WPS8432
Policy Research Working Paper 8432
Inequality of Opportunity in South Caucasus
Alan Fuchs
Sailesh Tiwari
Akhmad Rizal Shidiq
Poverty and Equity Global Practice
May 2018
Policy Research Working Paper 8432
Abstract
This paper discusses equality of opportunity in Armenia, among people who have access to these jobs, the share of the
Azerbaijan, and Georgia, with an emphasis on access to total inequality of opportunity that may be characterized
labor market opportunities. It develops an inequality of as unfair is relatively high. Armenia and Azerbaijan stand
opportunity index on access to good jobs and decomposes out for the significant share of inequality in access to good
the contributing factors in the prevailing inequality. Then, it jobs associated with gender differences. Fourth, the analy-
discusses the extent to which inequality in accessing human sis on access to education and basic human capital inputs
capital inputs among individuals during the early formative in the earlier, formative stages of life shows that learning
years may affect access to good jobs. The main takeaways performance in the South Caucasus tends to be poor and
are as follows. First, connections play an important role in unequal across the life circumstances of children. None-
obtaining access to good jobs in the South Caucasus, high- theless, the coverage rates of basic human capital inputs
lighting the unfairness in processes in the sub-region’s labor are generally high; the relatively narrow inequalities arise
markets. Second, access to good jobs—defined as work for mostly from spatial disparities. These results indicate that
20 hours or more a week and work under contract or with addressing the deep structural inequalities shaping the land-
tenure—is low in the South Caucasus in comparison with scape of opportunity in the South Caucasus must be a key
other parts of Eastern Europe and Central Asia. Third, even consideration in any strategy to share prosperity sustainably.
This paper is a product of the Poverty and Equity Global Practice. It is part of a larger effort by the World Bank to provide
open access to its research and make a contribution to development policy discussions around the world. Policy Research
Working Papers are also posted on the Web at http://www.worldbank.org/research. The authors may be contacted at
afuchs@worldbank.org.
The Policy Research Working Paper Series disseminates the findings of work in progress to encourage the exchange of ideas about development
issues. An objective of the series is to get the findings out quickly, even if the presentations are less than fully polished. The papers carry the
names of the authors and should be cited accordingly. The findings, interpretations, and conclusions expressed in this paper are entirely those
of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and
its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.
Produced by the Research Support Team
Inequality of Opportunity in
South Caucasus
Alan Fuchs, Sailesh Tiwari and Akhmad Rizal Shidiq1
Keywords: Equality of Opportunity, Intergenerational Mobility, Education, Employment
JEL Codes: I240, I280, I310, J620, O15
1 Fuchs (afuchs@worldbank.org), Senior Economist in the Poverty and Equity Global Practice, World Bank. Tiwari
(stiwari@worldbank.org), Senior Economist at the Poverty and Equity Global Practice, World Bank. Shidiq
(a.r.shidiq@hum.leidenuniv.nl), Assistant Professor, Faculty of Humanities, Leiden Institute for Area Studies, Leiden
University. The authors are grateful to Genevieve Boyreau, Cesar Cancho, Moritz Meyer and Ana Maria Munoz Boudet
for providing comments and support and Gabriel Lara Ibarra and Victor Sulla for providing comments to an earlier
version of this document. The work was carried out under the overall guidance of Mercy Miyang Tembon and Luis-
Felipe Lopez-Calva who provided substantive inputs at several stages of development of this work. The findings,
interpretations, and conclusions in this research note are entirely those of the authors. They do not necessarily represent
the views of the World Bank Group, its Executive Directors, or the countries they represent.
1. Introduction
For the countries of the South Caucasus sub-region, the issue of inequality is particularly
important because of their trajectory as economies in transition from a planning system with a
strong tendency to redistribution to a market-oriented economy. As in any country in transition,
the expectation of economic improvement is high. Reform takes time, and adjustment may lead
to adverse short-run distributional effects unknown in the past. In fact, a rise in inequality in
outcomes may reflect increasing returns to assets, including human capital, indicating
improvement in economic conditions and a functioning market mechanism. Nevertheless,
perceptions of inequality in opportunities or outcomes may also become more salient during
transition considering that populations in the South Caucasus regard their family life and the
lives of their parents before the transition as the crucial benchmark in evaluating current life
situations (Tiwari et al. 2018).
Inequality in outcomes such as income or consumption may influence other economic indicators
of progress. For example, pressure for political redistribution that distorts the labor market or
credit market imperfections that prevent the poor form investing in high return-human capital,
can impact rates of investment and economic growth, thus lowering average productivity
(Alesina and Rodrik 1994; Piketty 1997). However, the empirical findings are mixed. Barro (2000)
finds that initial inequality has adverse effects on growth in developing countries, but not in
developed countries. Forbes (2000) uses panel data to control for time-invariant characteristics
and eliminate potential sources of omitted-variable bias to find that the relationship between
inequality and growth is positive. Banerjee and Duflo (2003) note that it is not possible to interpret
any of this evidence as causal because variations in inequality are likely to be correlated with a
range of unobservable factors associated with growth.
Nonetheless, inequality in outcomes is probably not a pressing issue in South Caucasus. While
countries in Eastern Europe and Central Asia show some degree of heterogeneity in terms of
inequality, with Gini coefficients ranging from 0.25 to 0.45, the long-run trend of the Gini in
Georgia has shown no discernible sign of deterioration over the last 15 years (figure 1). In
Armenia, the Gini coefficient based on consumption has hovered between 0.30 and 0.35, which
puts the country at a moderate level of inequality within the region in 2000–15. Armenia also saw
a declining Gini in 2000–09. Data of the Commitment to Equity Institute show that, as of
November 2017, inequalities of outcomes in incomes in Armenia and Georgia are well below the
average among the 30 low- and middle-income countries the institute surveyed.2 For example,
the Gini coefficients of final income, treating contributory pensions as deferred income, in
Armenia and Georgia are 0.36 and 0.38, respectively, in comparison with the 0.41 average Gini of
the 30 countries.
2 SeeCEQ Standard Indicators (database), Commitment to Equity, Inter-American Dialogue, Washington, DC; Center
for Inter-American Policy and Research and Department of Economics, Tulane University, New Orleans,
http://www.commitmentoequity.org/data/.
2
Figure 1: Inequality in the Europe and Central Asia Region, 2000-2015
Moreover, citizen perceptions of inequality are often more important than a country’s actual
inequality because perceptions are more closely related to effective political action, such as voting
behavior or public policy, on preferences in redistribution (Engelhardt and Wagener 2014;
Niehues 2014). Figure 2 confirms that, in Eastern Europe and Central Asia, the correlation
between the net percentage share of the population perceiving that inequality has increased (left-
hand panel) and the demand for redistribution, defined as the net percentage share of the
population agreeing that the gap between the poor and the rich should be reduced, is greater than
the correlation between the country’s actual income inequality, measured by the Gini coefficient,
and the demand for redistribution. Nonetheless, despite quite unremarkable trends in actual
measured consumption inequality, the perception that inequality is widening is quite extensive
among Armenians and Georgians. More than 75 percent of respondents in Armenia and 60
percent in Georgia feel that inequality is widening. Also, the differences in the preference for
redistribution between these two countries are quite stark. Thus, Armenia stands out because the
net proportion of the population that believes the gap between the rich and the poor should be
reduced (more than 80 percent of respondents) is among the largest in the region, while Georgia
is among the countries in the region with a relatively smaller proportion of the population
holding this view.
3
Figure 2: Inequality and demand for redistribution, 2015
Figure 2 does not clarify the sort of inequality that people in South Caucasus perceive as
widening. However, given the prevailing governance issues, especially in Armenia and
Azerbaijan, it seems that inequality in opportunity and the less-transparent allocation of
resources, rather than inequality in outcomes, more likely create dissatisfaction among the
populations. This is particularly evident in Armenia, where the Gini coefficient based on income,
a measure of inequality in outcomes, is low, but the pressure for redistribution is high. Also, in
Armenia, the relevant gap responsible for the popular demand for redistribution is probably not
so much the gap between the poor and the rich, but the gap between the poor and the middle
class and the gap between the poor and the oligarchs who used excessive and uncontrollable
privatization during the early 1990s to accumulate incredibly vast amounts of wealth. The social
contract collapsed when the rest of the population realized they would face hard times because
of unequal market access in a context of a lack of competition and inadequate regulation, creating
a process wherein even a good education and substantial effort do not facilitate welfare
improvements among individuals given the severe governance problems generated by political
and economic elites. In many ways, such sentiments about inequalities in the process, rather than
outcomes, are likely relevant in Azerbaijan and Georgia as well because these countries share
similar institutional problems resulting from imperfect political and economic transition during
the 1990s.
Against the backdrop of the recognition that, in South Caucasus, inequality in opportunity
resonates more than inequality in outcomes, this paper joins a new direction in recent research
focusing on inequality in opportunity (Barros et al. 2009; Ferreira and Gignoux 2011). According
to one line of argument in this literature, the source of inequality may be divided into two
elements: first, differences in effort, choice, and talents (a good sort of inequality) and, second,
predetermined circumstances such as ethnicity or gender (a bad type of inequality). Thus,
inequality of opportunity denotes the extent to which inequality in outcomes can be attributed to
4
circumstances over which individuals have no control. Addressing inequality of opportunity—
the bad inequality—therefore has universal appeal regardless of any differences in political
spectrum. Early empirical findings also show that inequality in opportunity has negative effects
on income (Molina, Narayan, and Saavedra-Chanduví 2013).
