Размещено на http://www.allbest.ru/

Ethnic Diversity and Public Goods Provision in Turkey

Abstract

ethnic kurdish economic investment

This research paper examines the empirical relationship between ethnic diversity and publicly provided goods, on the sub-national level in Turkey. The aim is to analyze the effects of the increased economic growth in the early 2000s through the changes in the public delegation of goods in education and health. At the beginning of the 2000s a new political party took power in Turkey, promising a boost in economic growth and higher standards of living. The Justice and Development Party (AKP) leader, Recep Tayyip Erdoрan, highlighted the focus on the increase in economic activity in all regions, and investments in economic development. However, southeastern, Kurdish provinces still have significantly worse socioeconomic development scores. Thus, the research answers the question whether the delegation of the funds for health and education was economically, or ethnically based, and how did the AKP mandate affect the public provision of goods. Using the quantitative analysis on the panel data, findings imply that Kurdishness of a province has a significantly negative relationship with the public goods provision measured by proxy variables, and the Public Provision Index. Therefore, Kurdish provinces did on average receive less public goods. Conversely, looking only at the AKP mandate, Kurdish provinces profited significantly more than others. Because of the investment incentives created specifically to boost the economic growth in the underdeveloped Kurdish areas, AKP-led government managed to increase human capital. Annual increase in economic activities led to the higher public funding in education and health.

The main conclusion of the paper is that Kurdish provinces profited on average more than others, specifically because of the decisions made by the AKP government. Erdogan's method of solving the ethnic issues seems to be mostly related to the economic performance. Instead of dealing with the difficult process of political discussions which are also unpopular and could harm his election performance, he focused on boosting incomes and living standards, combined with slow sociopolitical changes.

Keywords

Economic development, public goods provision, ethnic diversity, central government, Turkish provinces, Kurds;

List of Abbreviations

CRC - Convention on the Rights of the Child

MHP - Nationalist Movement Party

CHP - Republican People's Party

DHS - Demographic and Health Survey

TURKSTAT - Turkish National Statistical Institute

TRY - Turkish Lira

GRP - Gross Regional Product

EFA - Exploratory Factor Analysis

OLS - Ordinary Least Squares

IMR - Infant mortality rate

KMO - Kaiser-Meyer-Olkin's test

UNHCR - The UN Refugee Agency

UNICEF - United Nations International Children's Emergency Fund

Introduction

This research focuses on disparities in the economic development on the subnational level in Turkey. The primary goal of the study is to explore the relationship between ethnic diversity and public goods delivery in the period 2009-2014. Turkey, a country characterized by many controversies based on human rights violations towards minorities, changed its political discourse at the beginning of the 2000s, with the new political party taking power and focusing on high economic growth in all provinces. Recep Tayyip Erdoрan, the prominent leader of the Justice and Development Party (AKP), is responsible for several reforms that allowed Turkey to develop faster. Education expenditure rose almost five times and became the most funded sector by the government budget (note that before, the biggest share of the budget was spent on the military). After 2008, all provinces had at least one university, and mandatory schooling rose from eight to twelve years. Infrastructure bloomed also. Starting in 2002, the number of airports in Turkey increased from 26 to 50. Railroads, highways, and high-speed railway lines were constructed, which helped reduce road accidents by 50%. Additionally, under Erdogan's lead, the government reformed the healthcare system, allowing poor and young people to have free access to health.

On the other hand, widely known ethnic issues persist. The second biggest ethnic group, Kurds, have not been officially recognized yet. Due to the complete elimination of all questions regarding ethnicity in population censuses, it has been very hard to determine the exact number of Kurds in Turkey. Estimates of the size of Kurdish population from various sources range from 15 to 25% of the total population, occupying mostly the eastern and south-eastern parts of the country.

Ethnic conflicts in this area started after the abolition of the Ottoman Empire. The official abolition of the Sultanate in 1922, and the establishment of the Republic of Turkey by the Treaty of Lausanne of 24 July 1923 (Axiarlis, 2014), kept the previous Kurdistan Emirates under the Turkish territory. Nationalization and state-building, two major strategies of the Turkish government, recognized only one ethnic group - Turks, and one language - Turkish. While the Turkish state recognized some non-Muslim minorities (Jews, Greeks, Armenians), by signing the Lausanne Agreement (Bulbul, 2016), the state excluded Kurds from this agreement making the Kurdish culture and language illegal, and their traditional - tribal way of life marginalized and oppressed.

