effect as well. There may also be unobservable differences between male
and female labor that play a role in the disparity.
Figure 5.6 depicts the difference between these two predicted earnings values along the female’s
earnings support, partitioned into 5% percentiles for ease of presentation for three years of the
data (1978, 91 and 1998). We can see that the discrimination component, as defined here,
decayed all along the earnings range, but the fall was especially sharp within the first 50%. The
“discrimination profile” went from having a steep negative slope along the initial 50%, to a
practically flat line around 0.25 log points.
This exercise concludes the urban part of the analysis. Now we switch our attention to the rural
sector, which will be analyzed in a briefer manner than the urban for a number of reasons, such as
the lack of information to carry out many of the exercises, but more importantly because the
already mentioned fact that the relationship among skill and earnings is not as strong in rural
areas. Also, as we saw in the descriptive section the changes in the distribution that occurred in
the period 1988-98 were not of great magnitude and did not show identifiable patterns.
We start by estimating equation (9) for the 24 skill groups defined for the rural sample only for
the first two demand shift indices as explained in the urban section. The results are presented in
Table 4.10. As expected, the Table shows that supply and demand for the various levels of skill
do not by themselves account for earnings changes. Surprisingly, however, the only periods in
which the model displays at least moderate explaining power are 1991-94 and 94-96, which cover
the opening up of the economy, and this result is particularly strong for the first interval. As we
can see in the Table, the values obtained for the elasticity of substitution fall within the correct
20
range, and are significant. Now we implement equation (3) of KM’s paper, which will help us in
cleaning the data from cell specific trends related to technical progress, which may have played a
very important role, especially during the 80s. This exercise simply consists of computing the
inner product of the vector of earning changes times the vector of net supplies changes, measured
in efficiency units (see also footnote 9 in their paper for a detailed and formal explanation of the
procedure). We can see the figures obtained from this exercise in Table 4.10a. They show that
once cell specific trends are partially removed, the periods for which changes in wages may be
adequately explained by supply and demand interaction are 1978-88 and 1991-94. This finding is
important for two reasons. First, the 1980s were a period of rapid modernization for the
Colombian agricultural sector (see for example Bejarano, 1992), and in 1986 a medium sized
coffee boom occurred. Thus, once we remove these effects from the price-quantity measures we
see that that earnings moved according to shifts in net demand for the various levels of skill.
Second, the fact that movements in relative earnings can be explained by such shifts during the
first half of the 90s, is evidence against the popular view that the deterioration of labor earnings
in rural areas was due to the increase in net trade flows. Finally, the results obtained for the
second half of the 1990s are consistent with the increase in violence coming from various
sources, falling international prices for exports and the recession.
Table 4.11 replicates Table 4.5 for the rural case. We compute the demand shift index with j
indexing only the 33 industrial sectors, because the occupational distribution has remained
practically unchanged, with almost 80% of the sample being either a family worker or selfemployed. The negative correlation between net supply and wage changes is almost non-existent
except for the period 1991-94 and, surprisingly, for the entire period. Over this period the supply
of men decreased by 5%, while women’s increased by almost 25%, but this rise occurred entirely
between 1978-91. Indeed, in the 90s female labor supply fell by 15 percentage points. Demand
for female labor displayed the same qualitative behavior but showed smaller changes, while
demand for male labor did not change neither during the entire period nor within individual
intervals. As a result, supply of female labor increased by 14%, and male’s fell by 6%. Women’s
fixed weight earnings rose despite the increase in net supply (by about 25%).
There has been a move toward a more educated rural labor force. Uneducated workers reduced
their share by more than 35%, while the highly educated increased theirs by 185%. High school
individuals’ supply grew by more than 80% and primary ones kept their share constant. The
majority of these changes occurred in the initial period and between 1994 and 1996 the shifts go
in the opposite direction from the pattern observed in other periods. Demand growth was close to
zero in the aggregate. The college group experienced a rise in demand for its services of about
70%, but during the 90s demand for college-educated workers fell more than 90%. The two lower
categories experienced stable demand within every sub-period, while the high school workers
experienced demand shifts that were qualitatively similar to those for college workers. The
supplies of the two higher groups increased dramatically (75 and 135%), while earnings
decreased for the high school group and increased for college educated workers. Earnings for
uneducated and high school workers thus seem to be adequately explained by supply and demand
movements over the entire period. The supply of educated women grows more than men’s, the
reduction of the share of uneducated female labor is also greater, and changes in wages are larger
as well. Hence, we have again two unexplained wage changes: college individuals, whose
earnings move almost randomly, and women. We looked for evidence of shifts in participation
21
rates, occupation and sector, among other things, seeking some explanation. Instead, what I found
was that in 1991 there were a few educated earners who made unusually high amounts of money.
Demand Shifts Coming from International Trade Flows: Table 4.12 presents the aggregated
supply of workers of different skill levels implicit in trade flows, as measured by equation (11).
In contrast to the urban sector, the rural one has been a net exporter of workers of every skill
category, except college educated men, which in principle constitutes an additional fact that goes
in the opposite direction of what we expected given the rise of these workers’ earnings. The
qualification “in principle” means that the relative demand shifts that will be presented shortly
constitute more reliable measures than this raw supply index. In the aggregate and for every subgroup, the implicit supply of workers increased over time (i.e. it became less negative), except for
the latest period analyzed when it went down. Demand shift indices for the more aggregated
groups are presented in Table 4.12a. Two facts are immediately apparent. First, external trade has
had a negligible positive impact on demand for male workers, while its effect on female’s has
been negative and quantitatively more important than the one observed for men. Second, across
educational categories trade has reduced demand for college educated individuals and increased
that for uneducated workers. Demand for high school individuals has also been reduced and
primary’s has not been affected importantly. Therefore, (i) import competing sectors in rural
areas are low skill when compared to export ones (the opposite situation to the one detected in
urban areas), and (ii) movements in labor quantities arising from trade flows were not consistent
with wage changes for college workers (except between 1991-94), while the opposite is true for
uneducated and high school workers. Thus, we conclude from a pure demand-supply perspective
that if trade had any effect on labor earnings of this group, it was to partially counteract the
increasing pattern caused by other factors15.
DECOMPOSITION OF CHANGES IN THE DISTRIBUTION OF
INCOME
In the previous section we related the changes in the distribution of labor income in Colombia
during the last 20 years to the skill composition of the labor force. In doing this, we sought to
explain changes in the rewards to different skills in a supply-demand framework. In this section
we carry out a complementary exercise in which we isolate and quantify the contribution of
single factors to changes in the distribution of labor income over time, holding other variables
constant. That is, we construct counterfactual distributions to answer questions such as “how
would the distribution of income have looked at time t had some factor x remained at its t-1 level,
and workers had been paid according to the time t earnings schedule?”. This exercise uses the
methodology proposed by DiNardo, Fortin and Lemieux (1996, DFL hereafter), which is in the
spirit of more traditional methods that decompose changes in the moments of random variables.
Examples of such procedures include the Oaxaca methodology that quantifies the contributions
15
Since the growth of women’s labor earnings remains unaccounted for, we carried the same exercise
undertaken for the urban sector. The results show that there has not been a reduction in the discrimination
component of the rural earnings schedule. In fact, women belonging to the initial 50% of the income range
were subject to more discrimination in 1998 than in 1978, while those along the 50th and 70th percentiles
were about in the same situation. Thus, we close the rural section noting that the earnings gains
experienced by women in rural areas are not accounted for in this analysis.
22
of individual variables to differences in means (Oaxaca, 1973), and variance decompositions. The
DFL procedure generalizes these methodologies by decomposing changes in the entire
distribution. This is done by nonparametrically estimating the density function, and then
reweighting it to produce a counterfactual density. This reweighting adjusts the time t density for
the change in probabilities between the two dates of observing one particular characteristic for
each observation in the sample.
We start with the weighted kernel estimate of the density function of income Y:
1 n
Y − y 
fˆ ( y ) =
wi K  i

∑
nh i =1
 h 
(20)
The variables h and n are the bandwidth and the number of observations, and wi is the
(re)weighting function, which is the product of two elements ui and Ri. The first of these is just
the sampling weight, but the second is the key to the procedure.
Suppose income and personal characteristics (including education, gender, experience, marital
status, city of residence, profession and trade) have a joint probability distribution at some point
in time. Each individual can be regarded as a vector x=(y, z) consisting of income and another
vector z of personal characteristics, indexed by time. Partition z into two components, r and v,
where r is any individual element of z that is of special interest, and v is the rest of the elements
of z. We want to decompose the change in f(.) between two dates t=1 and t=2, in which the actual
densities are given by (20) with Ri=1 for all i. Since f ( y, x) = f ( y | x) f ( x ) , we can express the
actual density at time 2 as:
f ( y | t y = 2, t r|v = 2, t v = 2) = ∫∫ f ( y | r , v, t y = 2) f (r | v, t r |v = 2) f (v | t v = 2)drdv
(21)16
The l.h.s summarizes DFL’s notation that for the most part will be used here as well. It tells us
that the observed density in time 2 depends on the income schedule at that time (ty), the
distribution of the attribute r given the particular realizations of the other attributes contained in z
(tr|v), and finally on the distribution of the latter (tv).
We seek the distribution that would have been observed at time 2 if the distribution of r given v
had remained as it was in t=1 and people were paid as they were in t=2. Hence, we are looking
for f ( y | t y = 2, t r|v = 1, t v = 2) . From (21) we have
f ( y | t y = 2, t r |v = 1, t v = 2) = ∫ ∫ f ( y | r , v, t y = 2) f (r | v, t v = 1) f (v | t v = 2)drdv =
= ∫ ∫ f ( y | r , v, t y = 2) Rr|v (r , v) f (r | v, t r |v = 2) f (v | t v = 2)drdv (21a)
for Rr|v (r , v ) = f (r | v, t r|v = 1) f (r | v, t r|v = 2)
16
(22)
Note that this notation implicitly assumes continuous random variables.
23
The lower part of the r.h.s of equation (21a) is identical to the r.h.s of (21), except for the R(.)
function. If we can estimate (22), then we can compute the counterfactual density for
characteristic r. The function defined in (22) measures the ratio of the probabilities of observing
r, given the realizations of v at dates t=1 and t=2. Therefore, if r is a dichotomous variable
R r |v ( r , v ) = r
Pr(r = 1 | v, t r|v = 1)
Pr(r = 1 | v, t r|v = 2)
+ (1 − r )
Pr(r = 0 | v, t r|v = 1)
(22a)
Pr(r = 0 | v, t r |v = 2)
Equation (22a) is the relationship that we use to measure the contribution of attribute r to the
observed change in the densities. Once the first counterfactual density is constructed, we can pick
other elements of v and carry out sequential decompositions. Once this sequential exercise is
over, we want to measure the effect of the remaining elements of v as whole. That is, we seek the
density f ( y | t y = 2, t r|v = 1, t v = 1) . Applying the same logic,
f ( y | t y = 2, t r |v = 1, t v = 1) = ∫ ∫ f ( y | r , v, t y = 2) f (r | v, t v = 1) f (v | t v = 1)drdv =
= ∫ ∫ f ( y | r , v, t y = 2) Rr|v f (r | v, t r|v = 2) Rv f (v | t v = 2)drdv
for Rv (v) = f (v | t v = 1) f (v | t v = 2)
(23)
Applying Bayes’ rule to (23) we get
Rv (v ) =
f (t v = 1 | v) Pr(t v = 2)
,
f (t v = 2 | v ) Pr(t v = 1)
(23a)
which simply measures the weighted probability of an observation falling in year 1 or in year 2
given the characteristics of the workers. The weight is the ratio of unconditional probabilities of
an observation belonging to either date. The final step of the decomposition exercise accounts for
the effect that supply and demand factors had in the evolution of labor income inequality.
Following DFL, the procedure used to compute the supply-demand counterfactual density is to
estimate the earnings shift for each skill group according to equation (9), subtract this quantity
from the actual earnings in time 2, and then estimate the density of these hypothetical earnings.
We estimate (9) in second differences to eliminate cell-specific effects. Also, two modifications
are introduced to (9): first, we allow for a constant term and second, we do not restrict the
changes in supply and demand to have equal coefficient in absolute value. Thus, the supplydemand counterfactual once we have controlled for individual attributes is given by
f ( y | t y = 2, t r |v = 1, t v = 1, s = 1, d = 1) = f ( y − ∆y | t y = 2, t r|v = 1, t v = 1)
(24)
where ∆y stands for the estimated shift in earnings for each cell (dropping the subscript to save
on notation). We finish this methodological section by listing the elements of z and by briefly
explaining the estimation procedure of (23a). The vector z consists of years of education,
experience, gender, occupation, city of residence, number of years there, marital status,
profession, hours worked, trade, and type of job (temporary/permanent). Expression (23a)
measures the probability of being in the sample in years 1 or 2, given the distribution of personal
24
characteristics. To estimate Rv(r,v) we fit a logit model with the date as the dependent variable
and a vector with the elements of v as the independent variable, and then compute the appropriate
fraction. The ratio of unconditional probabilities is calculated by dividing the number of
observations in period 2 by that of period 1 (adjusted by the sampling weights). Finally, the data
set used for the labor earnings decompositions is equal to the one used in the previous section17.
Results: The objective of the first decomposition is to isolate the contribution of trade flows to the
changes in the distribution of earnings. We assume that the distribution of workers across the
various net import penetration levels, as defined in the methodological section, is a random
variable that covaries with labor income. We are not implying any causality between these
variables, the only implication of the exercise is the assumption stated above.
The import coefficient was turned into a dichotomous variable taking on the values 1 if the
individual worked in a sector in which the net import coefficient was greater or equal than 0.1
and 0 otherwise. Then we fitted a logit equation of this variable on dummies for the 48 skill
groups, years of education and experience within these groups, a quartic in experience,
occupation, type of job, city and years of residence, hours worked and profession. This estimation
gives us the probability for each observation of belonging to a high importing sector in date 1,
given its individual characteristics. With these probabilities we constructed the reweighting
function Rr|v (r , v) to estimate the trade counterfactual density for date 2. With the same
covariates we estimated the probability of being in either date 1 or 2 and computed the function
Rv (.) as explained above, and the reweighting function R(.) = Rr|v Rv was computed. Finally,
the supply and demand counterfactual was computed according to the procedure explained
before. We undertook decompositions for the 15 possible combinations of the 6 years of the data,
but report results only for 1988-94, 94-96 and 96-98.The first of these intervals spans the trade
reform.
Figures 4.7 through 4.7b show the results of the exercise for these periods. Graphs are read as
follows: in the first panel we plot the two actual densities for dates 1 and 2. The second shows the
actual density for date 2 and the trade counterfactual, while the third displays the trade
counterfactual density and the personal attributes one. The fourth panel shows the personal
attributes counterfactual and the supply-demand hypothetical density, and the last panel depicts
this last density and the actual one for date 1, i.e. the residual or unexplained change. Thus, if the
elements of z=(r,v) were able to account for the entire change, this last panel would show two
identical functions. Finally, vertical lines are drawn at the 5th, 25th, 50th, 75th and 95th percentiles
of the actual distribution for date 1 to better interpret the changes.
Figure 4.7 plots the decomposition of the change in the density of labor income from 1988 to
1994. The figure shows that the changes induced by the increased level of net imports were
concentrated in the left and right tails of the density, and along 50-90th percentiles of the original
distribution. The increase in net imports reduced the probability mass at the very left of the
support and increased it at the opposite end. Additionally, the mass located above the median and
below the 75th percentile grew, while that along 75-95% of the earnings range was reduced. This
suggests that higher levels of net imports made the distribution more equal, not less so as it is
17
The labor earnings DFL decomposition is carried out only for urban areas.