This paper focuses on inequality of opportunity, particularly in labor markets, based on the idea
that the ability of individuals to access labor markets and obtain jobs based on their skills and
work experience (human capital), irrespective of their circumstances, is critical to economic
mobility and the reduction of inequality. In the context of transition economies, the effect of
market reform depended on whether the countries were able to reorganize the labor market into
more efficient and fair labor allocation processes. The discussion in this paper is structured
around the main concept that fairness in processes is determined by the underlying equality or
inequality of opportunity in society. Specifically, the paper examines, first, the extent to which
fairness—the role of skills and effort versus connections in obtaining job—is perceived to exist in
such processes by people in South Caucasus; second, whether access to good jobs is determined
by effort and choice or by circumstances independent of the control of individuals; and, third, the
extent to which inequality in the access to basic human capital inputs affects inequality in
obtaining good jobs.
The study finds that inequality of opportunity in labor markets is a serious issue among the
countries of South Caucasus. The access to good jobs in South Caucasus is not only limited
relative to other countries in Eastern Europe and Central Asia, but also circumstances beyond the
efforts individuals may undertake—the unfair element—have a substantial impact on inequality
in the access to good jobs. The paper also highlights that some parts of the unobserved inequality
in labor markets appear to be associated with inequalities in gaining access to education and basic
services during the formative stages of life.
2. Methodology
The data used in this paper are taken from the 2015–16 round of the Life in Transition Survey
(2015 LiTS) carried out by the European Bank for Reconstruction and Development in
collaboration with the World Bank. This round of the survey was conducted in 34 countries,
mainly in Eastern Europe and Central Asia, but also including Cyprus, Germany, Greece, Italy,
and Turkey, and covered around 51,000 households. This paper utilizes the household survey
not only because of the rich questions on the socioeconomic background and labor market status
of respondents, but also because of the extensive information on attitudinal perspectives and
perceptions of the social, political, and economic situation in the countries.
The first part of the analysis is devoted to identifying how people perceive fairness in processes.
A simple gauge of fairness in processes is the perceived role of hard work, effort, and skill as a
means to achieve one’s goals in life and to gain access to success in life. This concept of fairness
is contrasted with the elements of processes or social arrangements that may unfairly help
5
individuals achieve better lives. The paper shows how the population of a country may perceive
the balance between fair factors and unfair factors in accounting for the success of individuals.
A proxy for unfairness in processes is the perceived role of connections. Perceived fairness is
therefore reflected in the extent to which people believe that hard work and effort are more
important than connections in realizing success in life. In practice, connections are defined as,
first, political connections, whereby people can use political power to influence decisions. The
study involved observing whether people perceive such connections as generally helpful in
gaining success in life. Second, another, more general sort of connection, not necessarily
supported by political power, is associated with people who hold special functions or positions
within a community and who may be asked by community members for help in influencing
decision making. In particular, the paper examines the extent to which people believe this type
of connections can help secure access to government or private sector jobs. Acemoglu et al. (2016)
and Fisman (2001) have documented the role of political connections in providing people with
valuable noncompetitive market advantages. While there is a possibility of an endogeneity
problem associated with the reinforcing causality between belief in the role of connections and
the efficacy of connections in providing access to jobs, this paper does not parametrically estimate
the effect of connections on specific economic outcomes or apply a specific identification strategy
to deal with the endogeneity and omitted variable bias problems. Instead, it simply offers a
descriptive analysis of people’s perceptions of the role of connections in securing jobs to generate
preliminary ideas on popular sentiments about fairness in processes in South Caucasus.
The more substantial analysis is in the second part of the paper where the fairness in processes is
evaluated by measuring the inequality in opportunity in obtaining good jobs in the labor market.
The main analytical tool is the human opportunity index (HOI) framework developed by Barros,
Molinas Vega, and Saavedra-Chanduví (2010) and Barros et al. (2009), which has been adjusted
so it is more applicable to labor markets (box 1). The HOI for children in the original framework
is relabeled here as inequality-adjusted coverage in the labor market because, besides
circumstances beyond the control of individuals, the analysis explicitly recognizes the role of
effort and choice in obtaining good jobs.
Box 1. The HOI Methodology
The HOI has been developed as part of the World Bank initiative to measure the equitable
provision of opportunities among children. Since its introduction, the HOI methodology has
been used widely in the literature to analyze the inequality of opportunity among children.
(See, for example, Dabalen et al. [2015] for applications in Africa and Krishnan et al. [2016] for
applications to the Middle East and North Africa.) The key premise of equality of opportunity
among children is that basic services providing critical human capital development inputs,
such as quality education, good health care, or water and sanitation, should be available to all
children, irrespective of their birth circumstances, including gender, urban-rural residence,
parental wealth, and so on. Thus, equality of opportunity means that the playing field should
be level, and basic opportunities should be independent of initial circumstances. The HOI
methodology is one tool to measure the extent to which the reality deviates from this ideal.
6
In general, the HOI for a given opportunity—for example, access to quality education—is a
single index that captures both the opportunity’s universality (the share of children who enjoy
the opportunity) and any inequality in access (variations according to circumstances in access
among children to the opportunity). The penalty factor arises if the inequality in access is
calculated based on an index of dissimilarity (D-index), which equals zero if the access to
opportunity is independent of the circumstances of respondents. The underlying purpose of
constructing the HOI is to generate a scaled measure that rises as opportunity increases but
falls as inequality becomes wider in the coverage among groups characterized by differences
in circumstances. Once an index score has been obtained, a Shapley decomposition procedure
is applied to apportion the inequality across various circumstances.a Although causality cannot
be ascribed through the Shapley decomposition; quantitative statements may be made, such as
that a certain percent of the inequality in opportunity, for example, access to school, is
associated with children’s circumstances, such as gender, birth location, and so on.
More formally, the inequality-adjusted coverage rate, , is defined as follows:
̅ 1 ,
where ̅ represents the coverage rates of access to good jobs, and is the dissimilarity index
(D-index), calculated as follows:
̅
∑ | ̅ |,
where is a type of circumstances-group; is the coverage rate of group ; is the share of
group in the total labor force; and is the number of circumstances-groups.
a. The Shapley decomposition in Shorrocks (2013) is a method to overcome the problem that a change in a
dissimilarity scalar measure because of the addition of a circumstance depends on the initial set of circumstances
that are changed. In this procedure, intuitively, the effect of a circumstance is calculated as the average value of all
changes that occur if the circumstance is added to all possible subsets of initial circumstances.
This paper adopts the HOI methodology to measure the inequality of opportunity in obtaining a
good job in the labor market and labels it as inequality-adjusted coverage. It adopts this
terminology to accommodate two important considerations in interpreting inequality of
opportunity measures. First, it is necessary to shed the normative baggage of the notion of
opportunity in a labor market. In the original setting of the HOI among children, defining
opportunity as children’s access to education is untainted by the effect of the effort and agency of
individuals and thus has a universal appeal: every child should have access to quality education.
But any labor market outcome is a function of accumulated opportunities, plus what individuals
have done with the opportunities through effort and choices. The issue then becomes whether
everyone is entitled to employment and jobs in the same manner. Perhaps some people with a
certain desired, necessary, and appropriate level of education, skills, and temperament do
deserve some set of jobs, but not everyone. For this reason, this paper adopts the HOI
measurement tool sans the normative baggage. Second, the analysis maintains a sharp focus on
7
good jobs. Access to any kind of job is not always the most desirable state for everyone. People
who are working and who are observed as participating in employment because they have no
other choice are distinctly less well off than people who are not working and are thereby observed
as not participating in employment precisely because of their high reservation wage or because
they are seeking better outside options. To minimize this measurement error, the analysis focuses
on good jobs.
However, defining selected types of employment as good jobs is not a straightforward exercise.
Purely from the perspective of the development payoff, the definition of a good job varies by
context. For example, the World Bank (2012) finds that, in agrarian societies, a job may be more
productive if it is in smallholder farming or involves urban employment that is well connected
with global markets. In countries with high youth unemployment, it may be a job that is not
supported by rents or not allocated based on connections. In aging societies, it may be a job that
keeps the skilled active at older ages. From the perspective of the individual, a good job may have
desirable monetary and nonmonetary attributes such as good earnings, benefit stability, social
prestige, and dignity. But a good job is also associated with positive spillovers in society. The
World Bank (2012) addresses several potential mechanisms through which a good job may
promote a better society. Thus, jobs that are filled by women may empower women’s position in
society by rebalancing intrahousehold resource distribution and enhancing the role of women in
decision making, both of which benefit children. Jobs integrated in world markets may also
generate knowledge spillovers and help firms realize increasing returns to scale. Jobs that reflect
a sense of fairness, especially in fragile and conflict-affected societies, may help maintain social
cohesion. Meanwhile, the International Labour Organization (ILO 1999, 3) describes decent work
as “opportunities for women and men to obtain decent and productive work in conditions of
freedom, equity, security and human dignity.” In a slightly expanded version, the International
Labour Organization defines decent work as “work that is productive and delivers a fair income,
security in the workplace and social protection for families, better prospects for personal
development and social integration, freedom for people to express their concerns, organize, and
participate in decisions that affect their lives, and equality of opportunity and treatment for all
women and men.”3 Decent work is therefore instrumental in reducing poverty and achieving
equitable, inclusive, and sustainable development.
This paper relies on three desirable characteristics to define a good job, as follows: (1) a job that
allows the jobholder to work 20 or more hours per week, (2) salaried work through a contractual
arrangement, and (3) salaried work that provides some measure of tenure. The assumption
associated with the first criterion is that everyone would like to be employed full time if given the
opportunity. This is a strong assumption because there may be specific instances in which people
prefer to remain in part-time work or temporarily unemployed. However, the data on the types
of jobs held by individuals do not indicate whether some people may not be working because
they are waiting for better jobs or they may be staying at a job simply because there is no
alternative. The study classifies such people as participants in the labor force who do not have
appropriate jobs. The analysis also assumes that most people prefer jobs providing salaries under
3
See “Decent Work,” International Labour Organization, Geneva, http://www.ilo.org/global/topics/decent-
work/lang--en/index.htm.