The twentieth century represented a volatile period for the Kurdish population, characterized by oppression, discrimination, and violence by the Turkish government and local military organizations such as the Kurdistan Workers Party (PKK). The so-called process of assimilation had big consequences on the people living in eastern areas, claiming over 40 000 lives, as Hassanpour (1992) reports. After almost 90 years of ignorance and exclusion of the language from any public use, the government announced that the Kurdish language would be permitted as an elective course in universities, starting in 2009 (Bulbul, 2016). Therefore, AKP was the first government which made certain moves towards the resolution of the Kurdish question, despite the criticism that the government efforts are small and insignificant.

Hence, this research paper explores the relationship between public goods provision and ethnic structure in 81 Turkish provinces, in the period 2009-2014. The aim is to understand the empirical relationship between ethnic diversity and public goods provision on sub-national levels in Turkey, and the efficiency of that provision. Thus, this empirical study of answers two research questions - Was the delegation of public goods from the government ethnically, or economically driven? And, did Kurdish provinces have different treatment in public goods provision during the AKP mandate?

Firstly, the chapter on literature review gives an insight on the theories of ethnic diversity, and the empirical findings on the relationship between ethnic diversity and public goods provision. It, additionally, contains the history of Kurdish population in Turkey, and the characteristics of the Kurdish people. In the second section I explain the research design and main research questions, followed by the sub-chapter on variables, data collection and choice of the appropriate model for ethnic characteristics and economic performance. Section 4 is dedicated to the findings using the proxies, using index scores, and the changes in public goods provision. Lastly, I argue main findings in the discussion part, followed by concluding remarks on how AKP government changed the practice of the discrimination.

1. Literature Review

Public sector provision of goods may often be confused with public goods provision. The difference in these two concepts is that public goods might be naturally provided, while public sector provision reflects the state provided goods and services, such as education, health, and infrastructure. The public-sector provision of goods exists to provide equal opportunities to all individuals, increase economic development, and consequently the standard of living of its population. In this paper, public goods provision, therefore, represents publicly provided goods, or funds for investments in health and education in the provinces of Turkey.

Economic development disparities are very common and are influenced by numerous factors. Easterly and Levine's (1997), article initiates that more ethnically diverse countries experience up to 2% less growth annually than countries with a homogenous population. Following the trend over decades, this negative relationship between diversity and growth explains the so-called growth tragedy troubling many developing countries (Wimmer, 2016). Nonetheless, Turkey's success starting in the 2000s, especially after AKP took power, shows a significant increase in economic growth. Real GDP grew substantially in the period 2002-2015, except for the financial crisis 2008-2010. During this period, growth averaged 5 % (IMF, 2016). On the other hand, Yeldan and Unuvar (2015) argue that the real growth was artificial and questionable due to the rising unemployment and declining labor force participation rates. In the past decade, the growth rate has been modest for an emerging market, with a 3.5% annual increase. Economists call it a middle-income trap, where countries that recently moved up from poverty are trying to move to the rich-countries club. Turkey struggles with branding and efficiency issues because they have difficulties with adding new value through research, design, and branding (The Economist, 2016). In addition, government administration has been increasing in number, where “de jure and de facto control of the ruling cadre intensified, amplifying corruption and arbitrary, unpredictable decision-making” (Acemoglu & Ucer, 2015). However, while uneven, Turkey experienced economic growth in all regions. The impact of growth on socioeconomic development is mostly seen through public goods provision. The biggest challenge for Turkey in this period is the improvement of human capital through investments in education and health.

Alesina, Baquir, and Easterley (1999) found that public goods provision is lower in more ethnically diverse countries. Additionally, they found that ethnically diverse communities have lower rates of tax collection and lower spending on service items. In their article, they explain two mechanisms which influence this relationship.

Firstly, individuals might not want to share public goods with other ethnicities, which Wimmer (2016), calls ethnic egotism. Unwillingness to share especially happens when the ethnic groups are not represented proportionally (per share in population) in the government. In that case, only represented ethnic groups would receive public goods. In the case of Turkey, it is hard to capture the actual representation of each ethnic group, since other ethnicities are not officially recognized. However, KONDA (2015), reports that in the elections of 2015, the pro-Kurdish Peoples' Democratic Party (HDP) had the highest share of votes compared to previous elections - 6%. They entered the parliament, winning 59 out of 550 seats (The Grand National Assembly of Turkey, 2016), around 10.7%. If we assume the Kurdish population to be 15 to 20% of the total population (KONDA, 2006), the conclusion is that Kurds are, indeed, underrepresented.