25
widely believed. These changes are consistent with the demand shifts induced by trade studied in
the previous section. In particular, demand for less skilled workers grew relative to that for more
skilled individuals as imports increased, while the demand shifts induced by this variable were
unfavorable to workers with intermediate levels of ability. Regarding the remaining attributes of
the workers, the graph shows that the contributions made by these during this period were
negligible. Supply and demand interaction, on the other hand, reduced the probability mass in the
very left, and also in the middle-high sections of the distribution, while increasing it substantially
between the 5th and 25th percentiles of the 1988 distribution, and also along 75th-95th percentiles.
Hence, supply and demand factors probably worsened the overall distribution in this period.
The period 1994-96 is shown in Figure 4.7a. Trade had similar effects to those observed over
1988-94. The contribution of supply and demand factors was practically zero over this period.
The only noticeable effect of these factors was a minimal increase in the mass located at the very
left of the income range, accompanied by an equally small reduction at the opposite side. This
confirms the results found in the previous section that pointed to the neutrality of supply and
demand for skill in the wage formation process during this period.
The final two-year interval is shown in Figure 4.7b. During these years (1996-98) the
contributions made by trade and personal characteristics were very small. The next to last panel
of the Figure plots the supply and demand counterfactual, in which we notice that these factors
had unequalizing effects. More particularly, they increased the mass along the 25-75th percentiles,
and at the top 5% of the income range. The decomposition exercise carried out for the period
1978-88 is not shown. However, over this period changes in personal attributes account for most
of the change in the densities, due to the fact that changes in their distribution were very large.
After 1988, changes in the distribution of personal characteristics have less influence, although
their shifts were still important, especially for the occupation variable. Developments of the main
covariates are presented in Table 4.13. In summary, for the first period analyzed the covariates
account for most of the observed change, while the opposite is true for the latter two intervals.
The increased level of net imports of the 1990s reduced inequality, because it favored the poorest
segments of the distribution, whereas supply and demand factors generally increased inequality.
These findings (and the evolution of occupation in the earlier periods) prompted us to carry out
more detailed decompositions that would be able to account for a larger share of the changes.
We performed several robustness for the exercise just discussed. The first consisted of inverting
the order of the decomposition. That is, we started with supply and demand, continued with
personal attributes excluding trade, and finished with the latter. Since the personal attributes
counterfactual density is now constructed before having controlled for trade, while the trade
counterfactual is estimated unconditionally, we need to compute two new reweighting functions,
Rv|r and Rr. This computations are straightforward by noticing that f(X|Y)f(Y)=f(Y|X)f(X) and thus
R v| r ( r , v ) =
Rr |v (r , v) Rv (v )
Rr ( r )
,
Rr ( r ) =
Pr(t r = 1 | r ) Pr(t r = 2)
Pr(t r = 2 | r ) Pr(t r = 1)
Since the trade variable was turned into a dichotomous one, the function Rr, is a simple ratio of
individuals employed in high importing sectors, to those working in low importing ones. The
results were robust to the order of the decomposition, noting that the contributions made by
26
supply-demand and trade were slightly lower in every period (especially 1988-94) when the
reverse order was applied. This shows that when we remove changes attributable to personal
characteristics and trade, the effects of supply and demand are magnified because we allow the
counterfactual to pick up just the effects of those variables. That is, when supply and demand are
analyzed first, their action may be confounded with that of other attributes, especially education,
gender and experience. The slight reduction of the contribution of trade shows one point that has
been mentioned already, which is that it is very likely that trade and supply and demand for skill
interact in some way, especially in 1988-94. The second set of checks is related to the variables
included in v, which were changed and combined in several different ways. Next, we tried
various alternative estimation methodologies for the reweighting functions. For example, we
estimated the probability of an observation falling in a high importing sector by fitting a logit
equation, a probit one and nonparametrically. The latter means that I divided the sample in very
fine cells according the elements of v, and computed the proportion of high importing
observations in each cell at time 1, and then used this proportion to compute Rr|v. All the
decompositions were quite robust to the choice of covariates or the estimation methodology,
excepting most of the exercises involving the year 1978. Additionally, the exercises for this year
produced very jagged densities, and for these two reasons they are not shown. Finally, I carried
out the main exercise for hourly wages, obtaining exactly the same results.
The results discussed above suggest a role for the recession that started in 1996, because the
decomposition showed that personal attributes and supply-demand for different skill levels were
not able to account for most of the changes in the distribution of earnings. That is, the increase in
inequality may be partially explained by factors that lie outside the labor market. The presence of
factor biased technical change favoring the more skilled was shown in the previous section to be
the main factor lying behind their relative earnings performance, which combined to
macroeconomic shocks that affect the earnings and employment of the poorer groups may be
responsible for the increase in inequality. We sought for a decomposition exercise that would be
able to partially capture these effects. Also, we wanted to extract more r variables from v to check
if the lack of explanatory power of personal attributes is the result of aggregating the effect of
several characteristics into one overall function. Maybe some of the elements of v taken
individually interact with the outside shocks in different ways and explain a larger share of the
change in inequality (for example, education may play an important role due to its power to
isolate the highly educated from the effects of macroeconomic shocks).
Therefore, our next exercise consists of a more detailed sequential decomposition with
employment as the first step, proxying the macroeconomic shocks. We do this by incorporating in
the sample all active workers to compute the probability of being employed at date 1, given the
other personal attributes for each observation. This quantity is then used to calculate the function
Rr|v according to equation (22a). Next, we sequentially extract elements from v and compute
reweighting functions based on the probabilities of these variables taking on one particular value
at date 1. The r variables chosen are, after employment, trade (same definition as before),
education (11 or more years=1), occupation (self-employed=1) and gender. Then, with the
remaining elements of v we calculate the personal attributes counterfactual in the same way as
above. Finally, the supply and demand counterfactual is computed. In this exercise we are
assuming that wages and employment have a joint probability distribution. Under this assumption
we view unemployment as excess supply of workers that tends to drive earnings down. The
procedure is related to the Heckman two stage method (Heckman, 1976, 79).
27
The results from this exercise are presented quantitatively rather than graphically. From the
various estimated densities we computed the Gini coefficient (g) and several log percentile
differentials. In this way we quantify the contribution of each factor in isolation, and we keep the
nonparametric nature of the procedure The computation of various differentials allows us to study
the effects of each variable on different segments of the distribution. The differentials used are
the 90th-10th and 75th-25th. Since once we carry the decomposition we have the density values for
each point on the earnings support, it is straightforward to compute those quantities as
∞
g = 1 − 2 ∫ F1 ( y ) f ( y )dy where F1 ( x) =
0
x
1
yf ( y )dy and µ is the mean wage.
µ ∫0
The qth percentile of the distribution is given by the value of w for which the following is true
w
q = ∫ f ( y )dy
0
The results of this exercise are presented in Table 4.14 for the same three periods analyzed
before, and using the same format used when we presented the results graphically. Each cell
contains the change in the relevant measure of inequality between the two densities involved.
Below this number the percentage contribution of the factor to the overall change can be found.
The last column adds up the individual contributions.
Inequality rose along every segment of the distribution over 1988-94, especially for the 75th-25th
differential. The decomposition exercise explains 89, 128 and 71% of the change in the Gini, and
the 90th-10th and 75th-25th differentials, respectively. The factors that made positive contributions
to the increase in inequality were education, occupation and supply and demand. Employment
and trade worked to reduce earnings dispersion. The equalizing effect of employment was
especially large for the tails of the distribution, resulting from the fact that during this period the
uneducated were the only ones who endured growing unemployment. Trade flows had a strong
equalizing effect involving also the top and bottom 10% of the distribution, while they slightly
increased inequality between the lowest and highest quartiles. As a result, their contribution to the
change of the Gini coefficient was equalizing but small (this index is more sensible to changes in
the center of the distribution). Education was the strongest unequalizing factor everywhere in the
distribution, finding that accords with the results of Vélez, Bouillon and Kugler (1999) and
Sánchez and Núñez (1998, 1998a). As strange as it may seem, this result arises because the
educational earnings differential grew in favor of college individuals, at the same time that the
supply of high school educated workers kept growing. This behavior could be summarized as the
country having too many high school educated individuals. Occupation during this period was
associated to a rise in inequality concentrated in the tails, pointing again to a shift of the highly
educated and female labor towards highly paid occupations, such as white collar, and away from
low paid ones such as domestic service. However, occupation reduced the 25th-75th differential,
which is again related to the shifts observed in the female occupational distribution shown in
Table 4.13 and analyzed before. Regarding the gender variable, the shift to a larger share of
female labor contributed to reduce inequality. The remaining personal attributes (city and years of
residence, sector of employment, experience, type of work and profession) significantly reduced
28
the 90-10th differential. We ran regressions of the log of income on these variables and human
capital ones for both years and interpreted the parameters of the sectoral and city dummies as the
respective premia paid by each sector and city. The regional variable did not play an important
role. Changes in sectoral premia are shown in Table 4.13 compared to retail commerce, which is
the most numerous sector. We see that all premia but three were reduced during this period.
Additionally, note that the sectors that display the highest premia happen to be either the most
protected or non-tradable, and all of them reduced this premia after the opening of the economy.
This is an additional channel through which trade contributed to reduce earnings inequality.
Finally, the interaction of supply and demand for skills had unequalizing effects on earnings,
which were magnified for the measures that included the highest ability.
In the next period (1994-96), Table 4.14 shows that the decomposition exercise predicts increases
in inequality, irrespective of the measure used. However, the data displays the Gini coefficient
and 75th-25th differential decaying and the 90th-10th growing significantly. Occupation and supplydemand did not play any role in this period, while employment, trade and gender contributed to
inequality reduction. Education and personal attributes contributed to increase inequality. The
effect of personal characteristics was particularly strong in the tails, while that of education was
concentrated in the 75th-25th differential, but the equalizing effect of employment and gender was
large for this differential as well, offsetting the education effect. The fact that personal
characteristics acted in the direction of increasing inequality is an indirect reflection of the
downturn of the economic cycle, because the evolution of the different covariates presented in
Table 4.13 shows very modest developments and in the direction of reducing dispersion.
The final period (1996-98) is one of increasing inequality for all measures. The decomposition
picks up this pattern correctly, and over explains the increase of the Gini and of the 75th-25th
differential, while it accounts for more than 81% of the rise of the 90th-10th differential. The
equalizing effect of employment and trade is weaker, which results from unemployment
increasing more in this period for the highly educated segment than over any other interval. Also,
employment for the least educated increased. Regarding trade, net imports went down between
these two years. The unequalizing effect of education was weaker as well, reflecting the increase
in the growth rate of the supply of college individuals. Occupation again displayed unequalizing
effects, except for the 75th-25th differential, probably reflecting the increased share of selfemployment. Personal attributes and supply-demand had negligible effects.
In summary, the effect of the employment on inequality was always equalizing. Although this
may seem at first a trivial result, it goes to the bottom of the problem: demand for labor in urban
Colombia is becoming everyday more skill oriented, even during 1988-94 when the overall
unemployment rate fell from 9.8 to 7.4%. Although the supply of highly educated workers is
growing, except in the period 1996-98, it is doing so at a very slow rate. At the same time, the
supply of intermediate education keeps growing at a very fast pace. Second, it is clear that trade
did not have the adverse effects on the distribution of labor income that it is associated with. This
finding is consistent with the way imports increase demand for uneducated workers, while
reducing it for high school workers. This equalizing effect also is related to the fact that they
favor female labor demand. The methodology used in this section is different from the one used
in the previous section, and they yielded identical results pointing to trade lowering inequality.
However, trade may have other effects ignored here, among which we mention input-output
effects and an interaction with technical change (although this is partially captured by the
29
decomposition). We saw that the latter seemed to have sped up around the same time in which the
reform took place (1992/93), indicating a possible relationship between the two. Finally, the
evidence regarding occupation gives additional support to the technical progress hypothesis18.
We finish with an additional exercise intended mainly for testing the presence of factor biased
technical change. We run weighted regressions of the log of income on human capital variables
plus city of residence and hours worked for each of the four years used (1988, 94, 96 and 98).
The weights are given by the different R(.) functions times the sampling weights. Then we predict
earnings and residuals for each year. Thus, this exercise permits us to study the effect of each
covariate on the evolution of observed inequality (measured by the standard deviation of
predicted log wages), and unobserved dispersion (standard deviation of log residuals), which may
be thought of as measuring the “between” and “within” components of inequality. Additionally,
the second component is a good indicator of the presence of unobservable forces that affect
earnings, such as technical change. These quantities are presented in Table 4.15 in the same
format of Table 4.14.
The first important point is that over the earliest and latest time intervals both observable and
residual inequality increased, while during 1994-96 the rise only occurred in observable
inequality. In the section in which we analyzed technical change in detail we found that this
process slowed down around 1994/95, after having sped up two years before. The fact that
residual inequality decreased between 1994 and 96 is completely consistent with this finding.
More importantly, the Table shows that the decomposition exercise lacks any explaining power
for the residual component of inequality, except in the period 1994-96. Thus, only during this
period the variables used in the decomposition, which comprise practically all measurable factors,
are able to account for the changes in this component. In the remaining periods some other factor
is driving it. The second point is related to employment. The Table shows that the equalizing
effect of this variable during the period 1988-96 was restricted to the residual component,
meaning that observed inequality increased as a result of shifts in employment. Therefore, in the
latest period the way in which this variable influenced the distribution of earnings changed as a
result of the huge rise in the unemployment rate, especially among high school and primary
workers. Additionally, in the earlier periods (1988-96) employment contributed to increase
educational differentials. The next point has to do with trade and its relation with educational
differentials. As we discovered in the technology/education exercise, trade contributed to the
increase of the college-high school differential. In this decomposition exercise we find additional
proof of that fact. We see that trade accounts for 44 and 59% of the increase of the college/rest
differential. This figure is higher than the one found above, which leads me to conclude that trade
may be responsible for 30-40% of the increase in educational premia during the first half of the
1990s. If we analyze the education variable we see that, as a rule, the shifts of this variable
increased measured inequality and reduced unexplained dispersion. This point is another way to
see that the increases in the relative supply of college educated individuals is less than the one
needed. Finally, supply and demand factors are an important explaining factor for the observable
18
Again, we carried out several robustness checks. They were to change the order of the decomposition
between the r variables and to use alternative estimation methodologies for the different probabilities. The
results were robust, excepting again the decompositions involving 1978, and in some instances the results
regarding occupation in the period 1994-96 were not very robust. Also, we repeated the 1998-94 period
using hourly wages with no change in the conclusions.
30
component of inequality (especially in 1988-94), which is consistent with the results found all
along the paper.
CONCLUSION
This paper starts by investigating the evolution of the distribution of labor earnings from a skill
perspective. We confront the problem from two perspectives. First, using a simple supplydemand framework for different skills to evaluate the role these played, and how they affected the
distribution of labor earnings at different points in time. Skill is proxied by education, experience
and gender. Second, we relate income to several demographical, skill and labor market factors, to
quantify their effect on the evolution of that distribution along the entire income range, and on
other measures of dispersion and on skill differentials. This approach is a semi-parametric
technique to decompose changes in density functions over time.
In urban areas, we found a reduction in the differential between the least educated and the college
educated workers, an improvement of the relative position of younger entrants to the market, and
earnings gains for women. On the other hand, we found a worsening of the position of
intermediately educated individuals, and relative income gains for the more experienced workers
during the 1990s. Over time, 1978-88 was a period of reduced inequality between groups, while
after 1988 the opposite happened due to the rise in earnings of the highly educated, and of the
more experienced workers among the highly educated. In rural areas, the main conclusion was
that the increase in inequality was not linked to rising prices of higher skill levels.
