Akkoyunlu-Wigley & Wigley
1
A Regional Analysis of Capabilities and Education in Turkey
Arzu Akkoyunlu-Wigley Department of Economics
Hacettepe University arzus [at] hacettepe.edu.tr
Simon Wigley Department of Philosophy
Bilkent University wigley [at] bilkent.edu.tr
August 31, 2009
Abstract
The purpose of this paper is to assess education in Turkey based on capabilities approach pioneered by Amartya Sen (1999) and, more recently, Martha Nussbaum (2000). In keeping with capabilities approach, we argue that recognizes that educational attainment enables the acquisition of a set of basic cognitive functionings (e.g. problem solving, numeracy, literacy, etc) that are vital to realization of other human functionings (e.g. avoiding preventable illness and premature death). We contend that the value of education should be measured in terms of health outcomes because that measure directly or indirectly captures a number of other fundamental human functionings. With this purpose, we presented an overview of education in Turkey as well as of recent developments in the expansion and distribution of educational attainment in Turkey. We test claim that education has an income-independent effect on health by using a panel data set composed of economic, educational and mortality data for 64 geographic regions in Turkey for the years 1980, 1985, 1990 and 2000. We find that education (as indicated by secondary enrollment ratio) has a significant effect on health functionings (indicated by infant mortality). What we have found in the Turkish case is that the ability to avoid preventable illness and premature mortality constitutes an important basis for evaluating the basic cognitive functionings that are enabled by the education system.
Key words: capabilities approach, education in Turkey, regional analysis of education, effect of education on health
Dr. Arzu Akkoyunlu Wigley graduated from Middle East technical University (B.A, Economics). She
completed her M.A (Economics) at Middle East Technical University and Ph.D. (Economics) at Middle East
Technical University and her Ph.D. (Economics) at Hacettepe University. She was the recipient of a Jean
Monet Scholarship to carry out doctoral research at the London School of Economics. She is currently
working as an Associate Professor at the department of Economics, Hacettepe University. Her interest lies in
the following areas; regional economic integration, economics of European Union and education economics.
Her recent publications are; “Human Capabilities Versus Human Capital: Gauging the Value of Education in
Developing Countries”, Social Indicators Research, September 2006 (with Simon Wigley), “Effects of the
Customs Union with the European Union on the Market Structure and Pricing Behavior of the Turkish
Akkoyunlu-Wigley & Wigley
2
Manufacturing Industry”, Applied Economics, 2006, (with Sevinç Mıhçı), “Basic Education and Capability
Development in Turkey” in Education in Turkey (New York/ Münster: Waxmann Publishing) (co-edited with
Arnd-Michael Nohl & Simon Wigley “The Impact of the Customas Union with the European Union on Turkey’s Economic Growth” (Argumenta Oeconomica, 1(22), 2009.
Dr. Simon Wigley graduated from the Department of Philosophy and Politics at the University of Otoga
(New Zealand). He completed his masters and doctoral degrees in political philosophy at the London School
of Economics and Political Science. Dr. Wigley is currently working as an assistant professor in the
Department of Philosophy at Bilkent University. His interest lies in the following areas: democratic theory,
distributive justice and well-being. Among his recent publications are Parliamentary Immunity in
Democratizing Countries: The Case of Turkey (Human Rights Quarterly(forthcoming), Automaticity,
Consciousness and Moral Responsibility (Political Philosophy, 2007) and Voluntary Losses and Wage Compensation (Politics, Philosophy and Economics, 2006)
Introduction
In this paper we assess education in Turkey based on the capabilities approach pioneered by
Amartya Sen (1999) and, more recently, Martha Nussbaum (2000). That approach explicitly
recognizes that education has a positive effect on human well-being independently of its effect on
income growth or poverty reduction. Educational attainment enables the acquisition of a set of basic
cognitive functionings (e.g. problem solving, numeracy, literacy, etc) that are vital to realization of
other human functionings (e.g. avoiding preventable illness and premature death). We contend that
the value of education should be measured in terms of health outcomes because that measure
directly or indirectly captures a number of other fundamental human functionings (Sen, 1998;
Wigley and Akkoyunlu-Wigley, 2006). Moreover, we contend that education has a positive effect on
health both because of its effect on resource accumulation and independently of its effect on
resource accumulation (e.g. being adequately nourished requires both food and an understanding of
one’s nutritional requirements). Hence, a resourcist metric of the value of education (e.g. growth in
GDP per capita, reduction in income poverty, income returns to education) will only partially
capture the beneficial effect of education on life quality (as measured by population health). In
Section 5 we present statistical evidence in support of that claim. We begin by presenting an
Akkoyunlu-Wigley & Wigley
3
overview of education in Turkey present statistical evidence in support of that claim. We begin,
however, by presenting an overview of recent development in the expansion and distribution of
educational attainment in Turkey (Sections 1-2). We then turn to consider whether the expansion of
educational attainment has been accompanied by an improvement in learning outcomes as measure
by Programme for International Student Assessment (PISA) (Section 3). In subsequent section we
examine regional disparities in education, infant mortality and average income (Section 4). Finally,
we test claim that education has an income-independent effect on health by using a panel data set
composed of economic, educational and mortality data for 64 geographic regions in Turkey for the
years 1980, 1985, 1990 and 2000 (Section 5).