Secondly, preferences between different ethnic groups diverge, and different needs are very hard to satisfy, especially by a centralized government. Alesina, Baquir, and Easterley (1999) especially highlighted different languages as an important obstacle in public goods provision, which is the case with the Kurdish population and other minor ethnic groups. Additionally, Desmet, Ortuno-Ortin, and Weber (2009) show that redistribution by the government is lower in countries that have higher levels of linguistic diversity, which is a key indicator of cultural differences. Baldwin and Huber (2011) argue that economic differences between groups affect different group needs for public goods, feelings of alienation or discrimination by some groups, different attitudes toward redistribution across groups, and different “class” identities by different groups. However, empirical research on economic differences is limited. Alesina, Baqir, and Easterley (1999) argue that these differences in economic status lead to difficult decision making when it comes to which public goods should or should not be provided.

Article 42 of the Constitution of the Republic of Turkey prohibits educational institutions to teach any language other than Turkish as a mother tongue to Turkish citizens (Constitution of the Republic of Turkey, 2002). The Kurdish language, therefore, was banned from any public use since the early twentieth century. Additionally, only Jews, Greeks, and Armenians have been officially recognized as minorities, and any discussion about other languages and minority rights have been labeled as separatist propaganda (Human Rights Watch, 2002).By not changing this law, Turkey received a lot of criticism because they failed to recognize minority rights recommended by EU member states, the OSCE, and international human rights organizations. Nevertheless, the language obstacles and discrimination seem to be reducing with Erdogan on top. After almost 90 years of ignorance and exclusion of the language from public use, the government announced that the second most spoken language in Turkey, the Kurdish language, would be permitted as an elective course in universities, starting in 2009 (Bulbul, 2016).

A report on discrimination in the Turkish education system highlights the importance of education in its mother tongue and Turkey's ignorance towards certain paragraphs within international conventions. For example, they emphasize that Article 5 of the International Convention on the Elimination of All Forms of Racial Discrimination (CERD) stipulate the need to guarantee the equal rights to education to all citizens. However, Turkey does not recognize the authority of the Committee on the Elimination of Racial Discrimination to examine individual complaints. The same goes for Article 26 of the Universal Declaration of Human Rights (UDHR), article 13 of the International Covenant on Economic, Social and Cultural Rights (ICESCR), and article 28 of the Convention on the Rights of the Child (CRC) (UNICEF, n.d.).

Miguel and Gugerty (2005) analyzed public goods provision on the local level in Kenya and found that local ethnic diversity accounts for the sharp decrease in local school funding and the quality of school facilities in a sample of 84 schools. Moreover, ethnically diverse communities receive 20% less funding per pupil than schools in homogenous communities, and is 6 % less likely to have a functioning water well. This is an important finding since it directly highlights the issues related to economic growth and development through human capital. Authors argue that these findings can apply to all of Africa due to similar conditions (Miguel & Gugerty, 2005). Conversely, Egel (2012) found that tribal diversity is accompanied by higher levels of educational resources provided by the central government in Yemen. In Yemen's case, areas with more tribes had more teachers and more classrooms per capita. Polat (2012) similarly argues that higher ethnic heterogeneity in communities leads to more public goods in Jordan.

Baldwin and Huber (2011) in their cross-country study found that when economic inequality is along ethnic lines, it leads to different preferences for public goods and under-provision. Their work is in line with a study in 18 African countries where wealth inequality is high (Lieberman & McClendon, 2013), and it seems to be present when analyzing inequality between Turkish regions. The Income and Living Standards survey, conducted in 2008, in Turkey, was presented with the regional and national thresholds Using regional threshold refers to taking the mean of each region and comparing it with others. National threshold uses national mean, where regions are then compare to that score. Because of big disparities between western and eastern regions, national mean is deceiving because it undermines disparities between southeastern and western areas., due to large disparities between regions. Compared using the national threshold, Kurd-dominated Central-East, and South-East Anatolia's poverty rates have been 36.8 and 47.9 percentage points, respectively. Conversely, the Istanbul region's poverty rate is only three percentage points, estimated by the national threshold (Istanbul Notes, 2010). However, Waring and Bell (2013) note that for the negative effect of ethnic heterogeneity certain hierarchical stratification needs to be present.

Therefore, in this paper, I will examine the relationship between ethnic diversity and mechanisms of public goods delivery, on the subnational level in Turkey, where eastern and southeastern heterogeneous regions have poorer economic performance than ethnically homogenous Turkish provinces in the central and western Turkey.