The microeconomic model used provided a compact explanation for most of the changes in urban
areas, excepting the rise in earnings of female workers. More particularly, it was shown that the
interaction of supply and demand for skill fully explained the behavior of earnings for all
educational groups, but the college one. The secular worsening of the relative position of the
intermediate group (high school) was explained completely by poor demand performance
combined to increases in relative supply. To account for the increase in earnings of highly
educated workers we introduced in the model factor biased technical change, which was
suggested by the behavior of this group’s demand. We extended the model to permit different
factors to be affected in diverse ways by technical change and use the college-high school
earnings differential as the quantity whose evolution is to be explained. This exercise led us to
conclude that changes in relative supplies combined with an almost constant rate of increase in
the skill bias of technology towards highly educated workers explain the rise in relative earnings
of that group of workers. This can be also interpreted as technological change increasing demand
disproportionately for more educated workers in a constant fashion over time.
International trade was shown to exert an equalizing effect on the overall distribution of labor
earnings, because the demand shifts induced by trade were such that demand for less skilled
workers grew relative to that for more skilled individuals as imports increased. This equalizing
effect also is related to the fact that they favor female labor demand. However, trade contributed
to increase the college-high school earnings differential. Finally, I argued that the bulk of the
gains experienced by women were due to a reduction in labor market discrimination, and
estimated reduced form earnings equations for men and women, showing that indeed
discrimination is able to explain at least 70% of the gains made by women.
31
The semiparametric decomposition exercises carried out lend support to the above conclusions,
but also provided two important additional pieces of evidence. First, the unprecedented increase
in unemployment that started in 1996 has contributed to reduce inequality because it has
eliminated low paid individuals from the distribution. This conclusion goes to the bottom of the
problem: demand for labor in urban Colombia is becoming everyday more skill oriented, and
additionally unemployment affects in a stronger fashion high school workers. The second point is
related to education and its distribution across workers. As a rule the shifts of this variable
increased measured inequality and reduced unexplained dispersion. This point is another way to
see that the increases in the relative supply of college educated individuals is less than the one
needed.
32
APPENDIX
TABLE 4.1: CHANGE IN AVERAGE INCOME FOR AGGREGATTE GROUPS - URBAN
78-88
88-91
91-94
94-96
96-98
91-98
78-98
Groups
Abs.
Fw
Total
31.9%
-0.8%
Men
25.7%
-2.9%
Women
36.2%
4.2%
Low
13.8%
16.9%
7.5%
-0.7%
Medium
Abs.
Fw
Abs.
Fw
-4.0%
-8.7%
31.2%
28.6%
-2.0%
-6.7%
28.5%
27.2%
-6.3% -12.8%
37.5%
-6.3%
-7.4%
16.9%
Abs.
Fw
Abs.
Fw
Abs.
Fw
Abs.
Fw
-7.9%
-9.2%
12.2%
-0.2%
35.5%
16.6%
71.6%
5.7%
-8.4%
-9.6%
12.5%
-2.4%
32.6%
12.2%
63.3%
1.6%
31.9%
-6.5%
-8.2%
12.8%
4.6%
44.9%
26.6%
84.9%
15.0%
17.4%
-4.8%
-6.1%
-2.7%
-1.5%
8.3%
8.5%
17.7%
17.5%
-8.9% -10.7%
29.5%
27.6% -12.8%
-13.7%
2.4%
-1.5%
15.6%
8.5%
13.9%
-3.8%
High
-22.9% -11.7%
-5.1%
-5.8%
36.7%
39.5%
-2.3%
-3.8%
3.8%
2.5%
38.7%
37.6%
7.0%
14.4%
0-10 exp
25.5%
-3.9%
0.5%
-5.7%
24.5%
24.6%
-3.1%
-3.2%
10.9%
0.6%
33.8%
21.3%
68.9%
10.1%
10-25
34.9%
-1.3%
-4.2%
-8.7%
33.6%
30.8% -12.8%
-17.1%
17.0%
3.7%
36.4%
15.8%
76.2%
4.4%
25 +
26.6%
3.2%
-6.3% -11.4%
32.8%
29.3%
-5.2%
-6.8%
4.3%
-6.7%
31.2%
13.0%
55.7%
3.2%
16.1%
12.3%
-4.4%
-5.7%
14.0%
15.2%
-4.6%
-6.5%
-5.4%
-5.1%
3.0%
2.3%
14.2%
8.3%
8.6%
-1.1%
-7.5%
-9.8%
26.7%
24.6% -10.9%
-11.7%
1.6%
-3.9%
14.6%
5.7%
15.0%
-5.7%
-13.6% -15.3%
-1.3%
-1.9%
42.1%
41.1%
-8.2%
-8.6%
2.9%
1.4%
34.2%
30.8%
14.4%
8.7%
43.0%
Men
Low
Medium
High
Women
Low
Medium
High
22.7%
29.6%
-10.3% -11.3%
22.3%
22.9%
-3.7%
-5.3%
4.9%
7.0%
23.5%
24.5%
35.8%
9.6%
-0.1%
-9.6% -12.4%
34.9%
33.8% -16.1%
-17.6%
5.6%
3.5%
19.5%
14.1%
18.5%
-0.2%
-2.1% -10.7% -14.9%
36.0%
35.3%
9.8%
4.6%
4.8%
59.0%
55.7%
38.5%
29.7%
-2.4%
11.7%
33
TABLE 4.2: AVERAGE INCOME BY DECIL - URBAN
Decil
78
88
91
94
96
98
1
2
3
4
5
6
7
8
9
10
1,018.8
1,894.6
2,379.8
2,717.8
3,005.5
3,480.6
4,127.7
5,345.0
7,494.1
18,603.3
1,398.6
2,648.9
3,116.7
3,436.2
3,793.0
4,467.3
5,221.6
6,594.9
9,171.9
22,221.1
1,248.5
2,400.5
2,821.6
3,114.0
3,557.2
4,194.5
5,023.2
6,266.3
8,900.6
22,063.8
1,553.3
2,782.2
2,961.5
3,400.1
4,103.0
4,906.5
5,877.2
7,797.2
10,827.9
33,948.9
1,487.7
2,700.9
2,980.2
3,455.7
3,976.9
4,756.6
5,786.8
7,338.7
10,785.8
28,368.7
1,452.7
2,643.2
3,002.7
3,377.8
3,984.6
4,774.6
6,101.9
8,177.5
12,554.9
34,409.5
Total
4,666.7
6,155.7
5,911.8
7,757.8
7,141.1
8,010.3
TABLE 4.2A: CHANGES IN AVERAGE E ARNINGS DECIL - URBAN
Decil
78-88
88-91
91-94
94-96
96-98
91-98
94-98
78-98
1
2
3
4
5
6
7
8
9
10
37.3%
39.8%
31.0%
26.4%
26.2%
28.3%
26.5%
23.4%
22.4%
19.4%
-10.7%
-9.4%
-9.5%
-9.4%
-6.2%
-6.1%
-3.8%
-5.0%
-3.0%
-0.7%
24.4%
15.9%
5.0%
9.2%
15.3%
17.0%
17.0%
24.4%
21.7%
53.9%
-4.2%
-2.9%
0.6%
1.6%
-3.1%
-3.1%
-1.5%
-5.9%
-0.4%
-16.4%
-2.3%
-2.1%
0.8%
-2.3%
0.2%
0.4%
5.4%
11.4%
16.4%
21.3%
16.4%
10.1%
6.4%
8.5%
12.0%
13.8%
21.5%
30.5%
41.1%
56.0%
-6.5%
-5.0%
1.4%
-0.7%
-2.9%
-2.7%
3.8%
4.9%
15.9%
1.4%
42.6%
39.5%
26.2%
24.3%
32.6%
37.2%
47.8%
53.0%
67.5%
85.0%
Total
31.9%
-4.0%
31.2%
-7.9%
12.2%
35.5%
3.3%
71.6%
34
TABLE 4.3: FIXED WEIGHT RELATIVE CHANGE IN AVERAGE INCOME FOR AGGREGATTE GROUPS - RURAL
Groups
78-88
88-91
91-94
94-96
96-98
91-98
78-98
Actual Predict Actual Predict Actual Predict Actual Predict Actual Predict Actual Predict Actual Predict
Total
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
0.0%
Men
6.8%
1.7%
0.6%
1.0%
-6.9%
-2.0%
0.4%
-1.5%
-1.6%
1.0%
-5.6%
-2.9%
-2.0%
0.0%
-25.6%
-6.3%
-2.4%
-3.9% 26.1%
7.4%
-1.6%
5.7%
6.0%
-4.0% 21.3% 10.9%
7.7%
0.0%
-3.2%
8.9%
2.5%
6.0%
1.6% 14.3% 10.9% 29.2% 15.8% 21.4% 16.6%
1.8%
-7.1%
-1.7%
-0.2%
-1.0%
-4.0% 14.7%
3.8%
-7.1%
0.5%
Women
Uneducated
50.8%
3.9% -37.5%
Primary
-11.5%
3.5%
Hs
-16.2% -15.1% 20.5%
6.4%
College
-41.2% -28.8% 26.9% 12.0%
0-10 exp
-15.3%
6.1% -15.7%
10-25
-13.3%
0.7% 14.7%
1.9%
-3.9%
-2.0%
1.9%
25+
18.3%
-2.9%
-0.5%
-7.4%
-3.4%
-8.0% -11.8%
-9.8%
-8.0% -15.9% -26.2%
-2.6%
2.0%
9.5%
-2.9%
2.7% 10.2%
8.6%
0.2% 10.5%
8.6% -14.7% -23.6%
-5.9% 24.5%
3.1%
6.9%
9.3%
-3.6%
-6.5% 21.1%
7.7%
-8.8%
0.6%
2.7%
2.4%
7.8%
5.6%
-7.4%
-2.2%
-5.9%
-6.9%
-3.7%
-0.4% -13.0% -11.0% -15.3% -14.1%
1.6%
8.4%
7.6%
9.1% 14.4% 13.4%
Men
Uneducated
66.8%
3.9% -40.7%
0.5%
7.8%
0.1%
4.1%
3.3% 20.7%
4.5% 18.1% 10.1%
Primary
-6.8%
6.2%
1.1%
-9.7%
-3.5%
-0.2%
-1.9%
-5.4%
1.7% -11.3%
-4.0%
0.5% -18.8%
2.0%
-0.8%
-4.2%
-2.5%
-4.3% -18.0%
-7.2% -13.7% -24.2%
4.6%
Hs
-18.8% -20.0% 43.2%
College
-34.5% -25.4% 24.5% 12.8% 21.7%
1.5% 11.2%
-5.5%
2.7%
-3.6% -15.0%
-5.4% -14.4% -12.3% -22.1% -21.4% -36.4% -39.3%
Women
Uneducated
-21.3%
3.9% -22.9% -19.6% 14.2% 13.4% 14.1%
2.0% 28.2% 44.7% 67.2% 66.4% 36.3% 45.7%
Primary
-32.1%
-8.8% 14.0%
3.0%
Hs
-10.6%
-4.3% -29.0% -13.9% 88.0%
College
-49.8% -33.2% 30.0% 11.0%
4.7%
4.7%
-6.3%
6.0%
-0.1%
8.0% -10.6%
9.9%
-0.9% 10.6%
7.6% -20.9% 10.7% -19.8% -28.2%
8.0%
-9.6% -20.7% -30.5%
-2.1% 25.6% 30.4% 38.3% 16.4% 53.0% 47.6% 13.6%
-1.2%
-3.3%
35
TABLE 4.4: ESTIMATION RESULTS EQUATION (9) - URBAN
78-88
Biased
coeff
elasticity
significance
R2
c.i coeff
c.i elas
el=1-Beta1
coeff
elasticity
significance
R2
c.i coeff
c.i elas
el=1-Beta2
coeff
elasticity
significance
R2
c.i coeff
c.i elas
0.0013
0.9987
9.5%
10.0%
88-91
0.1686
0.8314
88.8%
15.8%
91-94
-0.1818
1.1818
73.4%
11.4%
94-96
96-98
0.0693
0.9307
26.4%
10.2%
-0.1533
1.1533
94.6%
15.5%
[-0.02,0.023]
[0.041,0.379]
[-0.507,0.143]
[-0.388,0.526]
[-0.366,-0.003]
[0.978,1.019]
[0.627,0.996]
[0.865,1.5]
[0.485,1.377]
[1,1.361]
0.0217
0.9783
100.0%
55.6%
0.2312
0.7688
98.3%
21.5%
-0.4303
1.4303
96.0%
18.7%
0.5584
0.4416
100.0%
21.8%
-0.2035
1.2035
98.1%
21.1%
[0.013,0.03]
[0.049,0.414]
[-0.772,-0.089]
[0.202,0.914]
[-0.391,-0.016]
[0.967,0.987]
[0.591,0.947]
[1.098,1.763]
[0.095,0.789]
[1.02,1.386]
0.0132
0.9868
63.9%
11.8%
0.2157
0.7843
97.2%
19.8%
-0.2953
1.2953
82.5%
13.9%
0.1554
0.8446
55.2%
11.2%
-0.1932
1.1932
97.0%
16.4%
[-0.017,0.043]
[0.025,0.406]
[-0.618,0.027]
[-0.306,0.617]
[-0.387,-0.012]
[0.956,1.016]
[0.599,0.97]
[0.991,1.61]
[0.395,1.294]
[1.003,1.383]
TABLE 4.4A: ESTIMATION RESULTS - URBAN (SECOND DIFFERENCES)
78-88
Biased
coeff
elasticity
significance
R2
c.i coeff
c.i elas
el=1-Beta1
coeff
elasticity
significance
R2
c.i coeff
c.i elas
el=1-Beta2
coeff
elasticity
significance
R2
c.i coeff
c.i elas
0.0051
0.9949
21.0%
10.1%
88-91
0.0298
0.9702
97.8%
14.0%
91-94
0.1155
0.8845
35.1%
10.5%
94-96
-0.1083
1.1083
39.7%
10.4%
96-98
-0.0471
1.0471
31.4%
20.2%
[-0.033,0.044]
[0.005,0.055]
[-0.392,0.623]
[-0.536,0.32]
[-0.412,0.318]
[0.957,1.032]
[0.946,0.995]
[0.39,1.379]
[0.691,1.625]
[.691,1.403]
0.0628
0.9372
99.2%
31.5%
0.0469
0.9531
98.9%
20.5%
0.5056
0.4944
92.6%
22.7%
-0.6043
1.6043
99.5%
25.7%
-0.3648
1.3648
99.9%
39.4%
[0.017,0.109]
[0.011,0.083]
[0.051,1.062]
[-1.016,-0.192]
[-0.578,-0.151]
[0.892,0.982]
[0.918,0.988]
[-0.048,0.974]
[1.203,2.006]
[1.157,1.573]
0.0137
0.9863
57.3%
10.6%
0.0412
0.9588
99.1%
17.8%
0.2265
0.7735
59.7%
12.1%
-0.2235
1.2235
71.6%
11.7%
-0.1131
1.1131
57.6%
21.3%
[-0.021,0.048]
[0.011,0.072]
[-0.314,0.767]
[-0.639,0.192]
[-0.476,0.25]
[0.953,1.02]
[0.929,0.988]
[0.247,1.3]