1. Cross-country Overview
A striking feature of development in Turkey is that it compares well amongst low and middle
income countries in terms of income poverty and income distribution, but not in terms of education
and health outcomes (Akkoyunlu-Wigley and Wigley, 2008, pp. 276-281).1 The divergence between
development as measured in terms of resource growth and development as measured in terms of the
expansion of educational attainment is apparent from Figures 1 and 2. In Figure 1 we take average
calorie intake as our indicator of resource poverty. Amongst developing countries increases in
average calorie consumption are more likely to benefit the poor because of the biological limit on
how much a person can consume and the difficulty of hoarding food.2
1 There is a similar discrepancy with respect to economic and human development in Turkey. Although, Turkey is classified as the seventeenth most industrialized country in the world, it ranked only seventy-sixth out of 179 countries on the UNDP Human Development Index in 2008 (a measure that combines life expectancy, literacy, educational attainment and average income) (UNDP, 2008). 2 See Blaydes and Kayser (2007) and Wigley and Akkoyunlu-Wigley (2009) for empirical evidence in support of that claim.
Akkoyunlu-Wigley & Wigley
4
*Data source: Food and Agriculture Organization, ‘Food Balance Sheets,’ data extracted from http://faostat.fao.org/.
For Figure 2 we take the proportion of the female population aged 15 and over with no schooling as
our measure of the shortfall in educational attainment. For both graphs we included representative
low and middle-income countries with a similar proportion of the population aged 14 and below
between the years 1960 and 2000.
Akkoyunlu-Wigley & Wigley
5
* Data source: Barro and Lee (2000), 'International Data on Educational Attainment: Updates and Implications,' (CID Working Paper No. 42, April 2000), Retrieved from: http://www.cid.harvard.edu/ciddata/ciddata.html. Note: data not available for China during the years 1960-1975.
What we observe is that even though Turkey has a significantly higher level of calorie consumption
throughout the last four decades of the twentieth century (Figure 1), only China had a (slightly)
higher proportion of non-schooled females in 2000 (Figure 2). The reduction in the share of the
female population without schooling in Turkey between 1960 and 2000 was significant. However,
many of the other representative countries were able to achieve similar results even though they are
poorer with respect to per capita calorie intake. Thus, according to a resource-based metric of
deprivation (calorie intake) Turkey compares favorably. However, according a capabilities-based
metric of deprivation (female schooling) the opposite is the case. Thus, a development strategy that
emphasizes the level of resource deprivation may conclude that there is not an urgent need to rectify
the shortfall in female education in Turkey. What we argue is that public policy should be
determined by assessing the impact of goods such as food holdings and educational attainment on a
generic measurement of human well-being, namely, ability to avoid preventable illness and
premature death.
Akkoyunlu-Wigley & Wigley
6
2. Recent Changes in the Distribution of Educational Attainment in Turkey
Since the turn of the century there has been a significant improvement in the level and distribution
of educational attainment in Turkey. In 1997 compulsory education was extended from grade 5 to
grade 8 (ages 11 to 13). Five year primary schooling was combined with three year lower secondary
schooling to create an eight-year basic education cycle.
* Data source: Filmer (2009) 'Educational Attainment and Enrollment around the World,' (Washington, DC: World Bank). Data retrieved from http://iresearch.worldbank.org/edattain/
In order to examine the impact of that reform on attainment levels we refer to the Demographic
and Health Surveys for the years 1993, 1998 and 2003. Those surveys allow us to pinpoint the
proportion of 13 year olds currently enrolled in primary education and, therefore, to estimate the
proportion of children actually completing 8 years of schooling. Figure 3 shows the net enrolment
ratio for all 13 year olds, those from the richest 20% of the population, those from the poorest 40%
of the population, those who are male or female, as well as the ratio of female enrolment to male
enrolment. Across all six categories there was only a marginal improvement between the 1993 and
1998 surveys, but a dramatic improvement after the expansion of compulsory schooling.