8
contracts and on a full-time, permanent basis. While the self-employed may have higher incomes
or better arrangements, the data do not permit the identification of the contractual or tenure status
of the self-employed. Arguably, jobs in the formal waged sector in labor markets characterized
by a high degree of informality are generally better jobs.
Figure 3 is a schematic description of the sample categories and the definition of a good job in the
employment module of the 2015 LiTS. This paper defines individuals as participants in the labor
force or within the labor market if they are between 18 and 64 years old and were working during
the 12 months previous to the survey or, if not working during the previous 12 months, were
discouraged in seeking work, were actively looking for work, or were waiting to learn if they had
been accepted for a job. Individuals are considered to hold a job if they report they were working
during the 12 months prior to the survey, at least one hour in the seven days prior to the survey,
and for more than at least one hour in a typical week. Individuals are considered to have lower-
or higher-quality jobs according to whether they were working for more than 20 hours a week,
in wage employment for 20 or more hours a week with a contract, or working in wage
employment for 20 or more hours a week permanently or with tenure.
Figure 3: Schematic definition of good jobs
In the next step, the analysis identifies and separates out cases in which the efforts or choice of
individuals (labeled behavior or characteristics) are important and cases in which circumstances
that may be beyond the control of individuals may drive inequalities in gaining access to good
jobs. The proxy for the characteristics of individuals is educational attainment and work
experience. To determine circumstances, the analysis investigates variations by gender, parental
educational attainment, parental political affiliation, and ethnicity. Based on the 2015 LiTS, the
educational attainment of individuals and their parents are divided into seven categories, from
9
no education to a postgraduate degree. For working experience, the ages of respondents are used
as a proxy on the assumption that older workers generally have more work experience.
Parental educational attainment is defined as the educational attainment of the father or mother,
whichever is greater, and this is also an aspect of the socioeconomic background of individuals.
Parental political affiliation is identified according to the responses to questions about whether
the parents of respondents were members of the Communist Party, which is particularly relevant
because Armenia, Azerbaijan, and Georgia were part of the Soviet Union, and this is also used as
an indicator of social status.4 There is no direct measure of the ethnicity of individuals; so the
language spoken by a respondent is taken as a cue, especially depending on the country of
residence of the respondent.5
Figure 4 illustrates the conceptual framework. An assumption supporting the framework is that
educational attainment—an important marker of skills and experience—should be a major
determinant of the ability to obtain a good job in the labor market. Therefore, a labor market that
places a premium on education and experience can be regarded as a fair labor market. In contrast,
a labor market that allocates opportunities based on the gender, ethnicity, paternal educational
attainment, or parental political affiliation is deemed an unfair labor market, although, in many
cases, this is not due to discrimination, but the unequal effect of an incomplete transition to
equitable access to opportunities. The World Bank (2015) finds that, in Armenia and Georgia,
education helps in gaining access to jobs, although the unemployment rate among the well-
educated is relatively high in comparison with the average in the Organisation for Economic Co-
operation and Development (OECD), and a significant share of workers (29 percent in Armenia
and 33 percent in Georgia) consider themselves overeducated for their current jobs. The World
Bank study also shows that, while skills matter more than educational attainment in explaining
the variance in hourly wages among youth (ages 15–29), the role of education in explaining wage
variance is higher among prime workers (ages 30–44).
4
Parental affiliation with the Communist Party can be considered an unfair advantage in the labor markets of former
communist countries especially among people in older birth cohorts who entered the labor market during the
communist era. Parental affiliation with the Communist Party might also affect labor market access negatively, an
unfair disadvantage, among people who entered the labor market in the first years after the collapse of the Soviet
Union. Yet, the analysis here is aligned with the view that social status is a durable institution that persists in the
aftermath of a political crisis, especially if the crisis leads to more labor market imperfections because of the destruction
of information networks. Another interpretation is that parental affiliation is a signal of motivation, especially in a
situation in which the only opportunity for upward social movement is associated with Communist Party membership.
5
This paper reflects an acknowledgment that there is a possibility that ethnicity does not play a major role in accessing
good jobs because of the effect of ethnic cleansing in the 1990s that effectively eliminated minorities.
10
Figure 4: Dimensions of inequality of opportunity in labor market
The next step in the analysis involves a longer, life-cycle view aimed at assessing the extent to
which inequalities in the earlier, formative stages of life may play a role in inequality of
opportunity in a labor market. It extends the conceptual framework by measuring inequalities of
opportunity during the formative years based on the notion that inequality in access to good jobs
may be even more unfair if one considers the indirect effects of circumstances on educational
attainment and other human capital inputs earlier in life or during childhood (figure 5). Although
the inequality of opportunity in labor markets that is attributable to an individual’s educational
attainment may be regarded as fair, not everyone enjoys equal access to good-quality education
during the formative years. Thus, the effects of inequality may also be evident earlier in life. The
differences in the access of individuals to education and human capital inputs at the moment of
entry into the labor market may reflect these unobserved and unfair sources of inequality.
Figure 5: Dimensions of inequality of opportunity in labor market,
extended framework
11
The scores achieved in the tests of the Program for International Student Assessment (PISA) of
the OECD in Georgia and Azerbaijan in 2009 and 2015 are used to estimate the level of access to
quality education. Gathered in 2015, data of the Integrated Living Conditions Survey in Armenia,
the Monitoring Survey for Social Welfare in Azerbaijan, and the Monitoring of Households
Survey in Georgia are used to estimate human capital inputs. The standard HOI among children
is applied to examine the dimensions along which the inequality of opportunity in access to basic
human capital inputs during the formative years may be most salient. Specifically, in this
framework, opportunities are defined as access to running water and sanitation among children
ages 0–16 and reflect the quality of health services in the early years of life.
The circumstances that may affect the access of children to running water and sanitation, are
measured with the following variables: the number of 0- to 15-year-olds in the household, the
educational attainment of the household head, the gender of the household head, the household
consumption quintile, the gender of the child, and the location of the household (urban or rural
area and province or region). Unlike the HOI framework for determining the access to a good job,
all factors contributing to inequality, that is, the circumstances, are independent of the choice of
children. Without a panel data structure, the HOI on human capital can only be generated for the
current young generation, which is not directly relatable to the current labor force on which the
HOI estimation of good jobs is focused.
3. Results
The empirical findings of the study are organized into three topics: the perceived fairness of the
process, inequality of opportunity in the labor market, and inequality in human capital inputs
(see the Methodology section). In this section, how people in South Caucasus generally perceive
fairness in the processes involved in obtaining access to good jobs is described by way of
perceptions on the role of connections. The next and more substantial part of the section
highlights the level of access to good jobs in South Caucasus, measures the inequality of access to
good jobs, and uses the HOI framework to decompose the sources of inequality and to determine
the level of unfair inequality in accessing good jobs. The last part of the section presents the results
produced by applying the HOI framework to determine the amount of the access of children to
basic human capital inputs and provide some early indication on the extent to which inequality
in access to good jobs is associated with inequality in access to human capital inputs.
3.1. Perceptions of the fairness of processes
The analysis generally finds that the populations of the countries of South Caucasus perceive
that connections play a crucial role in gaining access to good jobs. Among Eastern Europe and
Central Asia countries, the share of the population that believes connections are essential or
important in obtaining good jobs in both the private sector and the public sector is the largest in
12
Armenia. This represents an indication of the potential problem of unfairness in labor market
processes in the countries of South Caucasus.
In the broadest sense of opportunity defined as the ability to get ahead or achieve success in life,
a larger share of the population of South Caucasus still believe that an individual’s hard work
and effort play a greater role than political connections. Only 22 percent of Armenians, 11 percent
of Azerbaijanis, and 14 percent of Georgians believe political connections are the most important
vehicle for getting ahead (figure 6, panel a). In contrast, 64 percent of Armenians, 75 percent of
Azerbaijanis, and 78 percent of Georgians regard hard work, effort, skills, and intelligence as the
most important factors in achieving success. The numbers suggest that people in South Caucasus
consider society fair in providing opportunities in terms of upward mobility or achieving success
in life (figure 6, panel b).
Figure 6: Perceived role of connections and efforts in achieving success
in life, 2015
6.a Perceived role of political connections 6.b Perceived role of effort and skills in
in getting ahead in life achieving success
However, the perceived fairness in access to opportunities for good jobs on the labor market is
mixed among the people of South Caucasus. On the basis of a slightly expanded definition of
connections to include the support of people who have been asked to exert influence to obtain a
favorable decision about securing a good job, Armenia stands out with 83 percent of the
population rates connections as essential or important in acquiring a good government job, the
highest share among 34 countries surveyed (figure 7, panel a).6 In Azerbaijan and Georgia, the
6
In the 2015 LiTS, respondents rated the importance of connections from not important to essential. The ratings were
1 = not important, 2 = somewhat important, 3 = moderately important, 4 = very important, and 5 = essential. It was
also possible to respond “don’t know.”
13
shares were well below or close to the average among the 34 countries.7 A similar pattern also
emerges on the perceived role of connections in obtaining a good job in the private sector,
although the importance of connections is generally lower in this instance than in obtaining
government jobs (figure 7, panel b). These more typically negative perceptions on the role of
connections serve as an approximate indicator of inequality in the labor market in the region.
Figure 7: Perceived role of connections in getting job, 2015
7a: Perceived role of connections in 7b: Perceived role of connections in
getting government job getting private job
The data on the 34 countries surveyed in the 2015 LiTS show that the demand for redistribution—
defined as the share of a population believing that the gap between the rich and the poor needs
to be reduced—is positively correlated with the perceived importance of connections in obtaining
jobs. In contrast, it appears there is no correlation between perceived fairness in achieving success
in life and the demand for redistribution (figure 8). These correlations suggest that perceptions
about mechanisms—defined as connections that determine success in gaining access to good
jobs—are more strongly correlated with the preference for redistribution. The prevalence of
reliance on connections is not necessarily a pernicious symptom of an ailing economy because
connections may be important in compensating for imperfect or distorted labor markets that lack
formal means of intermediation. Nonetheless, not everyone possesses the same access to
connections regardless of the possible positive role of connections in an economy. The fact that
the correlation between the demand for redistribution and the role of connections in obtaining
jobs is more evident relative to more general perceptions about the role of political connections
in achieving success in life suggests that, in the countries surveyed, the demand for redistribution,
as an indicator of perceived unfairness, is sensitive to perceived opportunities to participate in
the labor market.