[0.819,1.628]
[.76,1.466]
36
TABLE 4.5: CHANGES IN NET SUPPLIES AND E ARNINGS FOR AGGREGATED LEVELS OF SKILL - URBAN
78-88
Groups
Men
Women
CS
CD
-0.07
-0.03
0.12
0.04
CNS
88-91
CW
CS
CD
-0.04
-0.03
-0.03
0.00
0.09
0.04
0.05
0.00
CNS
91-94
CD
CNS
94-96
CW
CS
CW
CS
CD
CNS
CW
-0.03
-0.07
0.00
0.02
-0.02
0.24
-0.01
0.00
-0.01
-0.10
0.05
-0.14
0.00
-0.03
0.03
0.28
0.01
0.00
0.02
-0.09
Un
-1.34
-0.22
-1.09
0.20
-0.14
-0.04
-0.10
-0.10
0.02
-0.01
0.03
0.25
-0.30
-0.02
-0.28
-0.15
Prim
-0.41
-0.05
-0.36
0.15
-0.12
-0.03
-0.10
-0.08
-0.05
0.00
-0.05
0.15
-0.04
-0.01
-0.04
-0.06
Hsd
0.13
-0.03
0.17
0.03
-0.02
0.00
-0.03
-0.10
-0.04
0.00
-0.03
0.17
-0.01
-0.01
0.00
-0.11
Hs
0.75
0.02
0.73
-0.06
0.10
0.02
0.08
-0.10
0.10
0.00
0.10
0.28
0.06
0.00
0.06
-0.19
Cd
0.73
0.09
0.65
0.05
0.15
0.02
0.12
-0.16
-0.09
0.00
-0.08
0.32
0.07
0.01
0.06
-0.14
C+
1.06
0.16
0.91
-0.12
0.14
0.02
0.12
-0.06
0.06
0.02
0.04
0.33
-0.01
0.03
-0.04
-0.04
0-5
-0.05
0.01
-0.06
0.00
-0.07
-0.02
-0.06
-0.04
-0.03
0.01
-0.04
0.20
-0.03
-0.03
0.01
0.00
5-10
0.14
0.00
0.14
-0.07
-0.05
0.00
-0.05
-0.07
-0.05
0.01
-0.06
0.24
-0.04
-0.02
-0.02
-0.05
10-25
25+
0.14
0.00
0.14
-0.01
0.06
0.00
0.06
-0.09
0.03
0.00
0.03
0.27
-0.01
0.01
-0.02
-0.16
-0.22
-0.01
-0.21
0.03
-0.01
0.01
-0.02
-0.12
0.01
-0.01
0.02
0.26
0.06
0.02
0.03
-0.07
-1.47
-0.16
-1.30
0.11
-0.12
0.00
-0.11
0.01
0.12
0.00
0.12
0.18
-0.36
-0.01
-0.34
-0.20
-0.50
-0.09
-0.41
0.12
-0.13
0.00
-0.12
-0.06
-0.03
0.02
-0.05
0.14
-0.06
0.00
-0.06
-0.06
0.13
-0.06
0.19
0.02
-0.07
0.00
-0.07
-0.09
-0.02
0.01
-0.04
0.18
-0.02
-0.01
-0.01
-0.12
0.74
0.00
0.74
-0.07
0.10
0.01
0.09
-0.08
0.08
0.01
0.07
0.28
0.07
0.00
0.07
-0.18
0.69
0.07
0.62
0.07
0.04
0.01
0.03
-0.17
-0.02
0.01
-0.03
0.18
0.05
0.01
0.04
0.01
0.82
0.18
0.66
-0.17
0.10
0.00
0.10
-0.02
-0.02
0.04
-0.06
0.34
-0.01
0.03
-0.04
-0.09
Men
Un
Prim
Hsd
Hs
Cd
C+
Women
Un
Prim
Hsd
Hs
Cd
C+
-1.19
-0.31
-0.82
0.34
-0.16
-0.08
-0.07
-0.28
-0.12
-0.03
-0.08
0.38
-0.22
-0.03
-0.19
-0.09
-0.25
0.02
-0.26
0.25
-0.12
-0.07
-0.05
-0.11
-0.07
-0.04
-0.03
0.19
-0.01
-0.02
0.01
-0.05
0.15
0.02
0.13
0.06
0.06
0.01
0.05
-0.13
-0.06
-0.03
-0.02
0.14
0.00
-0.01
0.01
-0.07
0.77
0.05
0.72
-0.05
0.11
0.04
0.07
-0.13
0.13
-0.02
0.15
0.28
0.05
0.00
0.05
-0.20
0.78
0.11
0.68
0.00
0.27
0.04
0.23
-0.15
-0.15
-0.02
-0.14
0.53
0.10
0.01
0.09
-0.35
1.68
0.14
1.54
-0.02
0.21
0.04
0.17
-0.16
0.17
-0.01
0.18
0.30
-0.02
0.04
-0.05
0.09
37
TABLE 4.5: CONTINUATION
Groups
Men
96-98
CS
CD
CNS
91-98
CW
CS
CD
CNS
78-98
CW
CS
CD
CNS
CW
-0.03
-0.02
-0.01
-0.02
-0.04
0.00
-0.04
0.12
-0.14
-0.02
-0.11
0.02
Women
0.04
0.02
0.02
0.05
0.06
-0.01
0.06
0.24
0.22
0.03
0.20
0.14
Un
0.05
0.01
0.04
0.07
-0.23
-0.02
-0.21
0.16
-1.71
-0.28
-1.38
0.26
Prim
-0.11
-0.02
-0.09
-0.02
-0.20
-0.02
-0.18
0.08
-0.74
-0.10
-0.63
0.15
Hsd
-0.15
-0.04
-0.12
-0.04
-0.19
-0.04
-0.15
0.02
-0.08
-0.07
-0.01
-0.05
Hs
0.05
0.00
0.04
0.02
0.21
0.00
0.21
0.10
1.06
0.05
1.02
-0.06
Cd
0.06
0.05
0.01
-0.03
0.05
0.06
-0.01
0.16
0.93
0.18
0.76
0.04
C+
0.32
0.07
0.25
0.02
0.37
0.12
0.25
0.32
1.57
0.31
1.30
0.13
0-5
0.00
0.00
0.00
-0.01
-0.05
-0.02
-0.03
0.18
-0.18
-0.03
-0.15
0.14
5-10
-0.11
-0.01
-0.10
0.02
-0.21
-0.02
-0.19
0.20
-0.12
-0.02
-0.10
0.06
10-25
0.04
0.00
0.04
0.04
0.06
0.00
0.06
0.15
0.26
0.00
0.26
0.04
25+
0.00
0.00
0.00
-0.07
0.07
0.02
0.05
0.12
-0.16
0.02
-0.18
0.03
Men
Un
Prim
Hsd
Hs
Cd
C+
0.05
0.01
0.04
0.12
-0.18
0.00
-0.18
0.10
-1.77
-0.16
-1.59
0.22
-0.15
-0.03
-0.12
-0.06
-0.25
-0.01
-0.23
0.02
-0.88
-0.11
-0.76
0.07
-0.19
-0.05
-0.14
-0.08
-0.23
-0.04
-0.19
-0.02
-0.17
-0.10
-0.07
-0.10
0.03
-0.01
0.04
0.02
0.17
-0.01
0.18
0.12
1.02
0.01
1.01
-0.04
0.07
0.03
0.04
-0.10
0.09
0.05
0.04
0.09
0.82
0.13
0.70
-0.01
0.34
0.06
0.28
0.01
0.31
0.13
0.19
0.27
1.23
0.31
0.96
0.08
Women
Un
Prim
Hsd
Hs
Cd
C+
0.05
0.02
0.04
-0.01
-0.28
-0.04
-0.24
0.28
-1.63
-0.44
-1.06
0.35
-0.05
0.02
-0.06
0.07
-0.12
-0.04
-0.09
0.21
-0.49
-0.08
-0.40
0.36
-0.09
-0.01
-0.07
0.04
-0.14
-0.05
-0.09
0.11
0.07
-0.03
0.09
0.05
0.07
0.02
0.05
0.00
0.25
0.00
0.25
0.08
1.12
0.09
1.04
-0.10
0.06
0.08
-0.02
0.09
0.00
0.06
-0.05
0.26
1.06
0.23
0.86
0.11
0.31
0.09
0.22
0.05
0.46
0.12
0.35
0.44
2.34
0.29
2.08
0.26
38
TABLE 4.6: RELATIVE DEMAND SHIFTS FOR AGGREGATED GROUPS - URBAN
Groups
78-88
88-91
91-94
94-96
96-98
Bet Within Tot
Bet Within Tot
Bet Within Tot
Bet Within Tot
Bet Within Tot
Men
-2.4% -0.3% -2.7% -0.4% 0.6% 0.2% 2.0% -0.3% 1.7% -0.9% 1.1% 0.2% -1.0% -0.8% -1.7%
Women
3.7% -0.1% 3.6% 0.7% -1.0% -0.3% -3.1% 0.4% -2.7% 1.4% -1.6% -0.2% 1.6%
Un
-9.4% -12.9% -22.3% -3.1% -0.6% -3.8% -1.0% -0.5% -1.5% -1.7% -0.2% -1.9% 1.5% -0.1% 1.4%
Prim
-2.7% -2.5% -5.2% -2.2% -0.5% -2.7% 0.6% -0.7% -0.1% -1.1% 0.6% -0.5% -1.3% -0.3% -1.6%
Hsd
1.1% -4.3% -3.2% -0.1% 0.6% 0.5% 0.0% -0.3% -0.3% -0.8% 0.3% -0.5% -1.9% -1.6% -3.5%
Hs
2.4% 0.0% 2.4% 1.7% 0.7% 2.4% -0.8% 0.5% -0.4% 0.2% -0.4% -0.3% 1.0% -0.6% 0.4%
Cd
2.4% 6.6% 8.9% 2.5% -0.2% 2.3% -0.7% 0.5% -0.2% 2.0% -0.9% 1.2% 3.7%
1.5% 5.3%
C+
1.9% 14.5% 16.4% 2.9% -1.2% 1.7% 0.5% 1.3% 1.8% 4.1% -0.8% 3.3% 3.5%
3.9% 7.3%
0-5
2.5% -1.5% 1.0% -0.4% -1.4% -1.8% -0.3% 1.0% 0.7% -0.2% -3.1% -3.2% 1.5% -1.2% 0.3%
5-10
1.5% -1.5% 0.0% 0.3% -0.5% -0.2% 0.1% 0.4% 0.6% -0.2% -1.6% -1.8% 0.3% -1.0% -0.7%
0.8% 2.4%
10-25
0.3% -0.6% -0.3% 0.3% 0.1% 0.4% 0.0% -0.2% -0.1% 0.3% 0.3% 0.6% -0.2% 0.0% -0.3%
25+
-3.0% 1.9% -1.0% -0.2% 0.9% 0.7% 0.0% -0.7% -0.7% -0.1% 2.5% 2.4% -0.7% 1.0% 0.3%
Men
Un
Prim
Hsd
Hs
Cd
C+
-20.7% 5.0% -15.7% -1.4% 0.9% -0.5% 1.9% -1.9% 0.0% -3.2% 1.9% -1.3% -0.2%
1.5% 1.2%
-6.5% -2.8% -9.2% -1.6% 1.3% -0.4% 3.2% -1.3% 1.9% -2.2% 2.3% 0.1% -2.2% -1.3% -3.5%
-0.4% -5.5% -5.9% -0.8% 0.9% 0.1% 1.9% -0.5% 1.4% -1.5% 1.0% -0.5% -2.2% -2.5% -4.7%
0.8% -0.6% 0.2% 0.7% 0.5% 1.2% 0.4% 0.3% 0.7% -0.7% 0.4% -0.3% 0.1% -1.1% -1.0%
2.0% 5.2% 7.2% 1.1% -0.5% 0.6% 0.7% 0.6% 1.3% 1.5% -0.6% 0.9% 2.3%
0.8% 3.2%
2.2% 15.8% 17.9% 1.9% -1.4% 0.5% 2.0% 1.9% 3.9% 4.0% -0.9% 3.1% 1.6%
4.4% 6.0%
Women
Un
Prim
Hsd
Hs
Cd
C+
5.4% -36.4% -31.0% -5.4% -2.7% -8.1% -4.7% 1.4% -3.3% 0.2% -2.8% -2.6% 3.8% -2.3% 1.6%
3.9% -2.0% 1.8% -3.1% -3.5% -6.7% -4.0% 0.4% -3.6% 0.8% -2.5% -1.6% 0.3%
1.3% 1.7%
3.8% -2.2% 1.6% 1.1% 0.1% 1.2% -3.3% 0.0% -3.4% 0.5% -1.1% -0.6% -1.3% -0.1% -1.5%
4.5% 0.7% 5.1% 3.1% 0.9% 4.0% -2.4% 0.7% -1.7% 1.3% -1.6% -0.3% 2.2%
0.1% 2.2%
2.8% 8.1% 10.9% 4.1% 0.2% 4.3% -2.2% 0.4% -1.8% 2.7% -1.2% 1.5% 5.3%
2.3% 7.6%
1.6% 12.6% 14.2% 4.5% -0.9% 3.6% -1.7% 0.5% -1.2% 4.3% -0.6% 3.7% 6.2%
3.0% 9.2%
39
TABLE 4.6: CONTINUATION
91-98
Groups
Bet
Within
78-98
Tot
Bet
0.3%
Tot
0.1%
0.0%
0.1%
Women
-0.1%
-0.5%
-0.6%
4.3%
-1.5%
2.8%
Un
-1.2%
-0.8%
-2.0%
-13.8%
-14.3%
-28.1%
Prim
-1.8%
-0.4%
-2.2%
-6.7%
-3.4%
-10.1%
Hsd
-2.7%
-1.7%
-4.4%
-1.8%
-5.3%
-7.1%
Hs
0.3%
-0.6%
-0.2%
4.5%
0.1%
4.6%
Cd
5.1%
0.5%
5.7%
10.0%
7.6%
17.6%
Men
-2.7%
Within
-2.4%
C+
8.1%
4.4%
12.5%
12.9%
17.7%
30.6%
0-5
1.0%
-3.2%
-2.2%
3.1%
-6.3%
-3.1%
5-10
0.2%
-2.3%
-2.0%
2.0%
-4.1%
-2.2%
10-25
0.1%
0.0%
0.1%
0.7%
-0.4%
0.2%
-0.8%
2.7%
1.9%
-4.0%
5.6%
1.6%
-1.6%
1.5%
-0.1%
-23.7%
7.4%
-16.3%
-1.2%
-0.2%
-1.4%
-9.3%
-1.7%
-11.0%
25+
Men
Un
Prim
Hsd
Hs
Cd
C+
-1.9%
-1.9%
-3.8%
-3.1%
-6.5%
-9.6%
-0.2%
-0.4%
-0.6%
1.3%
-0.4%
0.8%
4.5%
0.9%
5.3%
7.6%
5.6%
13.2%
7.6%
5.4%
13.0%
11.7%
19.7%
31.4%
-0.7%
-3.7%
-4.4%
-0.8%
-42.8%
-43.6%
-2.8%
-0.8%
-3.6%
-2.1%
-6.4%
-8.5%
-4.2%
-1.2%
-5.4%
0.7%
-3.3%
-2.7%
1.1%
-0.8%
0.2%
8.6%
0.7%
9.3%
Women
Un
Prim
Hsd
Hs
Cd
C+
5.8%
0.2%
6.0%
12.8%
9.8%
22.6%
8.7%
2.9%
11.7%
14.8%
14.7%
29.5%
40
TABLE 4.7: IMPLICIT SUPPLY IN INTERNATIONAL TRADE FLOWS - URBAN
Groups
1978
1988
1991
1994
1996
1998
All
7.56%
9.14%
4.30%
15.01%
11.73%
13.64%
Men
6.74%
0.82%
8.20%
0.94%
6.18%
-1.89%
13.26%
1.75%
10.34%
1.39%
11.19%
2.46%
-0.15%
1.31%
2.79%
2.24%
0.58%
0.78%
0.01%
1.97%
2.90%
2.44%
0.73%
1.08%
-0.04%
0.35%
1.20%
1.43%
0.51%
0.84%
0.08%
3.59%
4.86%
3.84%
1.07%
1.57%
0.07%
2.84%
3.77%
2.98%
0.84%
1.23%
0.09%
3.43%
4.46%
3.40%
0.92%
1.34%
-0.12%
1.40%
2.60%
1.83%
0.43%
0.61%
0.03%
2.05%
2.76%
2.00%
0.54%
0.83%
0.00%
1.38%
2.11%
1.60%
0.43%
0.67%
0.09%
3.54%
4.53%
3.13%
0.79%
1.19%
0.07%
2.80%
3.50%
2.42%
0.62%
0.93%
0.09%
3.03%
3.80%
2.61%
0.66%
1.00%
-0.02%
-0.09%
0.19%
0.41%
0.15%
0.18%
-0.01%
-0.07%
0.14%
0.44%
0.19%
0.26%
-0.05%
-1.03%
-0.90%
-0.16%
0.08%
0.17%
-0.01%
0.05%
0.33%
0.71%
0.28%
0.38%
-0.01%
0.05%
0.27%
0.56%
0.23%
0.29%
0.01%
0.40%
0.66%
0.79%
0.27%
0.34%
Women
Uneducated
Primary
Hs dropouts
Hs
Some college
College +
Men
Uneducated
Primary
Hs dropouts
Hs
Some college
College +
Women
Uneducated
Primary
Hs dropouts
Hs
Some college
College +
41
TABLE 4.7A: INTERNATIONAL TRADE DEMAND SHIFTS - URBAN
Groups
1978
1988
1991
1994
1996
1998
Total
0.00
0.00
0.00
0.00
0.00
0.00
Men
-0.15
-0.18
-0.23
-0.28
-0.22
-0.23
Women
0.23
0.27
0.36
0.43
0.34
0.35
Uneducated
0.38
0.35
0.22
0.50
0.40
0.51
Primary
0.12
0.12
0.13
0.16
0.12
0.14
Hs dropouts
-0.11
-0.09
-0.02
-0.17
-0.13
-0.19
Hs
-0.10
-0.10
-0.10
-0.13
-0.10
-0.12
Some college
-0.02
-0.05
-0.12
-0.01
-0.01
0.03
0.01
-0.03
-0.13
0.03
0.02
0.07
0-5 exp
-0.05
-0.05
-0.09
-0.06
-0.05
-0.04
5-10
-0.08
-0.08
-0.07
-0.10
-0.08
-0.10
10-25
-0.03
-0.03
-0.01
-0.04
-0.03
-0.04
25 +
0.12
0.12
0.11
0.16
0.12
0.15
College +
Men
Uneducated
0.38
0.30
0.15
0.41
0.31
0.42
Primary
-0.01
-0.03
-0.10
-0.09
-0.08
-0.05
Hs dropouts
-0.31
-0.32
-0.34
-0.53
-0.41
-0.47
Hs
-0.29
-0.31
-0.36
-0.46
-0.36
-0.40
Some college
-0.15
-0.20
-0.28
-0.24
-0.19
-0.18
College +
-0.08
-0.14
-0.24
-0.13
-0.10
-0.08
Uneducated
0.38
0.41
0.31
0.63
0.50
0.63
Primary
0.33
0.39
0.52
0.59
0.47
0.49
0.31
Women
Hs dropouts
0.23
0.30
0.55
0.46
0.36
Hs
0.14
0.18
0.24
0.29
0.23
0.24
Some college
0.14
0.13
0.07
0.26
0.20
0.27
College +
0.15
0.14
0.01
0.26
0.21
0.29
42
TABLE 4.8: CHANGES IN RELATIVE SUPPLIES DUE TO TRADE FLOWS
Year
Trade adj.