Akkoyunlu-Wigley & Wigley
7
Nevertheless, it remains the case that a significant proportion of those aged 13 were not completing
eight years of compulsory education in 2003 (approximately 9% of all 13 years olds, 18% of the
poorest 13 year olds and 15% of female 13 year olds).
3. The Distribution of Learning Outcomes in Turkey
We now turn to consider the quality of education that students receive if they do complete the eight
years of compulsory education. To evaluate learning outcomes in Turkey we refer to the Programme
for International Student Assessment (PISA) studies of 15-year-olds enrolled in formal education in
2003 and 2006 (OECD, 2004a; OECD, 2007b). The PISA studies show that socioeconomic
background has a significant impact on learning outcomes in Turkey. In 2003 Turkey recorded the
highest variance amongst the 40 participating countries in terms of between school variance in
mathematics proficiency. Moreover, most of that variation was explained by the socioeconomic
profile of schools (OECD, 2004a, pp. 162, 187-190). Similar results were recorded for the curricula
domain of science in 2006 (OECD, 2007b, pp.171, 187-189). In addition, a significant proportion of
15-year-olds in Turkey could not achieve more than the lowest proficiency level in the domains of
problem solving, mathematics, science and reading (OECD, 2004b, table 2.1; OECD, 2007b, tables
2.1a, 6.1a, 6.2a). In the domain of science, for example, 47 per cent of 15 year olds in 2006 could not
achieve above 1 on the 6 level proficiency scale. Moreover, 13 per cent could not even achieve level
1. According to the PISA criteria, students at level 1 “…have such a limited scientific knowledge
that it can only be applied to a few, familiar situations” (OECD, 2007b, p. 43).
One shortcoming of the PISA assessments is that they are based on the performance of 15-
year-olds that are enrolled in formal education. Thus, differential drop-out rates between countries
may significantly affect the results. Indeed Turkey has the lowest proportion of enrolled 15-year-olds
amongst those countries that participated in PISA 2003 and PISA 2006 (OECD 2004a, table A10.1;
OECD 2007b, table A3.1). Thus, it is very likely that the PISA assessments overstate the average
proficiency scores of those students in Turkey who have completed primary education and
understate the impact of socioeconomic status in Turkey (OECD, 2004a, p. 184; OECD, 2007b, p.
190).
4. Overview of Regional Disparities
In this and the following section we consider the effect of regional variations in educational
attainment on infant mortality in Turkey. The link between education (especially female education)
Akkoyunlu-Wigley & Wigley
8
and the mortality rates of infants and children is well-established in the literature (see e.g. Sen, 1999,
pp. 195-198; Cutler, Deaton, Lleras-Muney, 2006). Elsewhere we have argued that the incidence of
morbidity and mortality represents a more appropriate metric of the value of education than (micro
or macro level) income growth (Wigley and Akkoyunlu-Wigley, 2006). There we presented cross-
country evidence in support of the claim that education (as indicated by average years of schooling)
has a positive and significant impact on health (as indicated by life expectancy) independently of the
impact it has by way of income growth. Thus, income returns to education represents an
impoverished way of measuring the impact of education on human well-being (interpreted as the
ability to avoid preventable illness and premature death).
This study differs in two key respects. Firstly, our statistical analysis is based on a panel of
Turkish regions rather than panel of developing countries. Secondly, we take infant mortality
(number of children who perish during the first year of life, per 1,000 live births), rather than life
expectancy as our indicator of population health. Infant mortality typically afflicts the poorest
segment of the population and it is usually due to causes which are comparatively easier to prevent
than the causes of adult mortality. Thus, we would expect that it is particularly sensitive to changes
in the education levels of the most disadvantaged members of society. By extension we would
expect that the poorest geographic regions are characterized by higher infant mortality rates. Indeed,
as we will now see, that pattern is clearly evident in the Turkish case.
When we consider the regional variation in infant mortality and adult literacy between 1980
and 2000 (Table 1) we can see that both those indicators have improved within each region over
time. Even so there remains considerable disparity between the various regions. Perhaps,
unsurprisingly, those disparities are in keeping with regional differences in average income
(represented here as the ratio of GDP per capita for each region divided by national GDP per capita
for each year).3
3 There is also significant discrepancy between male and female enrolment (at both the primary and secondary levels) within each of the poorer regions, as well as between female enrolment in those regions and female enrolment in the more prosperous western regions and the country as a whole (Akkoyunlu-Wigley and Wigley, 2008, p. 285).