1.1 Theories of diversity

Alesina and La Ferrara (2005) highlight the importance of the basics of social identity theory by Tajfel et. al. (1971). He argued that individuals attribute positive utility to the well-being of members of their group, and negative utility to that of members of other groups, describing the patterns of behavior in heterogenic societies. Each society possesses a set of assets and resources, such as land and mineral resources. Hence, to acquire more of these assets, individuals form coalitions, or groups to compete against the rest of the population. Moreover, once these resources are won, non-members are enforced to turn them over. Non-members might try to infiltrate in the organization to participate in the distribution of the goods. However this infiltration reduces the dividend each member would get. Caselli and Wilbur (2006) argue that in large communities of millions of citizens, it can be quite costly to track the members and non-members. Hence, in heterogeneous societies, coalitions are usually formed along ethnic lines, therefore, ethnic identity represents the indicator of the distinction between them. Additionally, Castelli and Wilbur argue that a stronger ethnic group has more benefits from a conflict over a country's assets, than it would be for equally strong groups in an ethnically homogeneous country.

Alesina and La Ferrara (2000) argue that individuals dislike mixing between ethnic groups and that the bias towards homogeneity leads to lower participation in community activities in ethnically diverse areas. Moreover, the effect is more visible when the costs of moving to the homogenous area are too high. Alesina et. al.'s (1999) work is in line with this theory. They argue that funding in heterogeneous communities is going to be lower due to a low willingness to compromise when it comes to ethnically argumentative public goods. However, their model did not explain why the provision of public goods that are not ethnically contentious should depend on ethnic diversity. Furthermore, the model showed significantly lower funding in US heterogeneous municipalities for non-ethnically dependent public goods, such as libraries, roads, and others.

Previous literature on tolerance and non-coethnic trust in Turkey showed a negative relationship between education and levels of support for Kurdish ethnopolitics (e.g. Sarigil (2010)). However, Livny's research on coethnic and non-coethnic trust showed that education is a strong predictor of inter-ethnic bias in Turkey. Thus, more educated individuals show more trust towards both, their coethnics, and non-coethnics. She also found the lowest degree of trust for both groups among individuals who did not graduate from high school, while they displayed significantly more trust towards coethnics (Livny, 2015). These findings suggest an important effect of education and human development to the levels of tolerance and peaceful cohabitation. Other of Livny's findings suggest that Turks have the lowest levels of trust towards non-coethnics than either the Kurds or Arabs. Additionally, they have the highest coethnic premium Coethnic trust premium is a comparison of the non-coethnic trust to coethnic trust. (Robinson (2013); Livny (2015)) . Therefore, while non-coethnic trust among Kurds is negative, it is significantly higher than that among Turks. When tested, more ethnically diverse places have higher non-coethnic trust levels, which is line with contact theory. Furthermore, ethnic fractionalization and coethnic trust premium have a significantly negative relationship (Livny, 2015).

Miguel and Gugerty (2005) argue that ethnic and kinship relations characterize rural African communities inside groups, but not across groups. Additionally, these relations regulate the access to resources and the control and redistributions of public goods. Fearon and Laitin's (1996) study, however, contrasts this case, where repeated interaction between individuals across groups creates space for interethnic cooperation. After the creation of the Republic of Turkey, the government persistently tried to break kinship and ethnic relations in the eastern provinces of Turkey. Centralization and nationalization created a hostile environment for the Kurdish population, and by individual resistance made space for other informal authorities to arise. Such authorities were either religious leaders or powerful Kurds - agas, who found themselves in control over resources, such as land and other public goods (Belge, 2011).

Alesina and LaFerrara (2005) state that potential costs of diversity, such as conflicts of preferences, racism, and prejudices, and the oppression of minorities can cause political instability, or in the worst case, a civil war. Yegen (1996) argues that by ignoring the fact that Kurds existed (the so-called nationalization of Turkey and ignorance of ethnic diversity) in the period of 1930-1980 caused a wave of dissatisfaction in the population. He adds that when dealing with the Kurdish question, government addressed it as “reactionary politics, tribal resistance, and regional backwardness”, instead of treating it as an ethno-political issue. The justification of what Yegen seems to address as misinterpretation, ignorance of the Kurdish question, or the language of exclusion of Kurdish identity, government linked to the process of assimilation and “state-building” (Belge, 2011). On the other hand, Yegen (1996) identifies the usage of terms `reactionary politics' to the religion of Islam, `tribal resistance' to traditional ways of living, and the `regional backwardness', as a peripheral economy. He argues that these concepts were deeply ingrained in the Kurdish identity, and by pressuring any of these terms, the government was “excluding the Kurdish identity.”