7
In interpreting the responses in Azerbaijan, the fact that, among the countries covered in the 2015 LiTS, Azerbaijan
exhibits one of the largest population shares responding “don’t know” or only “moderately important” should be taken
into account. Thus, in this case, the results in Azerbaijan are probably underestimates.
14
Figure 8: Demand for redistribution and perceived fairness in access to
private sector job, 2015
3.2. Inequality in the labor market
This subsection extends the analysis beyond approximate measures of perceived unfairness in
the labor market to a more systematic examination of this area of inequality by implementing the
HOI framework. It starts with coverage rates, that is, the percent share of the labor force that has
access to good jobs, which is the simplest measure of beneficial access to the labor market. It then
presents the results of the application of the inequality index (the D-index), which is a measure
of between-group inequality among those people who obtain the access. This index is the basis
for adjusting the simple coverage rates by a measure of the inequality in access to generate an
augmented indicator of access to good jobs that is labeled the inequality-adjusted coverage rate.
The subsection presents the results of the decomposition of contributing factors in inequality and
assesses the extent to which the unfair component of circumstances contributes to inequality in
the access to good jobs. The study concludes that not only do people in South Caucasus have
15
relatively little and highly unequal access to good jobs, but also that this sort of inequality makes
a substantial contribution to the total inequality of opportunity characterized as unfair.
Coverage rates
Gauged by coverage rates, the labor force in South Caucasus exhibits low access to good jobs
relative to the 34 countries surveyed in the 2015 LiTS (figure 9). Based on the paper’s simplest
definition of a good job, the share of the labor force working 20 or more hours a week in South
Caucasus is small in comparison with the regional average across Eastern Europe and Central
Asia (61 percent). Only 26 percent of the labor force in Azerbaijan report they are working at least
20 hours a week. The shares are higher in Armenia (42 percent) and Georgia (41 percent), but still
well below the regional average. Indeed, the countries of South Caucasus, together with Kosovo,
the Kyrgyz Republic, Tajikistan, and Turkey, are among the countries in the region with the
lowest population shares working 20 hours or more a week.
Figure 9: Share of labor force population working in a good job
9a. Working 20 and more hours a 9b. Working 20 and more hours a
week week and with contract
9c. Working 20 and more hours a
week and with security of tenure
16
In the countries of South Caucasus, access is also much lower to salaried jobs of 20 hours or more
a week with contracts, a type of job that is more likely to be in the formal sector and to be
associated with benefits. Only 22 percent of the labor force in Armenia and Georgia and 11
percent of the labor force in Azerbaijan hold this type of job. These coverage rates are distinctly
lower than the regional average across Eastern Europe and Central Asia (45 percent). The same
pattern of low coverage emerges in the countries of South Caucasus in the case of jobs requiring
20 hours or more of work a week that are associated with tenure security and permanent
contracts. In Armenia, 27 percent of the labor force holds tenured jobs of 20 hours or more a
week, while, in Azerbaijan and Georgia, the shares are 20 percent and 13 percent, respectively.
The average coverage rate is 45 percent across the countries surveyed in the 2015 LiTS.
The coverage rates are measured using data from the 2015 LiTS. How do the coverage rates
compare with the results of the National Labor Force Survey? Applying similar age-group
classifications, the coverage rates for each definition of good jobs calculated using the LiTS are
close to the figures using the National Labor Force Survey in Armenia. However, the rates are
substantially different in Georgia, most probably because of the data limitation. Instead of using
the standard Labor Force Survey, the comparable coverage rates are calculated from the
Monitoring Household Survey, which is not specifically designed as a labor force survey.
Meanwhile, the Azerbaijan Labor Force Survey is not currently available. (See annex 1 for the
complete comparison of coverage rates and the accompanying notes.)
Inequality index and inequality-adjusted coverage rates
While coverage rates provide a simple measure of the universality of access to good jobs,
evaluating inequality among those people who gain access to good jobs is also important. Ideally,
access to good jobs should be independent of the circumstances over which workers have no
control. Following the HOI framework, this paper uses the D-index to measure between-group
inequality differentiated by individual effort and choice and by circumstances in accessing jobs.
It finds substantial inequality in the coverage among groups in South Caucasus that are
17
characterized by different circumstances. Thus, coverage rates among each circumstance-group
(age, educational attainment, parental educational attainment, parental political affiliation,
gender, and ethnicity) differ appreciably from the overall coverage rate of a country, indicating
that one’s position in circumstance-groups matters in gaining access to good jobs. Indeed, there
are sizable differences in access across population subgroups in South Caucasus.
The inequality index is remarkably high in Azerbaijan and, in Armenia and Georgia, well above
the average across the countries surveyed in the 2015 LiTS (figure 10). Within the possible range
of 0 to 1 on access to good jobs involving work for 20 hours or more a week, the D-index is 0.39
in Azerbaijan, the highest score among countries surveyed in the 2015 LiTS. The scores of
Armenia and Georgia are lower, at 0.18 and 0.16, yet these scores are still higher than the average
of 0.12 across surveyed countries. Similar patterns emerge in jobs involving 20 hours or more a
week under contracts and 20 hours or more a week with tenure, although, in the case of good jobs
with contracts, Azerbaijan has the third-highest inequality score after Kosovo and Tajikistan
among the countries surveyed. Therefore, the countries of the South Caucasus belong to a group
of countries with high between-group inequality in access to good jobs (according to effort and
according to circumstances). In addition, the stricter the definition of a good job, the greater the
inequality except in Azerbaijan, where, across all categories of good jobs, the inequality index is
highest in the case of the least strict category of good jobs, that is, jobs involving work for 20 hours
a week or more.
Figure 10: The inequality index (D-index)
10a. Working 20 and more hours a 10b. Working 20 and more hours a
week week and with contract
18
10c. Working 20 and more hours a
week and with security of tenure
The conversion of the inequality index into an effective penalty factor could be applied directly
to adjust the coverage rate. Such a penalty factor would be constructed by interacting the
inequality index with coverage rates for good jobs (the D-index, multiplied by the coverage rate).
Any differences between the coverage rate and the inequality-adjusted coverage rate—the
penalty factor—would arise from between-group inequality, illustrating the influence of
inequality on obtaining good jobs. Indeed, in Eastern Europe and Central Asia, Azerbaijan would
be associated with the highest between-group inequality penalty factor (10 percentage points) in
acquiring a good job involving work for 20 hours or more a week (figure 11). The penalty factors
for this type of good job in Armenia and Georgia would be 7.6 and 6.6 percentage points,
respectively, while the average in the 2015 LiTS countries would be 6.6 percentage points. The
high penalty factors in the countries of South Caucasus is remarkable because it implies that the
inequality in access overwhelmingly offsets the low coverage rates in these countries. The penalty
factors associated with good jobs involving work for 20 hours or more a week with contracts or
with tenure are lower than the one involving only work for 20 hours or more a week (figure 11).
In these stricter categories of good jobs, the penalty factors among the countries of South
Caucasus are generally below the average among the 34 countries surveyed in the 2015 LiTS,
which is as expected given the low coverage rates in the former.
19
Figure 11: The penalty factor (D-index times coverage rate)
11a. Working 20 and more hours a 11b. Working 20 and more hours a
week week and with contract
11c. Working 20 and more hours a
week and with security of tenure
Subtracting the penalty factor from the coverage rates gives the inequality-adjusted coverage rate,
an indicator of the access to good jobs that takes into account the inequality in access among
people who have the good jobs. As expected, the inequality-adjusted coverage rates are lower
than the simple coverage rates (see figure 9). For example, the inequality-adjusted coverage rate
among people working 20 hours or more a week is 16 percent in Azerbaijan, 34 percent in
Armenia, and 35 percent in Georgia. The coverage rates in these countries are 26 percent, 42
percent, and 41 percent, respectively. The relatively high between-group inequality in South
Caucasus does not help in improving the rank of the countries in the subregion in inequality-
adjusted coverage. In fact, the relative ranking of these countries in inequality-adjusted coverage
rates remains generally the same as the ranking on the coverage rate indicator, except for
20
Azerbaijan, where the inequality index is so high that it leads to a significant discount even on
the low coverage rate (see figure 9). Similar to the situation in the coverage rate indicator, these
countries belong to the group with low inequality-adjusted coverage rates as well.
Inequality index decomposition
This paper finds that countries in South Caucasus have low coverage rates for good jobs and high
inequality among population subgroups in the access to good jobs. The analysis now addresses
the extent to which differences in individual circumstances beyond personal effort and choice
(the unfair elements) contribute to the wide inequality in the access to good jobs. First, following
the Shapley decomposition method outlined in Shorrocks (2013), between-group inequality in
labor markets (D-index) is decomposed into two major contributing components: i) the
behavioral characteristics of individuals related to their efforts and experience and, ii) the
circumstances beyond the control of individuals, namely, gender, ethnicity, parental educational
attainment, and parental political affiliation. The first component (broadly, effort and
experience) is considered the fair element contributing to inequality in a labor market, while the
second component (circumstances) is the unfair element. To assess the relative fairness of labor
market inequality in the countries of South Caucasus, the shares of the two contributing factors
in inequality in these countries are compared with the average shares in all countries surveyed
in the 2015 LiTS, as well as with the averages in Belarus, Central Asia (Kazakhstan, the Kyrgyz
Republic, Mongolia, Tajikistan, and Uzbekistan), Moldova, and Ukraine; Southeastern Europe
(Albania, Bosnia-Herzegovina, Cyprus, Greece, Kosovo, FYR Macedonia, Montenegro, and
Serbia); the European Union–11 (Bulgaria, Croatia, Czech Republic, Estonia, Hungary, Latvia,
Lithuania, Poland, Romania, Slovak Republic, and Slovenia); the European Union (represented
by only Germany and Italy), the Russian Federation, and Turkey.