supply-all
True
coll/High S
Trade adj
coll/High S
77
1.096
0.145
0.151
78
1.076
0.118
0.123
79
1.105
0.147
0.152
80
1.142
0.168
0.170
81
1.149
0.176
0.177
82
1.147
0.189
0.188
83
1.154
0.184
0.181
84
1.121
0.195
0.191
85
1.084
0.202
0.200
86
1.074
0.204
0.201
87
1.073
0.204
0.202
88
1.096
0.213
0.211
89
1.075
0.249
0.245
90
1.095
0.234
0.230
91
1.065
0.237
0.236
92
1.086
0.243
0.238
93
1.132
0.235
0.228
94
1.151
0.245
0.235
95
1.167
0.223
0.216
96
1.118
0.236
0.229
97
98
1.139
1.137
0.294
0.342
0.277
0.318
43
TABLE 4.9: TRADE, RELATIVE SUPPLIES AND RELATIVE EARNINGS, COLLEGE VS HIGH SCHOOL
P r e d v a lu e s
P r e d v a lu e s
% change
% change
a d ju s t e d s u p
u n a d ju s t e d s u p
a d ju s t e d
u n a d ju s t e d
Year
77
0.9533
0.9575
na
na
78
1.0894
1.0854
14.3%
13.4%
79
1.0040
1.0089
-7.8%
-7.0%
80
0.9747
0.9757
-2.9%
-3.3%
81
0.9828
0.9810
0.8%
0.5%
82
0.9796
0.9752
-0.3%
-0.6%
83
1.0261
1.0177
4.7%
4.4%
84
1.0275
1.0195
0.1%
0.2%
85
1.0307
1.0307
0.3%
1.1%
86
1.0556
1.0545
2.4%
2.3%
87
1.0814
1.0826
2.4%
2.7%
88
1.0883
1.0923
0.6%
0.9%
89
1.0386
1.0452
-4.6%
-4.3%
90
1.1003
1.1049
5.9%
5.7%
91
1.1151
1.1272
1.3%
2.0%
92
1.1377
1.1435
2.0%
1.4%
93
1.2388
1.2312
8.9%
7.7%
94
1.2017
1.1973
-3.0%
-2.8%
95
1.2747
1.2698
6.1%
6.1%
96
1.2713
1.2722
-0.3%
0.2%
97
1.1998
1.1555
1.1954
1.1522
-5.6%
-3.7%
-6.0%
-3.6%
98
TABLE 4.10: ESTIMATION RESULTS - RURAL
7 8 -8 8
8 8 -9 1
9 1 -9 4
9 4 -9 6
9 6 -9 8
B ia s e d
c o e ff
-0 .0 6 6 1
-1 .3 8 2 5
0 .0 8 3 0
0 .4 9 0 6
-0 .2 1 0 5
1 .0 6 6 1
2 .3 8 2 5
0 .9 1 7 0
0 .5 0 9 4
1 .2 1 0 5
s ig n ific a n c e
6 6 .3 %
9 6 .0 %
2 3 .2 %
9 3 .1 %
6 6 .0 %
R 2
1 1 .1 %
2 3 .5 %
1 0 .1 %
1 6 .8 %
1 2 .6 %
e la s tic ity
e l= 1 -B e ta 1
c o e ff
-0 .3 0 1 0
-1 .5 4 3 0
0 .4 6 9 2
0 .8 8 3 3
-0 .4 4 1 2
1 .3 0 1 0
2 .5 4 3 0
0 .5 3 0 8
0 .1 1 6 7
1 .4 4 1 2
s ig n ific a n c e
9 9 .5 %
9 8 .7 %
9 9 .9 %
9 9 .4 %
9 9 .0 %
R 2
5 8 .0 %
3 1 .9 %
3 2 .4 %
3 5 .1 %
1 3 .4 %
e la s tic ity
TABLE 4.10A: INNER PRODUCTS DW*DNS
78-88
Biased
-0.025927
88-91
91-94
0.074176 -0.018937
94-96
96-98
0.042164
0.006232
44
TABLE 4.11: CHANGES IN NET SUPPLIES AND E ARNINGS FOR AGGREGATED LEVELS OF SKILL - RURAL
78-88
Groups
Men
Women
Un
Prim
88-91
CS
CD
-0.07
0.01 -0.08
0.36 -0.03
0.21
0.25
0.30
-0.62
0.03
CNS
CW
0.11
CS
0.11
CD
CNS
91-94
CS
CD
0.00 -0.03
0.08
0.00
0.01 -0.01 -0.24
0.02
0.03
0.00
0.04 -0.04
0.09
CNS
0.01 -0.63
0.71 -0.14
0.00 -0.14 -0.48 -0.10 -0.01 -0.09
0.01
0.31
0.00
0.02
0.00
94-96
CW
0.00
0.13 -0.03
CW
CS
CD
CNS
0.05 -0.01
0.15 -0.17 -0.13 -0.03 -0.36
0.01
0.23
0.01
0.22 -0.20
0.00 -0.03 -0.18
0.08
0.00
0.08 -0.28
Hs
1.13
0.12
1.02
0.19
0.11
0.02
0.09
0.32
0.11
0.03
Coll
1.72
1.50
0.80
0.03
0.30
0.14
0.17
0.32
0.30
0.55 -0.14 -0.14 -0.51 -1.13
0-10
0.04
0.07 -0.02
0.23
0.00
0.01 -0.01 -0.09 -0.04
0.02 -0.06
0.32
0.01
0.01
10-25
25+
0.06
-0.07
0.06
0.00
0.04 -0.11
0.41 -0.01
CW
0.06 -0.28
0.08 -0.28 -0.38 -0.09 -0.29 -0.39
0.63 -0.33
0.18 -0.10 -0.04 -0.06 -0.26
0.00
0.23
0.00
0.02 -0.02 -0.25
0.02 -0.04
0.06 -0.24
0.00 -0.01
0.03
0.03
0.01
0.04 -0.03
0.07 -0.35
0.02 -0.27
Men
Un
-0.65 -0.04 -0.61
0.76 -0.19
0.00 -0.19 -0.50 -0.04 -0.01 -0.03
0.01
0.27
0.03
0.24 -0.21
Prim
-0.04 -0.01 -0.03
0.33 -0.02
0.00 -0.02
0.12 -0.02
0.00 -0.02 -0.20
0.10
0.01
0.09 -0.28
Hs
1.17
0.05
1.12
0.16
0.08
0.01
0.07
0.46
0.07
0.02
0.05 -0.54 -0.32 -0.05 -0.27 -0.28
Coll
1.42
1.26
0.60 -0.21
0.23
0.12
0.12
0.30
0.29
0.50 -0.11 -0.05 -0.68 -1.04
0.36 -0.54
Women
Un
0.22 -0.69
0.21
0.03
0.01
0.02 -0.16 -0.30 -0.01 -0.30
Prim
-0.49
0.39
0.08
0.31
0.12
0.08
0.01
0.08
Hs
1.05
0.28
0.80
0.29
0.17
0.04
0.13 -0.24
0.18
0.06
Coll
2.52
1.81
1.49
0.79
0.39
0.16
0.24
0.32
0.62 -0.17 -0.33 -0.32 -1.25
0.20 -0.05
0.37
0.04 -0.08
0.12 -0.10
0.01 -0.06 -0.02 -0.02 -0.04
0.02 -0.28
0.12
0.05
0.55 -0.51 -0.19 -0.30 -0.62
0.93
0.03
TABLE 4.11: CONTINUATION
96-98
Groups
Men
Women
Un
Prim
Hs
Coll
0-10
10-25
25+
91-98
CD
78-98
CS
CD
CNS
CW
CS
CNS
CW
CS
CD
CNS
-0.001
0.00
-0.001
-0.04
0.000
0.04
0.127
0.19
0.046
-0.16
-0.006
-0.13
0.052
-0.03
-0.390
-0.02
-0.055
0.25
0.006
0.11
-0.061 0.054
0.15 0.26
CW
0.25
-0.06
-0.12
0.12
0.01
0.00
-0.02
-0.36
0.24
-0.06
-0.09
0.56
0.25
0.11
0.08
0.21
0.38
-0.01
-0.39
-0.09
0.01
-0.01
-0.09
-0.94
0.37
0.00
-0.30
2.72
0.07
-0.35
-0.59
-0.27
-0.38
0.02
0.85
1.93
0.02
0.00
0.06
0.70
-0.41
0.02
0.79
1.40
0.30
0.08
-0.08
0.09
-0.10
-0.01
0.06
-0.01
-0.01
-0.01
-0.09
0.00
0.07
0.11
0.19
0.11
-0.24
0.01
0.12
-0.03
-0.03
-0.03
-0.21
0.04
0.15
0.03
-0.30
-0.52
-0.20
0.08
0.05
0.05
0.03
0.01
-0.24
0.05
0.03
0.17
0.25
-0.09
0.25
-0.07
-0.11
0.30
0.00
0.00
0.00
-0.18
0.25
-0.07
-0.11
0.50
0.24
0.10
0.12
0.01
0.47
0.01
-0.36
-0.08
0.02
0.01
-0.03
-0.73
0.45
0.01
-0.33
1.21
0.03
-0.39
-0.70
-0.58
-0.37
-0.05
0.88
1.57
-0.02
0.00
0.03
0.66
-0.35
-0.04
0.86
1.06
0.29
0.06
-0.09
-0.49
0.28
-0.02
-0.12
-0.09
0.06
-0.02
-0.07
-0.58
0.22
0.00
-0.05
0.79
0.38
0.21
-0.04
0.42
0.02
-0.10
-0.46
-0.09
-0.02
-0.05
-0.20
-1.22
0.04
-0.04
-0.23
1.13
0.33
-0.09
-0.12
0.12
-0.45
0.37
0.77
2.81
0.21
0.03
0.12
0.75
-0.64
0.35
0.65
2.25
0.38
0.23
-0.07
1.29
Men
Un
Prim
Hs
Coll
Women
Un
Prim
Hs
Coll
45
TABLE 4.12: IMPLICIT SUPPLY IN INTERNATIONAL TRADE FLOWS - RURAL
Groups
All
Men
Women
Uneducated
Primary
Hs
College
Men
Uneducated
Primary
Hs
College
Women
Uneducated
Primary
Hs
College
1978
1988
1991
1994
-31.02%
-27.94%
-3.09%
-7.67%
-20.99%
-2.36%
-0.01%
-19.52%
-16.76%
-2.76%
-4.87%
-13.07%
-1.56%
-0.01%
-19.22%
-15.76%
-3.46%
-4.60%
-12.83%
-1.78%
-0.01%
-13.94%
-11.84%
-2.10%
-3.83%
-9.94%
-0.29%
0.12%
1996
-7.82%
-6.45%
-1.37%
-2.09%
-5.40%
-0.38%
0.04%
1998
-10.49%
-9.16%
-1.33%
-2.72%
-7.14%
-0.65%
0.02%
-6.89%
-18.97%
-2.08%
0.00%
-4.03%
-11.39%
-1.34%
0.00%
-3.77%
-10.68%
-1.31%
0.01%
-3.22%
-8.60%
-0.15%
0.13%
-1.65%
-4.58%
-0.27%
0.05%
-2.28%
-6.35%
-0.56%
0.03%
-0.78%
-2.02%
-0.28%
-0.01%
-0.84%
-1.69%
-0.22%
-0.01%
-0.84%
-2.14%
-0.46%
-0.02%
-0.61%
-1.34%
-0.14%
-0.01%
-0.44%
-0.81%
-0.11%
-0.01%
-0.44%
-0.80%
-0.09%
-0.01%
TABLE 4.12A: REALTIVE DEMAND SHIFTS DUE TO INTERNATIONA TRADE FLOWS - RURAL
Groups
Total
1978
0.00%
1988
0.00%
1991
0.00%
1994
0.00%
1996
0.00%
1998
0.00%
Men
Women
0.04%
-0.16%
0.02%
-0.06%
0.01%
-0.03%
0.01%
-0.04%
0.00%
-0.01%
0.01%
-0.04%
Uneducated
Primary
Hs
College
0.06%
0.03%
-0.16%
-0.30%
0.04%
0.02%
-0.10%
-0.19%
0.03%
0.02%
-0.08%
-0.19%
0.05%
0.02%
-0.12%
-0.21%
0.02%
0.01%
-0.05%
-0.10%
0.03%
0.01%
-0.06%
-0.12%
0-10 exp
10-25
25+
-0.06%
-0.05%
0.00%
-0.04%
-0.03%
0.00%
-0.03%
-0.02%
0.01%
-0.03%
-0.03%
0.01%
-0.02%
-0.01%
0.00%
-0.02%
-0.02%
0.00%
0.10%
0.07%
-0.12%
-0.31%
0.04%
0.03%
-0.07%
-0.19%
0.03%
0.02%
-0.07%
-0.20%
0.05%
0.03%
-0.13%
-0.27%
0.02%
0.01%
-0.05%
-0.13%
0.03%
0.02%
-0.05%
-0.13%
-0.10%
-0.13%
-0.25%
-0.29%
0.03%
-0.05%
-0.15%
-0.18%
0.03%
0.00%
-0.10%
-0.17%
0.02%
-0.02%
-0.11%
-0.13%
0.04%
-0.01%
-0.06%
-0.07%
0.01%
-0.03%
-0.09%
-0.10%
Men
Uneducated
Primary
Hs
College
Women
Uneducated
Primary
Hs
College
46
TABLE 4.13: EVOLUTION OF MAIN COVARIATES
1978
1988
1991
1994
1996
1998
Occupation
White
Blue
Domestic
Self-emp
Employer
Men
White
Blue
Domestic
Self-emp
Employer
Women
White
Blue
Domestic
Self-emp
Employer
Avg. years of
education
Avg. years of
experience
% women
6.2
8.2
8.7
8.9
9.0
9.6
18.2
17.8
17.8
18.1
18.5
18.7
34.3%
38.4%
40.2%
40.3%
40.9%
42.7%
1978
1988
33.1%
35.5%
6.1%
22.3%
3.0%
25.3%
39.6%
6.5%
23.5%
5.2%
1991
24.7%
40.9%
4.8%
25.4%
4.2%
1994
1996
1998
25.1%
41.3%
4.8%
24.0%
4.9%
23.6%
40.6%
4.1%
27.1%
4.6%
19.8%
42.3%
4.4%
29.4%
4.1%
33.48
33.84
0.34
25.56
6.79
33.64
33.60
0.18
26.37
6.21
12.07
48.73
16.47
20.21
2.51
12.37
52.78
11.57
20.43
2.85
Erarnings Premium by Sector Compared to Retail Commerce
Sector
Agriculture
coal
Oil, gas
Mining
Food, bev, to
Chemical
Metallic pr
Utilities
Water
Wholesale
Communic
Banking
Insurance
Real state
Governm
Services
Social serv
Entertainm
Multilateral
88
0.293
0.404
0.541
0.333
0.136
0.169
0.167
0.280
0.317
0.313
0.206
0.248
0.280
0.177
0.225
0.138
0.168
0.243
0.912
94
0.220
0.656
0.261
0.534
0.099
0.118
0.000
0.128
0.244
0.265
0.144
0.219
0.318
0.161
0.190
0.111
0.065
0.192
0.834
96
0.228
0.418
0.470
0.094
0.071
0.061
0.000
0.284
0.303
0.215
0.324
0.257
0.335
0.181
0.188
-0.036
0.162
0.205
0.318
47
TABLE 4.14: DECOMPOSITION OF CHANGES IN EARNINGS DISTRIBUTION
1988-94
Statistic
Gini
Actual
Emp
Trade
Educ
Occup
Gender
p.a.