When the PKK arose in 1978, their leader Ocalan proclaimed the fight against the regime of elites serving the Turkish government, due to the lack of strategy and willingness to industrialize and separate Kurdish areas and create Kurdistan (Belge, 2011). More to the point, the center formed a special provincial guard to protect the local ruling elites, which PKK wanted to eliminate. Their manifesto required the complete elimination of local authorities and dedication to only the Kurdish identity. Massive killings started at the end of 1979, with PKK militants killing clan leaders, resulting in the death of 21 persons in the first three months of 1980 (Ozdemir, 1997).

Kuzu (2016) states that when the AKP started with reformist policies in 2004, PKK incidents caused by conflicts increased steadily. McCord explains the situation as follows: in the reform period, it is necessary for the ethnic insurgency to remind people that the government is still their enemy. Because of the PKK-caused conflicts, Kurds started feeling stigmatized, especially because other ethnic groups “tend to think that any Kurd who is proud of being Kurd would necessarily support the PKK” (Kuzu, 2016).

1.2 History of the Kurdish population

Up to the sixteenth century, Kurdish tribes were organized under Emirates, or principalities that enjoyed de facto independence. Then, due to the conflicts between Iranian empires and the Ottoman Empire (between Shia's and Sunni's respectively), Sunni Kurdish emirates were included in the Ottoman Empire, but enjoyed a great deal of autonomy. Since Kurdistan protected the Ottoman empire from Iran, elite Kurdish families were rewarded with power to govern and regulate the tribes, which led to the different development of governing the region and tribes from the standard Ottoman practice (Bruinessen, From Ottomanism to Separatism: The Religious and Ethnic Background of the Sheikh Said Rebellion, 1992). Because of a very high degree of autonomy from the Ottoman Empire, Kurdish emirates created a space to develop economic, political, and social identity and structures. Yegen (1996) states that these emirates in the nineteenth century had their own “polito-social space” where “Kurdishness” was established. In the first half of the nineteenth century, the Ottoman Empire restructured the system, and abolished the autonomy of Kurdish emirates. Later, they failed to stop tribal conflicts, which resulted in the creation of new social actors, called sheikhs. As Bruinessen and Yegen highlight, these sheikhs had both power and legitimacy in the Kurdish regions, which appointed governors lacked (Yegen, 1996). Additionally, rebellious act in the nineteenth century were supported by religious sects, like tarikats and tekiyyes, which supported sheikhs. By linking the state and religion - Sultanate and Caliphate, the bonds between the palace and Kurdistan were weak and only based on religion. However, in 1924, the Caliphate was abolished and that event represented the second strike (after the elimination of the Emirates), and tyranny over the Kurdish population by the Turkish state (Yegen, 1996).

The collapse of the Ottoman Empire caused the rise of three countries in the Middle-East: Turkey, Syria, and Iraq. Hence, the territory of Kurdistan was separated by the national borders between these three newly formed states. The previously well-working economy of Kurdistan had very important centers of non-domestic economic activities, such as Aleppo, Damascus, and Baghdad, whose links were now blocked by national boundaries. Hence, a once well-functioning legal system turned into an illegal one, where smuggling across borders to maintain the economy became an important activity to maintain good relationships with people on the other side. Yegen (1996) argues that such activities represented a significant challenge for Turkey's nation-building plans. He also notes that smuggling represented the fight against Turkification, instead of building the Kurdish national economy, which Yegen (1996) highlights as an important factor in centralization politics and its process in Turkey. He concluded that the reorganization of the empire and centralization of Republican Turkey were both ethnic-nationally driven. Hence, the oppression of the peripheral areas for the sake of nationalization led to the persistence of ethnic identities. Therefore, Yegen links smuggling and illegal activities between different parts of Kurdistan to the ethnic identity building. Furthermore, a culture of Islam, traditional tribal structures, illicit economic activities and an ignorance towards taxation and military service became the main characteristics of what Kurdishness really is. Similarly, Gao (2016) argues that nation-building strategies in Jordan created a climate of a weak accommodationist state where public goods provision is centralized, but leaving certain autonomy for tribal identities to take over on a local level.