An important finding of this paper is that a sizable share of the inequality in access to good jobs
is unfair in Armenia and Azerbaijan, while the share is moderate in Georgia (figure 12). To
illustrate, on the access to the first category of good jobs (those involving work of 20 hours or
more a week), the share of the unfair element of inequality is small in Georgia, at 34 percent, but
it is not small for Armenia and Azerbaijan. At 63 percent and 85 percent of total between-group
inequality, respectively, the countries show the highest level of unfairness in the 34 countries of
Eastern Europe and Central Asia, where the average share of the circumstances component in
total inequality is 47 percent. The shares of circumstances in total inequality in Azerbaijan and
Armenia are also higher than the averages in other sub-regions. Similar patterns in the share of
the unfair element in inequality are observed in the access to good jobs involving work of 20
hours or more a week with contract and, especially, with tenure. In the category of good jobs
involving work of 20 hours or more a week with contracts, Georgia registered the smallest share
of the unfair element in between-group inequality.
21
Figure 12: Decomposition of inequality in access to good job – working 20
hours or more per week
12a. Working 20 and more hours a 12b. Working 20 and more hours a
week week and with contract
12c. Working 20 and more hours a
week and with security of tenure
Across different categories of good jobs, the unfair elements—the circumstances—matter less in
accessing good jobs involving work of 20 hours or more a week with contracts. Yet, the notion
of less unfairness in obtaining a better good job is difficult to establish because the share of unfair
elements in the strictest category of good jobs—the ones with tenure or permanent status—is not
substantially different from the share in the least strict category, jobs involving work of 20 hours
or more a week (see figure 12, panels a and c; table 1). Neither labor market discrimination or
market imperfection seems to diminish clearly with the quality of the job (such as jobs with better
contracts or better tenure security) in South Caucasus; the diminishing role of the unfair element
22
in inequality among better secured good jobs is evident in Armenia, but not in Azerbaijan or
Georgia.
Table 1: Contribution of effort and hard work and circumstances to total
inequality in access to good jobs (%)
Total unfair
Efforts and hard work Circumstances element
Parent's
Father's political
Age Education Gender education affiliation Ethnicity
Type of good job* 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3 1 2 3
Armenia 9 6 5 27 44 44 46 11 17 11 22 18 5 14 15 1 2 1 63 49 51
Azerbaijan 1 0 1 15 44 18 68 38 59 13 13 17 2 2 3 2 2 3 85 55 82
Georgia 21 1 5 45 74 55 9 0 5 20 17 29 1 2 1 4 5 5 34 24 40
.
BMU & Central
Asia 17 17 14 33 41 44 23 6 8 12 18 17 5 8 6 11 10 12 50 42 42
S.E. Europe 19 12 17 34 43 39 20 11 13 16 18 16 3 3 2 8 13 13 47 45 43
EU-11 21 16 22 38 40 40 7 8 5 18 21 18 4 4 4 11 11 10 41 44 38
GER, ITA, RUS,
TUR 24 24 28 27 28 21 22 14 22 18 19 16 1 3 1 9 14 13 50 49 51
Note: *Type 1 is working 20 hours or more a week; type 2, working 20 hours or more a week with contract; type 3 working 20
hours or more a week with tenure
Nonetheless, the share composition of the various factors contributing to the unfair element in
the inequality in access to good jobs in South Caucasus varies across categories of good jobs and
across countries. The gender factor plays a sizable role in Armenia and Azerbaijan, while
Georgia is among the countries with the smallest share of the gender factor in the inequality in
the labor market. Thus, 68 percent of the inequality in access to jobs involving work of 20 hours
or more a week in Azerbaijan derives from differences in gender, while this is true of a smaller,
but still considerable share of jobs of 20 hours or more a week with contracts or tenure (see table
1). Armenia shows a slightly different pattern: in jobs of 20 hours or more a week, the share of
the gender factor in inequality is 46 percent, but this declines significantly, to 11 percent and 17
percent, respectively, in jobs of 20 hours or more a week with contracts and tenure.
Parental educational attainment, one of the proxies for family social status, plays a considerable
role in inequality, especially in Georgia. The shares of the differences in parental educational
attainment in inequality in Georgia are 20 percent for jobs of 20 or more hours a week, 17 percent
for jobs of 20 or more hours a week with contracts, and 29 percent for jobs of 20 or more hours a
week with tenure. The contribution of parental educational attainment to labor market
inequality in Armenia and Azerbaijan is between 11 and 22 percentage points depending on the
category of good jobs. Meanwhile, parental political affiliation is a relatively important
23
contributor to inequality in Armenia, especially in the access to good jobs with contracts (14
percent) or tenure (15 percent). Parental political affiliation and the ethnicity of an individual
seem to have only a small role in inequality in the access to all categories of good jobs in the three
countries in South Caucasus.
Among the fair components of inequality, education generally matters a great deal, especially in
securing a good job with a contract or with permanent tenure (see table 1). In the case of the basic
definition of a good job, 20 hours or more a week, the contribution of educational attainment to
inequality in Armenia and Azerbaijan (27 percent and 15 percent, respectively) is limited relative
to the gender differential factor (46 percent and 68 percent, respectively). In the stricter category
of good jobs, the ones associated with contracts or tenure, the contribution of education generally
increases. Yet, it appears that, in South Caucasus, differences in educational attainment
contribute less to inequality in the access to good jobs with permanent tenure than to inequality
in securing jobs with contracts. Another fair component of inequality, age (as a proxy for work
experience), generally contributes less to inequality, except in Georgia in the access to jobs
involving work of 20 hours or more a week (21 percent). The educational attainment of both
respondents and parents appear to contribute substantially to inequality in the labor market
according to the HOI approach. To check the robustness of this finding, the analysis included a
probit regression run on access to good jobs with similar relevant variables (effort and
circumstances), which found that the education variables of respondents and their parents are
equally significant in driving the access to good jobs (see annex 2).
The research makes two assumptions about the fair component of inequality. First, work
experience is important in obtaining access to good jobs. However, among transition economies
such as those in South Caucasus, the massive disruption and exogenous economic restructuring
in the 1990s may have rendered the role of work experience in gaining access to good jobs
meaningless. For example, when Armenia became independent, the share of the manufacturing
sector in total output shrank from 70 percent to 20 percent; so, skills and work experience in
manufacturing suddenly became less relevant. To address this issue, the research also tests the
assumption that work experience remains relevant in the region. In particular, this study
observes the variations in the contributions of age as a proxy for the value of work experience in
gaining access to good jobs in industries in which experience is still important after the massive
restructuring (manufacturing) or in sectors characterized by human capital intensity, such as
high-quality services (finance and public administration). The research also compares the
contributions of age in accessing good jobs among younger cohorts among which work
experience may matter more (see annex 3). The evidence shows that the age of respondents as a
proxy for work experience does not contribute more to inequality in manufacturing and high-
quality services than in other industries. It also does not exhibit a greater share in inequality
among younger cohorts.
Second, the research assumes that educational attainment is a valid measure of skills. However,
many studies in the countries of South Caucasus point to a significant skills mismatch because of
poor teaching quality and outdated curricula that often do not allow for a smooth school-work
transition. To confront this problem, the research also tests if educational attainment among
individuals is correlated with work skills by observing differences in the contributions of
24
education to access to good jobs across industries showing skills mismatches. If there were a
skills-mismatch problem, industries in which the skills-mismatch is more severe, such as
agriculture and low-quality services, would exhibit relatively greater contributions of educational
attainment to labor market inequality, that is, workers would tend to be overeducated relative to
the skills required by their jobs in these industries. The research finds that educational attainment
among respondents, as a proxy for skills, does not have a greater effect on inequality in the access
to good jobs in industries with more severe skills mismatches (see annex 4, table A4;).
3.3. Inequality in human capital inputs
The observed fair inequality component of the access to good jobs, that is, the component
involving the effort, experience, and choice of individuals, may be affected by inequality in the
access to human capital inputs among individuals during their formative years. The research
extends the analysis by evaluating learning performance in Azerbaijan and Georgia based on the
PISA test scores to describe access to education. The analysis also relies on the application of the
HOI approach to the measurement of inequality in children’s access to basic utilities in South
Caucasus. The study finds that, first, educational performance appears relatively poor and
unequal depending on the life circumstances of children. Second, the coverage rates of basic
human capital inputs such as access to gas and water and sanitation are generally high, although
inequalities remain an issue in the countries, especially in higher-quality services. The inequalities
are mostly associated with geographical disparities.
Learning performance
While school attendance rates are high in the countries of South Caucasus, learning performance
shows mixed results.8 Only Azerbaijan and Georgia participate in the PISA test; the test was
conducted in Azerbaijan in 2006 and 2009 and in Georgia in 2009 and 2015. The average PISA
scores rose noticeably in Georgia between 2009 and 2015: by 38 points in science, 25 points in
mathematics, and 27 points in reading (World Bank 2016) (figure 13). Because every 30-point
difference in a PISA score corresponds to about one year of schooling in the 2015 round of the
PISA, the improvement in Georgia is substantial. Meanwhile, in the two rounds of the PISA test
in 2006 and 2009, the performance in Azerbaijan was less impressive. The country’s average
score in reading rose by 9 points, but the mathematics score dropped rather significantly, by 45
points. Overall, these levels of educational achievement in the countries of South Caucasus are
still far below the averages in the OECD and in Eastern Europe and Central Asia. For example,
the science score in Georgia in the 2015 round of PISA was 411, while the regional average was
454, and the OECD average was 493. More than half of 15-year-olds in Georgia scored below basic
proficiency in science, reading, and mathematics.