Akkoyunlu-Wigley & Wigley
9
Table 1: Regional Disparities in Infant Mortality, Education and Income
REGION YEAR
INFANT MORTALITY RATE
GDPPERCAPITA-RATIO
LITERACY RATE
MEDITERRENIAN 1980 112.00 0.99 67.46 1985 97.50 0.87 78.22 1990 59.67 0.95 81.00 2000 37.71 0.82 87.94 EASTERN ANATOLIA 1980 146.00 0.46 50.60 1985 132.64 0.44 63.28 1990 75.64 0.39 67.33 2000 52.77 0.33 78.36 SOUTH EAST 1980 128.00 0.50 42.08 1985 115.57 0.47 55.39 1990 71.14 0.63 59.80 2000 49.00 0.49 72.75 MARMARA 1980 113.40 1.38 76.91 1985 95.30 1.45 84.01 1990 60.50 1.40 86.22 2000 39.36 1.49 91.40 CENTRAL ANATOLIA 1980 139.40 0.81 68.70 1985 104.70 0.79 78.77 1990 69.10 0.77 82.09 2000 41.77 0.78 88.83 AGEAN 1980 123.50 1.05 71.43 1985 102.88 1.09 80.07 1990 65.50 1.05 82.82 2000 39.63 1.11 89.10 BLACK SEA 1980 135.33 0.72 62.98 1985 116.73 0.67 75.36 1990 69.47 0.67 78.46 2000 42.16 0.76 86.10
Source: State Institute of Statistics (2003)
Has the variation between regions decreased over time? To answer that question we
calculated the coefficient of variation for each indicator for 1980 and 2000 (Coefficient of variation
is calculated by dividing standard deviation with arithmetic mean).
Akkoyunlu-Wigley & Wigley
10
Table 2: Coefficient of Variations of Regional Indicators
Coefficient of Variations INFANT MORTALITY 0.0931 0.1193 LITERACY RATE 0.1812 0.074 SECONDARY ENROLLMENT RATIO 0.1755 0.1986 GDP PER CAPITA 0.3593 0.4277
According to the coefficient of variations in Table 2, the largest regional disparity is in GDP
per capita. The change in the coefficient of variations captures the evolution of the magnitude of the
disparities for the two time periods. The change in the coefficient of variations from 1980 to 2000
shows that only the coefficient of variation for literacy rate has declined among the four variables.
In other words regional diversity only declined with respect to literacy rates, whilst increasing with
respect to GDP per capita, infant mortality and the secondary enrollment ratio.
5. Estimation of the Income-Independent Effect of Education on Health in Turkey
In order to examine the relationship between education and infant mortality we use a panel data set
composed of 64 Turkish cities for the years 1980, 1985, 1990 and 2000. The main data set is
obtained from State Institute of Statistics. The estimation equation is specifically designed to clarify
the effects of education on health (as indicated by infant mortality). The dependent variable
INFANT (Infant Mortality) is calculated as the ratio of infant mortality of each city in the national
overall infant mortality for each year. Rather than using the absolute amounts, we prefer to use the
difference of each city from overall average since it shows the proportionate deviation from the
national average. In terms of the explanatory variables we take the secondary school enrolment ratio
(EDU) as our indicator of education levels in each region. As with the infant mortality variable this
variable is also the ratio of secondary school enrolment ratio of each city in the overall average
secondary school enrolment ratio for each year. GDP per capita (GDP) is the other explanatory
variable introduced to analyze the effects income on infant mortality. The GDP variable is obtained
Akkoyunlu-Wigley & Wigley
11
by dividing the GDP per capita of every city by national GDP per capita for every year. The last
explanatory variable introduced in our model is the unemployment rate (UNEMPLOYMENT) at
the city level. The main motivation for including this variable in our equation is that it is a good
proxy for poverty. This variable is also obtained by dividing each city’s unemployment rate to overall
unemployment rate. Accordingly following equation is estimated;
NTUNEMPLOYMEGDPEDUINFANT 3210 (1)
Table 3 shows the estimation results of the infant mortality equation for Turkey. The fixed-effect
specification of the panel data that we use helps to control for the possibility of unmeasured region-
specific factors. This is necessary given that regional education levels may be due to unmeasured
factors (e.g. proportion of parents that are tradition-bound) that are also correlated with mortality
rates.
Table 3: Estimation Results for Equation (1)
Variable Coefficient Std. Error t-Statistic Prob.