s/d
Total
0.074
-0.036
-0.009
0.054
0.021
-0.011
0.008
0.038
0.066
%cont
100.0%
-47.9%
-11.5%
72.2%
28.4%
-15.0%
10.7%
51.8%
88.5%
90-10
0.076
-0.186
-0.426
0.469
0.196
-0.044
-0.065
0.153
0.098
%cont
100.0%
-242.9%
-557.1%
614.3% 257.1%
-57.1%
-85.7%
200.0%
128.6%
75-25
0.186
-0.218
0.011
0.251
-0.076
0.011
-0.011
0.164
0.131
%cont
100.0%
-117.6%
5.9%
135.3%
-41.2%
5.9%
-5.9%
88.2%
70.6%
Statistic
Actual
Emp
1994-96
Gini
-0.033
%cont
90-10
%cont
Trade
Educ
Occup
Gender
p.a.
s/d
Total
-0.019
-0.027
0.112
-0.012
-0.012
0.021
0.004
0.067
100.0%
58.6%
82.4%
-345.2%
37.4%
36.8%
-64.1%
-12.8%
-206.8%
0.099
-0.154
-0.242
0.474
0.022
-0.066
0.154
-0.011
0.176
100.0%
-155.6%
-244.4%
477.8%
22.2%
-66.7%
155.6%
-11.1%
177.8%
0.044
75-25
-0.033
-0.264
0.397
0.000
-0.033
0.044
0.000
0.187
%cont
100.0%
800.0%
-133.3% -1200.0%
0.0%
100.0%
-133.3%
0.0%
-566.7%
Statistic
Actual
Emp
Trade
Gender
p.a.
1996-98
Gini
%cont
Educ
Occup
s/d
Total
0.041
-0.011
-0.031
0.103
0.012
-0.007
0.009
0.000
0.076
100.0%
-26.1%
-76.2%
251.2%
30.4%
-17.1%
22.7%
-0.7%
184.3%
90-10
0.215
-0.108
-0.343
0.421
0.186
0.029
0.000
-0.010
0.176
%cont
100.0%
-50.0%
-159.1%
195.5%
86.4%
13.6%
0.0%
-4.5%
81.8%
75-25
0.078
-0.245
-0.020
0.440
-0.039
0.010
0.000
0.010
0.157
%cont
100.0%
-312.5%
-25.0%
562.5%
-50.0%
12.5%
0.0%
12.5%
200.0%
p.a.: Personal Attributes
48
TABLE 4.15: DECOMPOSITION OF CHANGE IN E ARNINGS DISTRIBUTION
Statistic
Sd wages
%cont
sd resid
%cont
coll/rest
%cont
Actual
0.011
100.0%
Emp
0.005
Trade
-0.026
46.2% -233.7%
1988-94
Educ
Occup
0.072
-0.034
Gender
p.a.
s/d
Total
-0.018
0.023
0.058
0.080
640.9% -301.1% -162.9%
207.0%
516.5%
713.0%
0.004
-0.011
-0.001
-0.048
0.065
-0.001
-0.018
-0.008
-0.013
100.0%
-0.8%
-27.9%
-12.1%
-20.7%
5.9%
-17.0%
-0.8%
-73.3%
0.024
0.021
-0.018
0.174
-0.043
-0.062
0.090
0.184
0.346
100.0%
88.1%
-74.3%
719.5% -177.9% -257.7%
370.5%
Actual
Emp
Trade
Educ
758.4% 1426.8%
1994-96
Statistic
Sd wages
%cont
sd resid
%cont
coll/rest
%cont
Statistic
Sd wages
%cont
sd resid
%cont
coll/rest
%cont
Occup
Gender
p.a.
s/d
Total
0.127
0.001
-0.083
-0.021
-0.066
0.022
-0.049
-0.003
-0.197
100.0%
1.1%
-65.8%
-16.2%
-51.8%
17.4%
-38.4%
-2.2%
-155.8%
0.068
-0.002
-0.034
-0.007
-0.024
0.001
-0.009
-0.001
-0.075
100.0%
-3.1%
-49.5%
-9.6%
-35.0%
1.8%
-13.1%
-2.1%
-110.7%
-0.003
0.055
-0.008
0.019
-0.034
0.098
54.2% -129.0%
229.5%
-660.6%
0.056
0.012
100.0%
-0.015
20.5% -370.8% -381.0%
-84.0%
Actual
Emp
Trade
1996-98
Educ
Occup
Gender
p.a.
s/d
Total
0.089
0.021
-0.023
0.197
-0.056
-0.076
0.101
0.183
0.345
100.0%
23.1%
-25.7%
220.7%
-63.4%
-85.9%
113.8%
205.2%
387.9%
0.011
0.018
-0.074
0.245
-0.022
-0.072
0.124
0.269
0.488
100.0% 162.7% -678.6% 2232.1% -197.7% -652.6% 1133.0% 2457.8% 4456.7%
0.028
0.009
-0.014
0.173
100.0%
31.0%
-50.6%
622.9%
-0.009
-0.068
0.104
-32.5% -242.4%
373.2%
0.228
0.424
819.8% 1521.4%
49
FIGURE 4.1: CHANGES IN URBAN EARNINGS FOR AGGREGATE EDUCATION GROUPS-ABSOLUTE
1.3
Earnings
1.1
0.9
0.7
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
Low
Medium
High
FIGURE 4.1A: CHANGE IN URBAN EARNINGS FOR AGGREGATE EDUCATION GROUPS – FIXED WEIGHT
Fixed Weight Earnings
1.3
1.2
1.1
1
0.9
0.8
0.7
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
Low
Med
High
50
FIGURE 4.1B: URBAN EDUCATION DIFFERENTIALS-FIXED WEIGHT
1.2
Differential
1.1
1
0.9
0.8
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
Coll-low
Coll-int
FIGURE 4.1C: GENDER DIFFERENTIAL BY EDUCATION LEVEL-FIXED WEIGHT
0.7
Differential
0.6
0.5
0.4
0.3
0.2
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
Low
Med
High
51
FIGURE 4.2: OBSERVABLE INEQUALITY-URBAN
1.4
1.3
Differential
1.2
1.1
1
0.9
0.8
0.7
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
90-10
90-50
50-10
75-25
FIGURE 4.2A: RESIDUAL INEQUALITY-URBAN
1.3
Differential
1.2
1.1
1
0.9
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
90-10
90-50
50-10
75-25
52
FIGURE 4.3: OBSERVABLE INEQUALITY-RURAL
1.7
Differential
1.5
1.3
1.1
0.9
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
90-10
90-50
50-10
75-25
FIGURE 4.3A: RESIDUAL INEQUALITY-RURAL
1.9
Differential
1.7
1.5
1.3
1.1
0.9
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Year
90-10
90-50
50-10
75-25
53
FIGURE 4.4: PERCENTAGE CHANGES IN EARNINGS AND NET DEMAND BY SKILL GROUP.
1978-88
1.6
Earnings
1.2
0.8
0.4
0
-0.4
-10.5
-8.5
-6.5
-4.5
-2.5
-0.5
Net Demand
1988-91
Earnings
0.4
0.2
0
-0.2
-0.4
-0.6
-1
-0.8
-0.6
-0.4
-0.2
0
0.2
0.4
Net Demand
1991-94
Earnings
1.4
1
0.6
0.2
-0.2
-3.8
-3.3
-2.8
-2.3
-1.8
-1.3
-0.8
-0.3
0.2
Net Demand
54
FIGURE 4.4: CONTINUATION
1994-96
Earnings
0.8
0.4
0
-0.4
-0.8
-0.5
-0.3
-0.1
0.1
0.3
0.5
0.7
Net Demand
1996-98
0.4
Earnings
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
-0.3
-0.2
-0.1
0
0.1
0.2
0.3
0.4
0.5
0.6
Net Demand
55
FIGURE 4.5: ACTUAL VS PREDICTED COLLEGE-HIGH SCHOOL EARNINGS DIFFERENTIALS-FIXED WEIGHT
1.4
Differential
1.3
1.2
1.1
1
0.9
0.8
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
95
96
97
98
Time
Actual
Predicted
FIGURE 4.5A: COLLEGE-HS DIFFERENTIAL VS TREND
1.4
Log diff
1.2
1
0.8
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
Time
Diff
Trend
56
FIGURE 4.5B: TIME PATH OF G(T) AND TRENDS
0.8
0.4
g(t)
0
-0.4
-0.8
-1.2
77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98
Time
el=1.476
el=1.65
Trend
Trend
el=1.3
el=3
Trend
Trend
FIGURE 4.5C: RELATIVE SUPPLY OF COLLEGE VS HIGH SCHOOL
Log relative suppply
1.1
0.9
0.7
0.5
0.3
0.1
-0.1
-0.3
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
Time
Det supply
Supply
57
FIGURE 4.6: MEASURED DISCRIMINATION (MALE VS FEMALE)
1.9
1.8
Male/female
1.7
1.6
1.5
1.4
1.3
1.2
1.1
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
Female's Earnings Support-5% percentiles
1978
1991
1998
58
FIGURE 4.7: DECOMPOSITION OF CHANGES IN DENSITY FUNCTIONS, 1988-94
Actual Densities
1.6
1988
1994
1.2
f(.)
5th
0.8
25th
75th
0.4
95th
0
2.5
5
7.5
10
12.5
Log of Earnings
Trade Counterfactual
1.6
1994
trade
f(.)
1.2
5th
0.8
25th
75th
0.4
95th
0
2.5
5
7.5
10
12.5
Log of Earnings
Personal Attributes Counterfactual
1.6
trade
f(.)
1.2
p.a.
5th
0.8
25th
75th
0.4
95th
0
2.5
5
7.5
10
12.5
Log of Earnings
59
FIGURE 4.7: COTINUATION.
Counterfactual Supply & Demand
1.6
f(.)
1.2
p.a.
s&d
0.8
5th
25th
0.4
75th
95th
0
2.5
5
7.5
10
12.5
Log of earnings
Residual
1.6
s&d
1.2
f(.)
1988
5th
0.8
25th
75th
0.4
95th
0
2.5
5
7.5
10
Log of earnings
12.5
60
FIGURE 4.7A: DECOMPOSITION OF CHANGES IN DENSITY FUNCTIONS, 1994-96
Actual Densities
1.2
1994
1996
0.8
f(.)
5th
25th
0.4
75th
95th
0
2.5
5
7.5
10
12.5
Log of Earnings
Trade Counterfactual
1.2
0.8
f(.)
1996
trade
5th
0.4
25th
75th
0
95th
2.5
5
7.5
10
12.5
Log of Earnings
Personal Attributes Counterfactual
1.2
trade
p.a.
0.8
f(.)
5th
25th
75th
0.4
95th
0
2.5
5
7.5
10
12.5
Log of Earnings
61
FIGURE 4.7A: COTINUATION.
Counterfactual Supply & Demand
1.2
p.a.
0.8
s&d
f(.)
5th
25th
0.4
75th
95th
0
2.5
5
7.5
10
Log of earnings
12.5
Residual
1.2
s&d
0.8
1994
f(.)
5th
25th
0.4
75th
95th
0
2.5
5
7.5
10
12.5
Log of earnings
62
FIGURE 4.7B: DECOMPOSITION OF CHANGES IN DENSITY FUNCTIONS, 1996-98
Actual Densities
1.2
f(.)
0.8
1996
1998
0.4
0
2.5
5
7.5
Log of Earnings
10
12.5
Trade Counterfactual
1.2
f(.)
0.8
1998
trade
0.4
0
2.5
5
7.5
10
12.5
Log of Earnings
Personal Attributes Counterfactual
1.2
f(.)
0.8
trade
p.a.
0.4
0
2.5
5
7.5
10
12.5
Log of Earnings
63
FIGURE 4.7B: COTINUATION.
Counterfactual Supply & Demand
1.2
f(.)
0.8
p.a.
s&d
0.4
0
2.5
5
7.5
Log of earnings
10
12.5
Residual
1.2
f(.)
0.8
s&d
1996
0.4
0
2.5
5
7.5
10
Log of earnings
12.5
64
REFERENCES
Amemiya, T. (1985). Advanced Econometrics. Harvard University Press. Cambridge, MA.
Bhagwati, J. and V. Dehejia (1994). Freer Trade and Wages of the Unskilled, Is Marx Striking
Again?. In J. Bhagwati and M. Kosters (eds.). Trade and Wages: Leveling Wages Down?.
American Enterprise Institute. Washington, DC.
Bejarano, J. (1992). Productividad y Empleo en el Campo Colombiano. Tercer Mundo Editores.
Santafé de Bogotá, DC. Colombia.
Bound, J. and G. Johnson (1992). Changes in the structure of Wages in the 1980s: An Evaluation
of Alternative Explanations. American Economic Review. 82: 371-92.
Borjas, G., Freeman, R. and L. Katz (1992) (eds.). Immigration and the Work Force: Economic
Consequences for the United States and the Source Areas. University of Chicago Press.
Chicago, Il.