8
See, for example, UNESCAP (2017), which shows high attendance rates in secondary education.
25
Figure 13: Learning performance: PISA scores
13a. PISA scores by subject 13b. Proficiency among 15-year-olds,
Georgia
Even more concerning, the educational disadvantages among the working-age population
appear to emerge in Georgia early in life. While school attendance seems to be universal, the
quality gaps are rather large, given that learning outcomes vary dramatically by the
circumstances of children (figure 14). The 30-point difference in the PISA score equivalent in
science between students of the same age and grade at the top and the bottom of the scale of
economic, social, and cultural status is roughly equivalent to a difference of three years of
schooling. The score differences by urban-rural location (36 points), gender (17 points), and the
duration of exposure to preschool (23 points) are also substantial.
26
Figure 14: Equity profile, PISA science score, Georgia, 2015
Children’s access to basic utilities
The inequality in the access to basic human capital inputs among children ages 0–16 measured
using a standard HOI on access to water and sanitation is generally low and mostly derives from
spatial disparities.9 In this analysis, the life circumstances of children are based on the number of
children in the household; the educational attainment, age, and gender of the household head;
the household consumption quintile; the gender of the children; and the urban-rural and
provincial location of the household.
Among the three countries, the share of children who have access to water appears to be highest
in Armenia (figure 15). In 2015, 96 percent of Armenian children had access to water, and the
HOI for access to running water was approximately the same, at 94 percent, highlighting the
narrow between-group inequality in access to the service. However, these large shares may
9
Because the data are generated though separate household surveys across the countries of South Caucasus, the
definitions of access to water and sanitation are also slightly different. In Azerbaijan, the definition revolves around
whether children have access to water in their dwellings, though the survey does not specify the source of the water.
In Armenia, children are considered to have access if their households obtain water from a central supply. In Georgia,
access refers to whether the households have water systems within the home or outside in the vicinity. Likewise, in
Azerbaijan, children are considered to have access to sanitation if there are toilets in their dwellings. In Armenia, access
refers to whether the households have local or centralized sanitation utilities. In Georgia, where all households claim
to have toilets, children are considered not to have access to sanitation if the household toilets are latrines emptying
into a river, channel, or ravine, and so on.
27
derive from the broader definition of access to water in the case of Armenia. In Azerbaijan, using
the definition of access to a central water supply, the coverage rate is 90 percent, while the HOI
is 85 percent. Using the definition based on whether households have access to a water system,
Georgia’s coverage rate is 79 percent, and the HOI is 68 percent. These relatively high rates of
access are in line with the data of the United Nations Economic and Social Commission for Asia
and the Pacific (UNESCAP 2017) on household access to improved drinking water among 41
Asia-Pacific countries that show coverage rates in Armenia and Georgia at nearly 100 percent
and in Azerbaijan at above 80 percent. All three countries also managed to meet the relevant
Millennium Development Goal (WHO and UNICEF 2015).10
Figure 15: Access to water and sanitation (0-16 years old)
a. Armenian b. Azerbaijan
10
The notion of coverage rates of the United Nations Economic and Social Commission for Asia and the Pacific
(UNESCAP 2017) and the World Health Organization and the United Nations Children’s Fund (WHO and UNICEF
2015) is different from the notion here. The definition in the former two sources is the percentage of the population
having access to water or sanitation, while the definition here is the percentage of children with access. The definitions
of access to water and sanitation differ as well.
28
c. Georgia
Among lower-middle-income countries, indicators of service quality, such as whether running
water is available for more than 12 hours a day, are more relevant. Based on these indicators, the
coverage rates are lower, and the inequality gaps are wider than the results of indicators simply
on access to running water. The coverage rate declines substantially, from 96 percent to 76
percent in Armenia, from 87 percent to 65 percent in Azerbaijan, and from 79 percent to 60
percent in Georgia (table 2). Inequality (D-index) also rises considerably, and, combined with the
lower coverage rates, this leads to an increase in the penalty factors associated with greater
inequality, that is, the gap between coverage rates and the inequality-adjusted HOI.
Table 2: Contribution of effort and hard work and circumstances to total
inequality in access to good jobs (%)
Running water - 12hrs more
Running water more/day Sanitation
D- D- D-
Coverage HOI Penalty index Coverage HOI Penalty index Coverage HOI Penalty index
Armenia 0.96 0.94 0.02 0.02 0.75 0.67 0.08 0.11 0.90 0.85 0.05 0.06
Azerbaijan 0.87 0.82 0.06 0.06 0.65 0.58 0.06 0.10 0.92 0.90 0.02 0.02
Georgia 0.79 0.68 0.11 0.14 0.60 0.47 0.13 0.22 0.93 0.89 0.04 0.04
Despite the differing definitions of access to sanitation, sanitation coverage rates are high in
South Caucasus: 90 percent in Armenia, 92 percent in Azerbaijan, and 93 percent in Georgia (see
table 2). Between-group inequality in children’s access to sanitation also appears to be narrow as
shown by the low D-index. Indeed, the United Nations Economic and Social Commission for
Asia and the Pacific (UNESCAP 2017) also assigns Armenia and Georgia among the countries
with the highest household sanitation coverage rates in the Asia-Pacific region. Regarding
29
coverage rates between 1990 and 2015, only Azerbaijan managed to meet its Millennium
Development Goal target, while Armenia and Georgia fell short (WHO and UNICEF 2015).
The striking feature of the composition of factors contributing to the penalty index is the
substantial role of spatial differences—whether children reside in urban or rural areas or in a
particular province—in opportunities in all three South Caucasus countries (figure 16). For
example, in access to water for at least 12 hours a day, 33 percent and 53 percent of the total
between-group inequality in Armenia arises from differences in urban-rural and provincial
residence, respectively. In Azerbaijan, the shares are 11 percent and 67 percent, and, in Georgia,
28 percent and 55 percent. The relatively high contribution of location may suggest a spatial
discrepancy in the supply of basic human capital inputs in the South Caucasus countries. In
Azerbaijan, the discrepancy reflects a regional disparity between Baku and the Absheron region
and the rest of the country. Moreover, while the contributions of other circumstances are
relatively small, differences in per capita incomes appear to play a considerable role in
Azerbaijan, where the differences in household income quintiles account for more than 10
percent of the inequality in access to water and sanitation among children.
Figure 16: Inequality decomposition (0- to 16-year-olds)
30
The research thus finds that educational performance still requires major improvement, and
children’s access to education is unequal. Meanwhile, children’s access to basic utilities are
generally high, although there is a spatial disparity in supply. Therefore, the fair component of
inequality appears to derive from a learning gap associated with children’s circumstances.
4. Conclusion and Policy Discussion
The main takeaways from the document are as follows. First, connections play an important role
in obtaining access to good jobs in South Caucasus, highlighting the unfairness in processes in
the sub-region’s labor markets. Second, access to good jobs—defined as work for 20 hours or
more a week and work under contract or with tenure—is low in South Caucasus relative to other
parts of the Europe and Central Asia region. Third, even among the people who have access to
good jobs, the share of the total inequality of opportunity that may be characterized as unfair is
relatively large. Armenia and Azerbaijan stand out for the significant share of inequality in access
to good jobs associated with gender differences. Fourth, the analysis on access to education and
basic human capital inputs in the earlier, formative stages of life shows that learning performance
in South Caucasus tends to be poor and unequal across the life circumstances of children.
Nonetheless, the coverage rates of basic human capital inputs are generally high; the relatively
narrow inequalities arise mostly from spatial disparities.
To some extent, these prevailing inequalities in access to good jobs suggest that there is a labor
market failure in South Caucasus, that is, the inability of the labor market to allocate the labor
force to the most efficient uses based on skills, efforts, and talents. The accompanying labor
market policies or reform in South Caucasus needs therefore to be directed toward addressing
this issue and ensuring that opportunities to participate in the labor market are not determined
by circumstances such as gender or parental social status, but, rather, skill and effort. It is
important that non-merit–based and discriminatory mechanisms do not distort the entry and
exit associated with good jobs (promotion), and wage rate policies.
In practice, policy makers in the countries of the South Caucasus need to promote policies that
enhance advancement based on the skills and effort of labor market participants, without
overreliance on potentially distorting signals, to sort out who deserves the access to good jobs.
One example of such policies is to ensure that school diplomas reflect the quality of an education
system that provides the skills necessary for the labor market. Various smaller initiatives and
campaigns to promote fairness in worker recruitment and promotion are also feasible and can
be effective in reducing inequality of opportunity in the labor market, especially if such
initiatives are based on credible evidence on the efficiency loss arising from unequal access
associated with circumstances beyond effort and skills. Technology and scientific findings can
also be used and promoted to help labor market actors design more efficient mechanisms to
identify available skills and match them with demand.
31
Labor codes and laws can also be enforced to eliminate discrimination in the labor market,
thereby reducing inequality of opportunity in the labor market. A legal framework fostering
equal opportunity and equal treatment in employment is available in South Caucasus, for
example, Armenia’s 2004 Labor Code (amended in 2011) and Azerbaijan’s 1999 Labor Code,
2001 Employment Act, and 2006 Law on the Guarantee of Gender Equality (ILO 2011, 2012).
These legal codes explicitly include legal equality in working relationships regardless of gender,
social status, political affiliation, and so on. However, reports of the International Labour
Organization (ILO 2011, 2012) suggest that the effective implementation of labor codes remains
an issue. Thus, enforcing the already available legal codes is the logical next step in labor market
reform to reduce the inequality in access to good jobs.