C 1.077269 0.032733 32.91063 0.0000
GDP -0.042127 0.024837 -1.696147 0.0915
EDU -0.087021 0.024121 -3.607698 0.0004
UNEMPLOYMENT 0.057216 0.018752 3.051282 0.0026
R2 0.995972
Adj.R2 0.994566
Durb.Wat. 2.081894
F-stat
708.1248
Note: Standard errors and t-statistics of coefficients are computed using White’s heteroscedasticity consistent variance-covariance estimator.
Akkoyunlu-Wigley & Wigley
12
From Table 3, one can see that all of our explanatory variables for infant mortality are
statistically significant at the 1% level. EDU variable has a negative sign in line with the theoretical
expectations and indicates a negative relationship between education and infant mortality. In other
words, an increase in the education level which is measured by the secondary enrollment ratio
decreases the infant mortality in the Turkish case. We can conclude, therefore, that educational
functioning which is represented by the secondary school enrolment ratio (EDU) has a positive
effect on health functionings captured by infant mortality. As we expected our income variable
GDP per capita has a negative sign indicating that relatively high income cities have a low infant
mortality rate. As expected our indicator or income poverty (UNEMPLOYMENT) is negatively
associated with infant mortality. Taken together these results suggest that education has a reductive
effect on infant mortality independently of its effect on average income and poverty.
Conclusion Our analysis of geographic regions in Turkey provides further support for the claim that income
growth represents an impoverished metric of the value of education. The ability to avoid
preventable illness and premature mortality is crucial to human well-being (more precisely it
encapsulates heath functionings that are fundamental to leading a life worth living). Thus, it
constitutes an important basis for evaluating the basic cognitive functionings (problem solving,
numeracy, literacy etc) that are enabled by the education system. What we have found in the
Turkish case is that the impact of education on resource accumulation (i.e. average income and
poverty reduction) does not fully capture the full effect of education on health.
References
Akkoyunlu-Wigley, A. and S.D. Wigley,(2008) A. Basic Education and Capability Development in Turkey, in Education in Turkey, A.-M. Nohl, A. Akkoyunlu-Wigley, and S. Wigley, eds., Waxmann Publishing, New York/Münster
Barro, R.J and Jong-Wha Lee, (2000) 'International Data on Educational Attainment: Updates and
Implications,' (CID Working Paper No. 42, April 2000),
Blaydes, L. and M. Kayser, (2007) ‘Counting Calories: Democracy and Distribution in the
Developing World’, paper delivered to American Political Science Association Annual
Meeting 2007, retrieved from http://www.stanford.edu/~blaydes/Calories.pdf.
Akkoyunlu-Wigley & Wigley
13
Cutler, D., A. Deaton and A. Lleras-Muney, (2006) ‘The Determinants of Mortality’, Journal of
Economic Perspectives, 20(3), 97-120
Filmer, D. (2009) Educational Attainment and Enrollment Around the World, Development Research Group (World Bank, Washington DC), Data retrieved from http://econ.worldbank.org/projects/edattain
Food and Agriculture Organization, ‘Food Balance Sheets,’ data extracted from
http://faostat.fao.org/.
Nussbaum, M., 2000, Women and Human Development: The Capabilities Approach (Cambridge University Press, Cambridge)
OECD, 2004a, Learning for Tomorrow’s World – First Results from PISA 2003 (OECD, Paris)
OECD, 2004b, Problem Solving for Tomorrow’s World – First Measures of Cross Curricular Competences from PISA 2003 (OECD: Paris)
OECD, 2007a, Reviews of National Policies for Education: Basic Education in Turkey (OECD, Paris)
OECD, 2007b, PISA 2006 Science Competences for Tomorrow’s World (OECD, Paris)
Sen, A. (1999) Development as Freedom (New York: Anchor Books, 1999)
Sen, A. (1998) ‘Mortality as an Indicator of Economic Success and Failure’, The Economic Journal 108
(446), 1-25
State Institute of Statistics (2003), Provincial Indicators, 1980-2003
UNDP (2008) Human Development Report (UNDP, New York)
Wigley, S. and Akkoyunlu-Wigley, A., (2006), ‘Human Capabilities versus Human Capital: Gauging the Value of Education in Developing Countries,’ Social Indicators Research, 78(2): 287-304.
Wigley, S. and Akkoyunlu-Wigley, A. (2009) ‘The Political Determinants of Health: A Cross-
National Study,’ Paper presented to the Annual Meeting of the American Political Science
Association, Toronto, Canada, September 3-6, 2009