Borjas, G., Freeman, R. and L. Katz (1996). Searching for the Effect of Immigration in the Labor
Market. American Economic Review. 84(may): 10-16.
Borjas, G., Freeman, R. and L. Katz (1998). How Much do Immigration and Trade Affect Labor
Market Outcomes?. Brookings Papers on Economic Activity. 1, 1-67.
Caligiuri, P. and C.R. Rao (1993). Analysis of Ordered Data through Appropriate Scaling. In
C.R. Rao (ed.) Handbook of Statistics 9. North-Holland, Amsterdam, Netherlands.
Collins, S. (ed.), (1998). Imports, Exports and the American Worker. The Brookings Institution.
Washington, D.C.
Cline, W. (1997). Trade and Income Distribution. Institute for International Economics.
Washington, D.C.
Deardorff, A. and R. Staiger (1988). An Interpretation of the Factor Content of Trade. Journal of
International Economics. 24: 93-107.
DiNardo, J., Fortin, N. and T. Lemieux (1996). Labor Market Institutions and the Distribution of
Wages, 1973-1992: A Semiparametric Approach. Econometrica. 64(5): 1001-1044.
Oxford, UK.
DNP (1995). Guía del Usuario del Banco de Fuentes Primarias, BFP. DNP. Santafé de Bogotá
DC, Colombia.
DNP (1991-99). Indicadores de Coyuntura Económica. Various numbers. Santafé de Bogotá,
DC.
65
Freeman, R. (1975). Overinvestment in College Training?. Journal of Human Resources. X, 287311.
Freeman, R. (1980). An Empirical Analysis of the Fixed Coefficient Manpower Requirements
Model, 1960-1970. Journal of Human Resources. XV, 176-99.
Freeman, R. (1998). Will Globalization Dominate Labor Market Outcomes?. In S. Collins (ed.).
Imports, Exports and the American Worker. Brookings Institution. Washington, DC.
Greene, W.H. (1997). Econometric Analysis. Prentice Hall. London, U.K.
Heckman, J. (1976). The Common Structure of Statistical Models of Truncation, Sample
Selection, and Limited Dependent Variables and a Simple Estimators for Such Models.
Annals of Economic and Social Measurement. 5: 475-492.
Heckman, J. (1979). Sample Selection Bias as a Specification Error. Econometrica.47: 153-161.
Hogg, R.V. and A.T. Craig (1978). Introduction to Mathematical Statistics, Fourth Edition.
Macmillan Publishing Co., Inc. and Collier Macmillan Publishers.
Juhn, C., Murphy, K. and B. Pierce (1993). Wage Inequality and the Rise in Returns to Skill.
Journal of Political Economy. 101(3), 410-42.
Katz, L. and K. Murphy (1992). Changes in Relative Wages 1963-87: Supply and Demand
Factors. Quarterly Journal of Economics. 107: 35-78.
Kosters, M. (ed.), (1991). Workers and Their Wages: Changes Patterns in the United States.. The
AEI Press. Washington, D.C.
Krugman, P. (1995). Technology, Trade and Factor Prices. NBER Working Paper No 5355.
Cambridge, MA.
Murphy, K., Riddell, C. and P. Romer. (1998). Wages, Skill and Technology in the United States
and Canada. NBER Working Paper 6638. National Bureau of Economic Research.
Cambridge, MA.
Murphy, K. and F. Welch (1991). The Role of International Trade in Wage Differentials. In M.
Kosters (ed.) Workers and Their Wages: Changing Patterns in the United States. The
AEI Press, Washington, DC.
Oaxaca, R. (1973). Male-Female Wage Differentials in Urban Labor Markets. International
Economic Review. 14: 673-709.
Sánchez, F. and J. Núñez (1998). Educación y Salarios Relativos en Colombia 1976-1995.
Determinantes, Evolución e Implicaciones para la Distribución del Ingreso. Archivos de
Macroeconomía. Documento 74. DNP. Santafé de Bogotá, DC. Colombia.
Sánchez, F. and J. Núñez (1998a). Descomposición de la desigualdad del Ingreso Laboral Urbano
66
en Colombia: 1976-1997. Archivos de Macroeconomía. Documento 86. DNP. Santafé de
Bogotá, DC. Colombia.
Santamaría, M. (1997). Describing the Evolution of Income Inequality: Colombia 1978-95. First
Dissertation Paper. Mimeo. Washington, D.C.
Santamaría, M. (1999). Nonparametric Density Estimation and Regression: an Application to the
Study of Income Inequality and Poverty, Colombia (1978-1996). Second Dissertation
Paper. Mimeo. Washington, D.C.
Stata Corp (1999). Reference Manual, Release 6.0. Stata Press. College Station, TX.
Stata Corp (1999). User’s Guide, Release 6.0. Stata Press. College Station, TX.
Vélez, C.E., Bouillon, C. and A. Kugler (1999). The Reversal of Inequality Gains in Urban
Colombia, 1978-1995: A Combination of Persistent and Fluctuating Forces. Mimeo.
Interamerican Development Bank. Washington, D.C.
67
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
1
La coyuntura económica en Colombia
y Venezuela
Andrés Langebaek
Patricia Delgado
Fernando Mesa Parra
Octubre 1992
2
La tasa de cambio y el comercio
colombo-venezolano
Fernando Mesa Parra
Andrés Langebaek
Noviembre 1992
3
¿Las mayores exportaciones colombianas
de café redujeron el precio externo?
Carlos Esteban Posada
Andrés Langebaek
Noviembre 1992
4
El déficit público: una perspectiva
macroeconómica
Jorge Enrique Restrepo
Juan Pablo Zárate
Carlos Esteban Posada
Noviembre 1992
5
El costo de uso del capital en Colombia
Mauricio Olivera
Diciembre 1992
6
Colombia y los flujos de capital privado
a América Latina
Andrés Langebaek
Febrero 1993
7
Infraestructura física. “Clubs de
convergencia” y crecimiento
económico
José Dario Uribe
Febrero 1993
8
El costo de uso del capital: una nueva
estimación (Revisión)
Mauricio Olivera
Marzo 1993
9
Dos modelos de transporte de carga por
carretera
Carlos Esteban Posada
Edgar Trujillo Ciro
Alvaro Concha
Juan Carlos Elorza
Marzo 1993
10
La determinación del precio interno del
café en un modelo de optimización
intertemporal
Carlos Felipe Jaramillo
Carlos Esteban Posada
Edgar Trujillo
Abril 1993
11
El encaje óptimo
Edgar Trujillo Ciro
Carlos Esteban Posada
Mayo 1993
12
Crecimiento económico, “Capital
humano” y educación: la teoría y el
caso colombiano posterior a 1945
Carlos Esteban Posada
Junio 1993
13
Estimación del PIB trimestral según los
componentes del gasto
Rafael Cubillos
Junio 1993
Fanny Mercedes Valderrama
14
Diferencial de tasas de interés y flujos
de capital en Colombia (1980-1993)
Andrés Langebaek
15
Empleo y capital en Colombia: nuevas
estimaciones (1950-1992)
Adriana Barrios
Septiembre 1993
Marta Luz Henao
Carlos Esteban Posada
Fanny Mercedes Valderrama
Diego Mauricio Vásquez
Agosto 1993
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
16
Productividad, crecimiento y ciclos en
la economía colombiana (1967-1992)
Carlos Esteban Posada
Septiembre 1993
17
Crecimiento económico y apertura en Chile
y México y perspectivas para Colombia
Fernando Mesa Parra
Septiembre 1993
18
El papel del capital público en la
producción, inversión y el crecimiento
económico en Colombia
Fabio Sánchez Torres
Octubre 1993
19
Tasa de cambio real y tasa de cambio
de equilibrio
Andrés Langebaek
Octubre 1993
20
La evolución económica reciente:
dos interpretaciones alternativas
Carlos Esteban Posada
Noviembre 1993
21
El papel de gasto público y su
financiación en la coyuntura actual:
algunas implicaciones complementarias
Alvaro Zarta Avila
Diciembre 1993
22
Inversión extranjera y crecimiento
económico
Alejandro Gaviria
Javier Alberto Gutiérrez
Diciembre 1993
23
Inflación y crecimiento en Colombia
Alejandro Gaviria
Carlos Esteban Posada
Febrero 1994
24
Exportaciones y crecimiento en Colombia
Fernando Mesa Parra
Febrero 1994
25
Experimento con la vieja y la nueva teoría
del crecimiento económico (¿porqué crece
tan rápido China?)
Carlos Esteban Posada
Febrero 1994
26
Modelos económicos de criminalidad y la
posibilidad de una dinámica prolongada
Carlos Esteban Posada
Abril 1994
27
Regímenes cambiarios, política
macroeconómica y flujos de capital en
Colombia
Carlos Esteban Posada
Abril 1994
28
Comercio intraindustrial: el caso
colombiano
Carlos Pombo
Abril 1994
29
Efectos de una bonanza petrolera a la luz
de un modelo de optimización
intertemporal
Hernando Zuleta
Juan Pablo Arango
Mayo 1994
30
.
Crecimiento económico y productividad
en Colombia: una perspectiva de largo
plazo (1957-1994)
Sergio Clavijo
Junio 1994
31
Inflación o desempleo:
¿Acaso hay escogencia en Colombia?
Sergio Clavijo
Agosto 1994
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
32
La distribución del ingreso y el sistema financiero
Edgar Trujillo Ciro
Agosto 1994
33
La trinidad económica imposible en
Colombia: estabilidad cambiaria,
independencia monetaria y flujos de
capital libres
Sergio Clavijo
Agosto 1994
34
¿’Déjà vu?: tasa de cambio, deuda externa
y esfuerza exportador en Colombia.
Sergio Clavijo
Mayo 1995
35
La crítica de Lucas y la inversión en
Colombia: nueva evidencia
Mauricio Cárdenas
Mauricio Olivera
Septiembre 1995
36
Tasa de Cambio y ajuste del sector
externo en Colombia
Fernando Mesa Parra
Dairo Estrada
Septiembre 1995
37
Análisis de la evolución y
composición del Sector Público
Mauricio Olivera G.
Septiembre 1995
Manuel Fernando Castro Q.
Fabio Sánchez T.
38
Incidencia distributiva del IVA en un
modelo del ciclo de vida
Juan Carlos Parra Osorio
Fabio José Sánchez T.
Octubre 1995
39
Por qué los niños pobres no van a la
escuela? (Determinantes de la asistencia
escolar en Colombia)
Fabio Sánchez Torres
Jairo Augusto Núñez M.
Noviembre 1995
40
Matriz de Contabilidad Social 1992
Fanny M. Valderrama
Javier Alberto Gutiérrez
Diciembre 1995
41
Multiplicadores de Contabilidad
Derivados de la Matriz de Contabilidad
Social
Javier Alberto Gutiérrez
Fanny M. Valderrama G.
Enero 1996
42
El ciclo de referencia de la economía
colombiana
Martin Maurer
María Camila Uribe S.
Febrero 1996
43
Impacto de las transferencias intergubernamentales en la distribución
interpersonal del ingreso en Colombia
Juan Carlos Parra Osorio
Marzo 1996
44
Auge y colapso del ahorro empresarial
en Colombia 1983-1994
Fabio Sánchez Torres
Abril 1996
Guillermo Murcia Guzmán
Carlos Oliva Neira
45
Evolución y comportamiento del gasto
público en Colombia 1950-1994
Cielo María Numpaque
Ligia Rodríguez Cuestas
Mayo 1996
46
Los efectos no considerados de la
apertura económica en el mercado
laboral industrial
Fernando Mesa Parra
Javier Alberto Gutiérrez
Mayo 1996
47
Un modelo de Financiamiento óptimo de un
aumento permanente en el gasto público:
Una ilustración con el caso colombiano.
Alvaro Zarta Avila
Junio 1996
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
48
Estadísticas descriptivas del mercado laboral
masculino y femenino en Colombia: 1976 -1995
Rocío Ribero M.
Carmen Juliana García B.
Agosto 1996
49
Un sistema de indicadores líderes para Colombia
Martín Maurer
María Camila Uribe
Javier Birchenall
Agosto 1996
50
Evolución y determinantes de la productividad
en Colombia: Un análisis global y sectorial
Fabio Sánchez Torres
Jorge Iván Rodríguez
Jairo Núñez Méndez
Agosto 1996
51
Gobernabilidad y Finanzas Públicas en Colombia
César A. Caballero R
Noviembre 1996
52
Tasas Marginales Efectivas de Tributación
en Colombia
Mauricio Olivera G.
Noviembre 1996
53
Un modelo keynesiano para la economía
colombiana
Fabio José Sánchez T.
Clara Elena Parra
Febrero 1997
54
Trimestralización del Producto Interno Bruto
por el lado de la oferta.
Fanny M. Valderrama
Febrero 1997
55
Poder de mercado, economías de escala,
complementariedades intersectoriales y
crecimiento de la productividad en la
industria colombiana.
Juán Mauricio Ramírez
Marzo 1997
56
Estimación y calibración de sistemas flexibles de gasto.
Orlando Gracia
Gustavo Hernández
Abril 1997
57
Mecanismos de ahorro e Inversión en las Empresas
Públicas Colombianas: 1985-1994
Fabio Sánchez Torres
Guilllermo Murcia G.
Mayo 1997
58
Capital Flows, Savings and investment in Colombia
1990-1996
José Antonio Ocampo G.
Camilo Ernesto Tovar M.
Mayo 1997
59
Un Modelo de Equilibrio General Computable con
Competencia imperfecta para Colombia
Juan Pablo Arango
Orlando Gracia
Gustavo Hernández
Juan Mauricio Ramírez
Junio 1997
60
El cálculo del PIB Potencial en Colombia
Javier A. Birchenall J.
Julio 1997
61
Determinantes del Ahorro de los hogares.
Explicación de su caída en los noventa.
Alberto Castañeda C.
Gabriel Piraquive G.
Julio 1997
62
Los ingresos laborales de hombres y
mujeres en Colombia: 1976-1995
Rocío Ribero
Claudia Meza
Agosto 1997
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
63
Determinantes de la participación laboral de
hombres y mujeres en Colombia: 1976-1995
Rocío Ribero
Claudia Meza
Agosto 1997
64
Inversión bajo incertidumbre en la Industria
Colombiana: 1985-1995
Javier A. Birchenall
Agosto 1997
65
Modelo IS-LM para Colombia. Relaciones de
largo plazo y fluctuaciones económicas.
Jorge Enrique Restrepo
Agosto 1997
66
Correcciones a los Ingresos de las Encuestas de
hogares y distribución del Ingreso Urbano en
Colombia.
Jairo A. Núñez Méndez
Jaime A. Jiménez Castro
Septiembre 1997
67
Ahorro, Inversión y Transferencias en las Entidades
Territoriales Colombianas
Fabio Sánchez Torres
Mauricio Olivera G.
Giovanni Cortés S.
Octubre 1997
68
Efectos de la Tasa de cambio real sobre la Inversión
industrial en un Modelo de transferencia de precios
Fernando Mesa Parra
Leyla Marcela Salguero
Fabio Sánchez Torres
Octubre 1997
69
Convergencia Regional: Una revisión del caso
Colombiano.
Javier A. Birchenall
Guillermo E. Murcia G.
Octubre 1997
70
Income distribution, human capital and economic
growth in Colombia.
Javier A. Birchenall
Octubre 1997
71
Evolución y determinantes del Ahorro del
Gobierno Central.