The empirical results described in this paper show that gender inequality is the main contributor
to inequality in access in South Caucasus. The International Labour Organization (ILO 2011,
2012) finds that Armenia and Azerbaijan have low female employment rates and wide gender
pay gaps. This highlights that, among the types of inequality hindering the access to good jobs,
it is necessary to prioritize policies that address gender-based inequality. The empirical findings
outlined in the paper also suggest that it is important to increase the quality of learning and to
focus on eliminating spatial disparities in the supply of education and basic utilities. Similarly,
policies should be oriented towards improving connections and matching of specific skills with
employment where these are needed. Finally, constant updating and upgrading of human
capital through ‘on the job training’ can help improve perceptions and actual quality of
employment in the South Caucasus.
As in any reform effort, the challenges are numerous. However, the benefits of addressing the
inequalities in the labor market are tangible in South Caucasus. First, reforms would be
economically rewarding because they would remove labor market distortions and increase the
incentives for investment in human capital and thus promote long-run growth. Second, reforms
would also be politically rewarding because the perceived inequality in the labor market is
positively associated with the public’s demand for redistribution. As the forces of globalization,
technological change, and aging continue to shape the economic and social landscape of the
broader Eastern Europe and Central Asia region, perceptions of reduced social mobility and
inequality of opportunity are generating distributional tensions that will likely exert greater
pressure on the existing social contract. Addressing the deep structural inequalities that are
shaping the landscape of opportunity in South Caucasus must be a key consideration in any
strategy to share prosperity sustainably.
32
References
Acemoglu, Daron, Simon Johnson, Amir Kermani, James Kwak, and Todd Mitton. 2016. “The
Value of Connections in Turbulent Times: Evidence from the United States.” Journal of Financial
Economics 121 (2): 368–91.
Alesina, Alberto F., and Dani Rodrik. 1994. “Distributive Politics and Economic Growth.”
Quarterly Journal of Economics 109 (2): 465–90.
Banerjee, Abhijit Vinayak, and Esther Duflo. 2003. “Inequality and Growth: What Can the Data
Say?” Journal of Economic Growth 8 (3): 267–99.
Barro, Robert J. 2000. "Inequality and Growth in a Panel of Countries." Journal of Economic Growth
5 (1): 5–32.
Barros, Ricardo Paes de, Francisco H. G. Ferreira, José R. Molinas Vega, and Jaime Saavedra-
Chanduví. 2009. Measuring Inequality of Opportunities in Latin America and Caribbean. With Mirela
de Carvalho, Samuel Franco, Samuel Freije-Rodríguez, and Jérémie Gignoux. Latin American
Development Forum Series. Washington, DC: World Bank; New York: Palgrave Macmillan.
Barros, Ricardo Paes de, José R. Molinas Vega, and Jaime Saavedra-Chanduví. 2010. “Measuring
Progress toward Basic Opportunities for All.” Brazilian Review of Econometrics 30 (2): 335–67.
Dabalen, Andrew L., Ambar Narayan, Jaime Saavedra-Chanduví, and Alejandro Hoyos Suarez.
2015. Do African Children Have an Equal Chance? A Human Opportunity Report for Sub-Saharan Africa.
With Ana Abras and Sailesh Tiwari. Directions in Development: Poverty Series. Washington, DC:
World Bank.
Engelhardt, Carina, and Andreas Wagener. 2014. “Biased Perceptions of Income Inequality and
Redistribution.” CESifo Working Paper 4838 (June), Center for Economic Studies and Ifo Institute,
Munich.
Ferreira, Francisco H. G., and Jérémie Gignoux. 2011. “The Measurement of Inequality of
Opportunity: Theory and an Application to Latin America.” Review of Income and Wealth 57 (4):
622–57.
Fisman, Raymond. 2001. “Estimating the Value of Political Connections.” American Economic
Review 91 (4): 1095–1102.
Forbes, Kristin. 2000. “A Reassessment of the Relationship between Inequality and Growth.”
American Economic Review 90 (4): 869–87.
ILO (International Labour Organization). 1999. “Decent Work: Report of the Director-General,
International Labour Conference, 87th Session.” International Labour Office, Geneva.
———. 2011. “Decent Work Country Profile: Armenia.” International Labour Office, Geneva.
———. 2012. “Decent Work Country Profile: Azerbaijan.” International Labour Office, Geneva.
33
Krishnan, Nandini, Gabriel Lara Ibarra, Ambar Narayan, Sailesh Tiwari, and Tara Vishwanath.
2016. Uneven Odds, Unequal Outcomes: Inequality of Opportunity in the Middle East and North Africa.
Directions in Development: Poverty Series. Washington, DC: World Bank.
Molina, Ezequiel, Ambar Narayan, and Jaime Saavedra-Chanduví. 2013. “Outcomes,
Opportunity, and Development: Why Unequal Opportunities and Not Outcomes Hinder
Economic Development.” Policy Research Working Paper 6735, World Bank, Washington, DC.
Niehues, Judith. 2014. “Subjective Perceptions of Inequality and Redistributive Preferences: An
International Comparison.” IW-Trends 2 (June 25), Cologne Institute for Economic Research,
Cologne.
Piketty, Thomas. 1997. “The Dynamics of the Wealth Distribution and Interest Rates with Credit
Rationing.” Review of Economic Studies 64 (2): 173–89.
Shorrocks, Anthony F. 2013. “Decomposition Procedures for Distributional Analysis: A Unified
Framework Based on the Shapley Value.” Journal of Economic Inequality 11 (1): 99–126.
Tiwari, Sailesh, César A. Cancho, Moritz Meyer, and Alan Fuchs. 2018. “South Caucasus in
Motion: Economic and Social Mobility in Armenia, Azerbaijan, and Georgia.” Policy Research
Working Paper 8329, World Bank, Washington, DC.
UNESCAP (United Nations Economic and Social Commission for Asia and the Pacific). 2017.
“Sustainable Social Development in Asia and the Pacific: Toward a People-Centred
Transformation.” UNESCAP, Bangkok.
World Bank. 2012. World Development Report 2013: Jobs. Washington, DC: World Bank.
———. 2015. “Skills Gaps and the Path to Successful Skills Development: Emerging Findings
from Skills Measurement Surveys in Armenia, Georgia, FYR Macedonia, and Ukraine.” Report
ACS14318 (July), World Bank, Washington, DC.
———. 2016. “PISA 2015: Europe and Central Asia.” World Bank, Washington, DC.
WHO (World Health Organization) and UNICEF (United Nations Children’s Fund). 2015.
“Progress on Sanitation and Drinking Water: 2015 Update and MDG Assessment.” Joint
Monitoring Programme for Water Supply and Sanitation, WHO, Geneva; UNICEF, New York.
34
Annex 1
How do the coverage rates based on the 2015 LiTS compare with the results of
the national Labor Force Survey?
In Armenia, the LiTS-based coverage rates are similar to the ones calculated from Labor Force
Survey. In Georgia, the coverage rates differ, but it is important to keep in mind that the
comparing coverage rates are calculated using Monitoring Household Survey, which is not
specifically designed as a Labor Force Survey. Azerbaijan’s Labor Force Survey, or similar survey,
is not available. See Table A1 for the complete figures.
Table A1: Coverage comparison (% of labor force, 18-64 years old)
Worked 20hrs/week, with Worked 20hrs/week, with
Worked 20hrs/week
contract tenure
LiTS LFS/HH survey* LiTS LFS/HH survey* LiTS LFS/HH survey*
Armenia 0.42 0.43 0.22 0.23 0.27 NA**
Georgia 0.41 0.55 0.22 0.35 0.13 0.53
Azerbaijan 0.26 NA*** 0.11 NA*** 0.20 NA***
Note: Armenia: *Use LFS 2016. NA**, data on tenure/permanent status for main job not available in the LFS dataset.
Georgia: *Use Monitoring of Household Survey module on Employment and Income (Shinda 05-1). By design, this
survey is not a specific Labor Force Survey. The major problem is that the question about working hours is asked for a
typical week (not the last seven days) and it is grouped into categories (less than 20 hours, 21–40 hours, 41–60 hours,
and depending on season). Also, the high permanent status of employment figures (92 percent of the total number of
people working in the last seven days) seems problematic. Azerbaijan: NA***, neither LFS or household monitoring
survey is available.
35
Annex 2
Respondent’s and parent’s education appear to have a large contribution to inequality in the labor
market using the HOI approach, especially in Georgia. Do other variables play a role in the
previous stage, of accessing good jobs? To check this, this paper runs probit regression analysis
of access to good jobs on respondent’s age, education, gender, parents’ education, parents’
political affiliation, and ethnicity.
The observable implication if respondent’s and parent’s educational attainment drives most of
the dispersion in accessing good jobs is that these variables has jointly significant effect on the
dependent variable of access to good jobs, also adding these variables to other variables
significantly increase the probit model goodness of fit (pseudo R-square). The regressions show
that in all specifications, the jointly hypothesis test for zero effect of respondents’ and parent’s
education on good jobs is rejected (see the result of the test at the bottom of Table A2.1 to A2.3).
In addition, adding respondent’s and parent’s educational attainment generally increases the
pseudo R-squared substantially.