Fabio Sánchez Torres
Ma. Victoria Angulo
Noviembre 1997
72
Macroeconomic Perforrmance and Inequality in
Colombia: 1976-1996
Raquel Bernal
Mauricio Cárdenas
Jairo Núñez Méndez
Fabio Sánchez Torres
Diciembre 1997
73
Liberación comercial y salarios en Colombia:
1976-1994
Donald Robbins
Enero 1998
74
Educación y salarios relativos en Colombia: 1976-1995
Determinantes, evolución e implicaciones para
la distribución del Ingreso
Jairo Núñez Méndez
Fabio Sánchez Torres
Enero 1998
75
La tasa de interés “óptima”
Carlos Esteban Posada
Edgar Trujillo Ciro
Febrero 1998
76
Los costos económicos de la criminalidad
y la violencia en Colombia: 1991-1996
Edgar Trujillo Ciro
Martha Elena Badel
Marzo 1998
77
Elasticidades Precio y Sustitución para
la Industria Colombiana
Juán Pablo Arango
Orlando Gracia
Gustavo Hernández
Marzo 1998
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
78
Flujos Internacionales de Capital en Colombia:
Un enfoque de Portafolio
Ricardo Rocha García
Fernando Mesa Parra
Marzo 1998
79
Macroeconomía, ajuste estructural y equidad en
Colombia: 1978-1996
José Antonio Ocampo
María José Pérez
Camilo Ernesto Tovar
Francisco Javier Lasso
Marzo 1998
80
La Curva de Salarios para Colombia.
Una Estimación de las Relaciones entre el Desempleo,
la Inflación y los Ingresos Laborales, 1984- 1996.
Fabio Sánchez Torres
Jairo Núñez Méndez
Marzo 1998
81
Participación, Desempleo y Mercados
Laborales en Colombia
Jaime Tenjo G.
Rocio Ribero M.
Abril 1998
82
Reformas comerciales, márgenes de beneficio y
productividad en la industria colombiana
Juán Pablo Arango
Orlando Gracia
Gustavo Hernández
Juán Mauricio Ramírez
Abril 1998
83
Capital y Crecimiento Económico en un Modelo
Dinámico: Una presentación de la dinámica
Transicional para los casos de EEUU y Colombia
Alvaro Zarta Avila
Mayo 1998.
84
Determinantes de la Inversión en Colombia:
Evidencia sobre el capital humano y la violencia.
Clara Helena Parra
Junio 1998.
85
Mujeres en sus casas: Un recuento de la población
Femenina económicamente activa
Piedad Urdinola Contreras
Junio 1998.
86
Descomposición de la desigualdad del Ingreso laboral
Urbano en Colombia: 1976-1997
Fabio Sánchez Torres
Jairo Núñez Méndez
Junio 1998.
87
El tamaño del Estado Colombiano Indicadores y
tendencias 1976-1997
Angela Cordi Galat
Junio 1998.
88
Elasticidades de sustitución de las importaciones
Para la economía colombiana.
Gustavo Hernández
Junio 1998.
89
La tasa natural de desempleo en Colombia
Martha Luz Henao
Norberto Rojas
Junio 1998.
90
The role of shocks in the colombian economy
Ana María Menéndez
Julio 1998.
91
The determinants of Human Capital Accumulation in
Donald J. Robbins
Colombia, with implications for Trade and Growth Theory
Julio 1998.
92
Estimaciones de funciones de demanda de trabajo
dinámicas para la economía colombiana, 1980-1996
Alejandro Vivas Benítez
Stefano Farné
Dagoberto Urbano
Julio 1998.
93
Análisis de las relaciones entre violencia y equidad
Alfredo Sarmiento
Lida Marina Becerra
Agosto 1998.
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
94
Evaluación teórica y empírica de las exportaciones
no tradicionales en Colombia
Fernando Mesa Parra
María Isabel Cock
Angela Patricia Jiménez
Agosto 1998.
95
Valoración económica del empleo doméstico femenino
no remunerado, en Colombia, 1978-1993
Piedad Urdinola Contreras
Agosto 1998.
96
Eficiencia en el Gasto Público de Educación.
María Camila Uribe
Agosto 1998.
97
El desempleo en Colombia: tasa natural, desempleo
cíclico y estructural y la duración del desempleo.
1976-1998.
Jairo Núñez M.
Raquel Bernal S.
Septiembre 1998.
98
Productividad y retornos sociales del Capital humano:
Microfundamentos y evidencia para Colombia.
Francisco A. González R.
Carolina Guzmán R.
Angela L. Pachón G.
Noviembre 1998.
99
Reglas monetarias en Colombia y Chile
Jorge E. Restrepo L.
Enero 1999.
100
Inflation Target Zone: The Case of Colombia
1973-1994
Jorge E. Restrepo L.
Febrero 1999.
101
¿ Es creíble la Política Cambiaria en Colombia?
Carolina Hoyos V.
Marzo 1999.
102
La Curva de Phillips, la Crítica de Lucas y
la persistencia de la inflación en Colombia
Javier A.Birchenall
Abril 1999.
103
Un modelo macroeconométrico para la economía
Colombiana
Javier A.Birchenall
Juan Daniel Oviedo
Abril 1999.
104
Una revisión de la literatura teórica y la experiencia
Internacional en regulación
Marcela Eslava Mejía
Abril 1999.
105
El transporte terrestre de carga en Colombia
Documento para el Taller de Regulación.
Marcela Eslava Mejía
Abril 1999.
Eleonora Lozano Rodríguez
106
Notas de Economía Monetaria. (Primera Parte)
Juan Carlos Echeverry G.
Abril 1999.
107
Ejercicios de Causalidad y Exogeneidad para
Ingresos salariales nominales públicos y privados
Colombianos (1976-1997).
Mauricio Bussolo
Orlando Gracia
Camilo Zea
Mayo 1999.
108
Real Exchange Rate Swings and Export Behavior:
Explaining the Robustness of Chilean Exports.
Felipe Illanes
Mayo 1999.
109
Segregación laboral en las 7 principales ciudades
del país.
Piedad Urdinola
Mayo 1999.
110
Estimaciones trimestrales de la línea de pobreza y
sus relaciones con el desempeño macroeconómico
Colombiano. (1977-1997)
Jairo Núñez Méndez
Fabio José Sánchez T.
Mayo 1999
111
Costos de la corrupción en Colombia.
Marta Elena Badel
Mayo 1999
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
112
Relevancia de la dinámica transicional para el
crecimiento de largo plazo: Efectos sobre las tasas de
interés real, la productividad marginal y la estructura
de la producción para los casos de EEUU y Colombia..
Alvaro Zarta
Junio 1999
113
La recesión actual en Colombia: Flujos, Balances y
Política anticíclica
Juan Carlos Echeverry
Junio 1999
114
Monetary Rules in a Small Open Economy
Jorge E. Restrepo L.
Junio 1999
115
El Balance del Sector Público y la Sostenibilidad
Fiscal en Colombia
Juan Carlos Echeverry
Gabriel Piraquive
Natalia Salazar
Ma. Victoria Angulo
Gustavo Hernández
Cielo Ma. Numpaque
Israel Fainboim
Carlos Jorge Rodriguez
Junio 1999
116
Crisis y recuperación de las Finanzas Públicas.
Lecciones de América Latina para el caso colombiano.
Marcela Eslava Mejía
Julio 1999
117
Complementariedades Factoriales y Cambio Técnico
en la Industria Colombiana.
Gustavo Hernández
Juan Mauricio Ramírez
Julio 1999
118
¿Hay un estancamiento en la oferta de crédito?
Juan Carlos Echeverry
Natalia Salazar
Julio 1999
119
Income distribution and macroeconomics in Colombia.
Javier A. Birchenall J.
Julio 1999.
120
Transporte carretero de carga. Taller de regulación.
DNP-UMACRO. Informe final.
Juan Carlos Echeverry G. Agosto 1999.
Marcela Eslava Mejía
Eleonora Lozano Rodriguez
121
¿ Se cumplen las verdades nacionales a nivel regional?
Primera aproximación a la construcción de matrices de
contabilidad social regionales en Colombia.
Nelly.Angela Cordi Galat
Agosto 1999.
122
El capital social en Colombia.
La medición nacional con el BARCAS
Separata N° 1 de 5
John SUDARSKY
Octubre 1999.
123
El capital social en Colombia.
La medición nacional con el BARCAS
Separata N° 2 de 5
John SUDARSKY
Octubre 1999.
124
El capital social en Colombia.
La medición nacional con el BARCAS
Separata N° 3 de 5
John SUDARSKY
Octubre 1999.
125
El capital social en Colombia.
La medición nacional con el BARCAS
Separata N° 4 de 5
John SUDARSKY
Octubre 1999.
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
126
El capital social en Colombia.
La medición nacional con el BARCAS
Separata N° 5 de 5
John SUDARSKY
Octubre 1999.
127
The Liquidity Effect in Colombia
Jorge E. Restrepo
Noviembre 1999.
128
Upac: Evolución y crisis de un modelo de desarrollo.
Juan C Echeverry
Orlando Gracia
B. Piedad Urdinola
Diciembre 1999.
129
Confronting fiscal imbalances via intertemporal
Economics, politics and justice: the case of Colombia
Juan C Echeverry
Verónica Navas-Ospina
Diciembre 1999.
130
La tasa de interés en la coyuntura reciente en Colombia.
Jorge Enrique Restrepo
Edgar Trujillo Ciro
Diciembre 1999.
131
Los ciclos económicos en Colombia. Evidencia
Empírica (1977-1998)
Jorge Enrique Restrepo
José Daniel Reyes Peña
Enero 2000.
132
Colombia'natural trade partners and its bilateral
Trade performance: Evidence from 1960 to 1996
Hernán Eduardo Vallejo
Enero 2000.
133
Los derechos constitucionales de prestación y sus
Implicaciones económico- políticas. Los casos del
derecho a la salud y de los derechos de los reclusos
Luis Carlos Sotelo
Febrero 2000.
134
La reactivación productiva del sector privado colombiano Luis Alberto Zuleta
(Documento elaborado para el BID)
Marzo 2000.
135
Geography and Economic Development:
A Municipal Approach for Colombia.
Fabio José Sánchez T.
Jairo Núñez Méndez
Marzo 2000.
136
La evaluación de resultados en la modernización
del Estado en América Latina. Restricciones y
Estrategia para su desarrollo.
Eduardo Wiesner Durán
Abril 2000.
137
La regulación de precios del transporte de carga por
Carretera en Colombia.
Marcela Eslava Mejía
Abril 2000.
138
El conflicto armado en Colombia.
Una aproximación a la teoría de juegos.
Yuri Gorbaneff
Flavio Jácome
Julio 2000.
139
Determinación del consumo básico de agua potable
subsidiable en Colombia.
Juan Carlos Junca Salas
Noviembre 2000.
Incidencia fiscal de los incentivos tributarios
Juan Ricardo Ortega
Noviembre 2000.
Gabriel Armando Piraquive
Gustavo Adolfo Hernández
Carolina Soto Losada
Sergio Iván Prada
Juan Mauricio Ramirez
.
140
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
141
Exenciones tributarias:
Costo fiscal y análisis de incidencia
Gustavo A. Hernández
Carolina Soto Losada
Sergio Iván Prada
Juan Mauricio Ramirez
Diciembre 2000
142
La contabilidad del crecimiento, las dinámicas
transicionales y el largo plazo:
Una comparación internacional de 46 países y
una presentación de casos de economías tipo:
EEUU, Corea del Sur y Colombia.
Alvaro Zarta Avila
Febrero 2001
143
¿Nos parecemos al resto del mundo?
El Conflicto colombiano en el contexto internacional.
Juan Carlos Echeverry G.
Natalia Salazar Ferro
Verónica Navas Ospina
Febrero 2001
144
Inconstitucionalidad del Plan Nacional de Desarrollo:
causas, efectos y alternativas.
Luis Edmundo Suárez S.
Diego Mauricio Avila A.
Marzo 2001
145
La afiliación a la salud y los efectos redistributivos
de los subsidios a la demanda.
Hernando Moreno G.
Abril 2001
146
La participación laboral: ¿qué ha pasado y qué
podemos esperar?
Mauricio Santamaría S.
Norberto Rojas Delgadillo
Abril 2001
147
Análisis de las importaciones agropecuarias en la
década de los Noventa.
Gustavo Hernández
Juan Ricardo Perilla
Mayo 2001
148
Impacto económico del programa de Desarrollo
alternativo del Plan Colombia
Gustavo A. Hernández
Sergio Iván Prada
Juan Mauricio Ramírez
Mayo 2001
149
Análisis de la presupuestación de la inversión de
la Nación.
Ulpiano Ayala Oramas
Mayo 2001
150
DNPENSION: Un modelo de simulación para estimar
el costo fiscal del sistema pensional colombiano.
Juan Carlos Parra Osorio
Mayo 2001
151
La oferta de combustible de Venezuela en la frontera
con Colombia: una aproximación a su cuantificación
Hernando Moreno G.
Junio 2001
152
Shocks fiscales y términos de intercambio en el caso
colombiano.
Ómer ÖZAK MUñOZ.
Julio 2001
153
Demanda por importaciones en Colombia:
Una estimación.
Igor Esteban Zuccardi
Julio 2001
154
Elementos para mejorar la adaptabilidad del
mercado laboral colombiano.
Mauricio Santa María S.
Norberto Rojas Delgadillo
Agosto 2001
155
¿Qué tan poderosas son las aerolíneas
colombianas? Estimación de poder de
mercado de las rutas colombianas.
Ximena Peña Parga
Agosto 2001
ARCHIVOS DE ECONOMIA
No
Título
Autores
Fecha
156
Elementos para el debate sobre una nueva reforma
pensional en Colombia.
Juan Carlos Echeverry
Andrés Escobar Arango
César Merchán Hernández
Gabriel Piraquive Galeano
Mauricio Santa María S.
Septiembre 2001
157
Agregando votos en un sistema altamente
desistitucionalizado.
Francisco Gutiérrez Sanín
Octubre 2001
158
Eficiencia -X en el Sector Bancario Colombiano
Carlos Alberto Castro I
Noviembre 2001
159
Determinantes de la calidad de la educación en
Colombia.
Alejandro Gaviria
Jorge Hugo Barrientos
Noviembre 2001
160
Evaluación de la descentralización municipal.
Descentralización y macroeconomía
Fabio Sánchez Torres
Noviembre 2001
161
Impuestos a las transacciones: Implicaciones sobre
el bienestar y el crecimiento.
Rodrigo Suescún
Noviembre 2001
162
Strategic Trade Policy and Exchange Rate Uncertainty
Fernando Mesa Parra
Noviembre 2001
163
Evaluación de la descentralización municipal en
Colombia. Avances y resultados de la descentralización
Política en Colombia
Alberto Maldonado C.
Noviembre 2001
164
Choques financieros, precios de activos y recesión
en Colombia.
Alejandro Badel Flórez
Noviembre 2001
165
Evaluación de la descentralización municipal en
Colombia. ¿Se consolidó la sostenibilidad fiscal de los
municipios colombianos durante los años noventa.
Juan Gonzalo Zapata
Olga Lucía Acosta
Adriana González
Noviembre 2001
166
Evaluación de la descentralización municipal en
Colombia. La descentralización en el Sector de
Agua potable y Saneamiento básico.
Maria Mercedes Maldonado Noviembre 2001
Gonzalo Vargas Forero
167
Evaluación de la descentralización municipal en
Colombia. La relación entre corrupción y proceso
de descentralización en Colombia.
Edgar González Salas
Diciembre 2001
168
Evaluación de la descentralización municipal en
Colombia. Estudio general sobre antecedentes,
diseño, avances y resultados generales del proceso de
descentralización territorial en el Sector Educativo.
Carmen Helena Vergara
Mary Simpson
Diciembre 2001
169
Evaluación de la descentralización municipal en
Colombia. Componente de capacidad institucional.
Edgar González Salas
Diciembre 2001
170
Evaluación de la descentralización municipal en
Colombia. Evaluación de la descentralización en
Salud en Colombia.
Iván Jaramillo Pérez
Diciembre 2001
171
External Trade, Skill, Technology and the recent
increase of income inequality in Colombia
Mauricio Santa María S.
Diciembre 2001
Descargar