36
Table A2.1: Probit regression – dependent variable: working 20 and more hours a week
Dep. Var: Worked 20 hours and more a week
Armenia Azerbaijan Georgia
(1) (2) (3) (4) (5) (6)
Age 0.0125** 0.00945 0.00782 0.00942 0.0144*** 0.0118*
(0.00437) (0.00491) (0.00515) (0.00563) (0.00431) (0.00467)
Male=1 0.588*** 0.692*** 1.374*** 1.365*** 0.183 0.199
(0.114) (0.120) (0.122) (0.136) (0.109) (0.114)
Parent communist=1 0.198 0.126 0.424* 0.406* -0.0285 -0.190
(0.155) (0.151) (0.180) (0.180) (0.140) (0.147)
Dummy ethnicity Yes Yes Yes Yes Yes Yes
Dummy respondent's
education Yes Yes Yes
Dummy parent's education Yes Yes Yes
Observations 898 855 992 855 904 869
Pseudo R-squared 0.0537 0.0955 0.197 0.227 0.0247 0.0800
Joint hypothesis test:
respondent's and parents'
education=0
df 23.43 16.49 54.41
chi2 9 7 10
Prob > chi2 0.01 0.02 0.00
Standard error in parenthesis, robust, * p<0.05 ** p<0.01 ***
p<0.001
37
Figure A2.1: Decomposition of inequality in access to good job, by work-experience relevant
industry
a. Working 20 hours or more per week
a. All industries b. Industries where experience c. Industry where experience less
matters matters
b. Working 20 hours or more per week, with contract
a. All industries b. Industries where experience c. Industry where experience less
matters matters
c. Working 20 hours or more per week, with tenure
a. All industries b. Industries where experience c. Industry where experience less
matters matters
38
Table A2.2: Probit regression – dependent variable: working 20 and more hours a week, with
contract
Dep. Var: Worked 20 hours and more a week, with contract
Armenia Azerbaijan Georgia
(1) (2) (3) (4) (5) (6)
Age 0.00817 0.00598 0.00117 0.00417 0.00257 -0.00207
(0.00475) (0.00551) (0.00579) (0.00665) (0.00489) (0.00548)
Male=1 0.213 0.297* 0.610*** 0.591*** 0.00694 0.0518
(0.127) (0.131) (0.138) (0.150) (0.128) (0.130)
Parent communist=1 0.433** 0.281 0.272 0.118 0.0669 -0.152
(0.167) (0.163) (0.219) (0.218) (0.148) (0.163)
Dummy ethnicity Yes Yes Yes Yes Yes Yes
Dummy respondent's education Yes Yes Yes
Dummy parent's education Yes Yes Yes
Observations 885 844 992 855 904 844
Pseudo R-squared 0.0285 0.0928 0.0516 0.118 0.00951 0.0878
Joint hypothesis test:
respondent's and parents'
education=0
df 33.52 31.27 41.43
chi2 9 7 8
Prob > chi2 0.00 0.00 0.00
Standard error in parenthesis, robust, * p<0.05 ** p<0.01 ***
p<0.001
Figure A2.2.: Decomposition of inequality in access to good job, by age cohort
a. Working 20 hours or more per week
39
a. All industries b. 18-40 years old c. 41-64 years old
b. Working 20 hours or more per week, with contract
a. All industries c. 41-64 years old
b. 18-40 years old
c. Working 20 hours or more per week, with tenure
a. All industries b. 18-40 years old c. 41-64 years old
40
Table A2.3: Probit regression – dependent variable: working 20 and more hours a week, with
tenure
Dep. Var: Worked 20 hours and more a week, with tenure
Armenia Azerbaijan Georgia
(1) (2) (3) (4) (5) (6)
Age 0.00643 0.00308 0.00455 0.00708 -0.00393 -0.00751
(0.00465) (0.00517) (0.00519) (0.00562) (0.00575) (0.00608)
Male=1 0.256* 0.308* 1.167*** 1.129*** -0.0977 -0.128
(0.125) (0.129) (0.127) (0.138) (0.144) (0.146)
Parent communist=1 0.371* 0.281 0.501** 0.460* 0.0417 -0.141
(0.162) (0.156) (0.181) (0.181) (0.162) (0.183)
Dummy ethnicity Yes Yes Yes Yes Yes Yes
Dummy respondent's
education Yes Yes Yes
Dummy parent's education Yes Yes Yes
Observations 898 855 992 855 895 837
Pseudo R-squared 0.0237 0.0676 0.154 0.188 0.00724 0.119
Joint hypothesis test:
respondent's and parents'
education=0
df 25.11 23.62 43.53
chi2 9 7 8
Prob > chi2 0.00 0.00 0.00
Standard error in parenthesis, robust, * p<0.05 ** p<0.01 ***
p<0.001
41
Annex 3
Is work experience correlated with access to good jobs?
The paper uses respondent’s age to indicate her work experience as a `fair’ labor market
inequality component. However, for transition economies, a massive economic restructuring in
the 1990s, may render the role of work experience for getting access to good jobs meaningless in
South Caucasus countries.
The first observable implication if work experience had no meaningful effect on access to good
jobs is that respondent’s age, as the proxy of work experience, would have relatively greater
contribution to inequality in access to good job in industries where experience remains mattered
even after the massive restructuring (manufacturing) or characteristically specific-human capital
intensive such as high-quality services (finance and public administration), than it would in other
industries. For this first observable implication, the evidence shows that respondent’s age, as
proxy for working experience, does not have a greater contribution in manufacturing and high-
quality services industries than it does in other industries (see the comparison of the percentage
of age contribution to labor market inequality on Table A2.1, and Figure Table A2.1); therefore,
rejecting the observable implication for the idea that work experience is not correlated with access
to good jobs.
42
Figure A.3: Decomposition of inequality in access to good job, by skill-mismatch industry level
a. Working 20 hours or more per week
a. All industries b. Skill-mismatch industries c. Not kill-mismatch industries
b. Working 20 hours or more per week, with contract
a. All industries b. Skill-mismatch c. Not skill-mismatch
industries industries
c. Working 20 hours or more per week, with tenure
a. All industries b. Skill-mismatch c. Not skill-
industries mismatch
industries
43
Table A3.1: D-index and contribution of age to labor market inequality by work-experience
relevant industry
Manufacturing & high
All industries services industries Other industries
D-index Age D-index Age D-index Age
contribution contribution contribution
(%) (%) (%)
Working >=20 hrs a week
Armenia 0.18 9 0.16 3 0.21 10
Azerbaijan 0.39 1 0.38 1 0.41 1
Georgia 0.16 21 0.37 3 0.16 19
Working >=20 hrs a week,
with contract
Armenia 0.24 6 0.21 6 0.27 11
Azerbaijan 0.32 0 0.36 1 0.32 1
Georgia 0.23 1 0.39 1 0.25 0
Working >=20 hrs a week,
with tenure
Armenia 0.19 5 0.13 1 0.24 6
Azerbaijan 0.38 1 0.40 0 0.40 1
Georgia 0.31 5 0.43 3 0.30 5
The second observable implication if massive industry restructuring in 1990s attenuated the link
between work experience and access to good job is that for the younger cohort of labor forces,
aged 40 years old and less, work experience would have more profound effect on access to good
job than for the older cohort. The evidence shows that respondent’s age, as the proxy for work
experience, does not have a greater effect on inequality for access to good job for younger cohort
of labor forces than it does for older one (see the comparison of the percentage of age contribution
to labor market inequality on Table A2.2, and Figure A2.2); therefore, rejecting the observable
implication for the idea that work experience has no meaningful effect on access to good job.
44
Table A3.2: D-index and contribution of age to labor market inequality, by age cohort
All ages Young workers Old workers
D- Age D- Age D- Age
index contribution index contribution index contribution
(%) (%) (%)
Working >=20 hrs a week
Armenia 0.18 9.00 0.36 48.00 0.48 71.00
Azerbaijan 0.39 1.00 0.47 19.00 0.54 56.00
Georgia 0.16 21.00 0.39 64.00 0.45 68.00
Working >=20 hrs a week,
with contract
Armenia 0.24 6.00 0.37 38.00 0.56 60.00
Azerbaijan 0.32 0.00 0.46 35.00 0.56 56.00
Georgia 0.23 1.00 0.46 52.00 0.47 55.00
Working >=20 hrs a week,
with tenure
Armenia 0.19 5.00 0.38 44.00 0.50 65.00
Azerbaijan 0.38 1.00 0.48 18.00 0.52 58.00
Georgia 0.31 5.00 0.52 47.00 0.54 41.00
45
Annex 4
Is educational attainment associated with work skill?
The paper assumes educational attainment is an important marker for skill, as a `fair’ labor
market inequality component. However, many studies in South Caucasus countries point out
significant skill mismatch due to poor teaching quality and outdated curricula that often do not
allow for good job transitions.
The observable implication if there were a skill-mismatch problem is that industries where skill-
mismatch are more severe would witness relative higher contribution of educational attainment
to labor market inequality – the workers are more over-educated, relative to the required level of
skills for working in these industries. In South Caucasus, industries with severe skill-mismatch
are agriculture sector since working in agricultural sector was often not really a voluntary choice
but rather the first best mitigation strategy in times of shocks. In addition, another typical
industry with skill mismatch is the low-quality service sector.
The evidence shows that respondent’s educational attainment, as the proxy for skill, does not
have a greater effect on inequality for access to good job in industries associated with more severe
skill mismatch, relative to sectors with less-severe skill mismatch (see the comparison of the
percentage of educational attainment to labor market inequality on Table A3, and Figure A3.);
therefore, rejecting the observable implication for the idea that educational attainment is not a
proper measure for skill.
46
Table A.4: D-index and contribution of age to labor market inequality, by skill-mismatch
industry level
Skill-mismatch
All industries industries Other industries
D- Age contribution D- Age D- Age
index (%) index contribution index contribution
(%) (%)
Working >=20 hrs a week
Armenia 0.18 9 0.16 3 0.21 10
Azerbaijan 0.39 1 0.38 1 0.41 1
Georgia 0.16 21 0.37 3 0.16 19
Working >=20 hrs a week,
with contract
Armenia 0.24 6 0.21 6 0.27 11
Azerbaijan 0.32 0 0.36 1 0.32 1
Georgia 0.23 1 0.39 1 0.25 0
Working >=20 hrs a week,
with tenure
Armenia 0.19 5 0.13 1 0.24 6
Azerbaijan 0.38 1 0.40 0 0.40 1
Georgia 0.31 5 0.43 3 0.30 5
47