1.2.1 Characteristics of the Kurdish population

Kaya describes Zaza Kurds as a tribal society characterized by significant social disparities between “prominent village leaders” - tribe members, and the majority, consisting of small farmers and landless tenants. Division of land and other public goods depends on the village heads, who delegate the land to farmers. The quality of life is considered very low due to many factors. Most people are poor and undernourished, and their houses are built in the medieval style, lacking running water, and, often, toilets, and women do laundry on the riverbanks. Their life expectancy is very low, around 50 years at birth, and the birth rates are high, with 5 to 8 children per nuclear family. Mortality rates are high due to several factors. Firstly, villages do not have medical facilities. For example, Kaya mentions Solhan district where there is a hospital, but people do not get adequate treatment because of high prices or the lack of expertise by the local medical staff. Furthermore, their perception of Islam is that God determines one's fate, and such cannot be changed by a man. Hence, sick people sometimes do not demand healthcare. Kaya (2011) argues that no effort has been made to educate the population on the use of medical care. Thus the development of this area is very low. Additionally, he claims that most of the population never attended school or are illiterate. Only one percent of the population has educational attainment higher than the primary level in the Solhan district. Due to these characteristics, they base their decisions on their experience, which is in contrary with the modern way of governance. Most important aspects of social life are related to kinship relations. Thus, economic, social, political, and religious contexts are not separated, but looked at collectively, through biological relationships. Zaza society, like most other Islamic societies, is organized through patrilineal relations, and the paternal side of the family is more important than the maternal. Through the paternal side, one obtains rights and obligations (Kaya M. S., 2011).

1.3 Politics of Multiculturalism, changes, and obstacles

Kuzu (2016) highlights that in the last two decades, state policies towards the Kurdish question have slowly begun to change. However, the weak multiculturalists policies failed to adapt to the needs of the population, which would include the provision of state-funded education in the Kurdish language. Previous studies usually relate these failures to the Islamist AKP government's incoherent and reluctant approach towards the language issue (Zeydanliouglu, 2013) and the constant opposition of the MHP (Nationalist Movement Party) and the Kemalist CHP (Republican People's Party) (Gunes, 2013) (Kuzu, 2016). Additionally, Kyumlicka and Opalski argue that on the state-building nationalist policies ethnic minorities in the country usually responded by seeking greater autonomy to “engage in their competing nation-building, to protect their societal culture throughout their traditional territory” (Kymlicka & Opalski, 2001). However, Kuzu (2016) argues that Kurds diverge in the views of what they mean by constitutional recognition and autonomy solution in the first place. Therefore, the take-up of the goods provided by the central government might be lower simply because of the failed multiculturalism policies and the opposite reactions they might create when it comes to the oppressed minorities, such as the Kurds. On the other hand, because of the divergence in the views on the recognition and autonomy mentioned above, the lack of collective action might not decrease the take-up.

2. Methodology

In this chapter I discuss the process of research, main research questions, and hypothesis which we will test. Additionally, I will explain the method of data collection, the variables used in regressions, and the within-between model, or the hybrid model for panel data analysis.

2.1 Research design

The goal of this study is to understand the empirical relationship between ethnic diversity and public goods provision on sub-national levels in Turkey, and the efficiency of that provision. During the perdiod of 1990-2015, there has been a shift from non-economic growth (or slow economic growth) before AKP, to high economic growth after AKP won key positions. The government's position towards the Kurdish question and ethnic minorities' recognition has not changed. However, the central government did focus on investments in human capital. Consequently, southeastern and eastern areas of Turkey, majorly populated by Kurdish tribes, still have obvious disparities in living standards, in comparison with homogenous Turkish regions.

While all regions had a similar increase in their regional product after AKPs victory, did economic growth influence the better provision of public goods to all parts evenly? To understand the process of the delegation of public goods, I will focus on the major factors that can affect the funds delegated to each province. One factor can be economic, which means that higher economic performance of the provinces leads to higher public goods provision. The second factor, ethnic characteristics of the provinces, theoretically would result in a lower provision for provinces with a higher concentration of Kurdish population, meaning provinces with the bigger share of the Kurdish population would have lower state provided goods concentration. Additionally, I will test these two factors on the changes in public goods provision, which would account for the previous disparities that might affect results. If historically Kurdish provinces had lower public funding from the central government, then changes between the current provision and previous one would be neutralized, showing only the improvements in each province. Using the statistical software Stata, I will conduct a quantitative analysis on panel data with 81 provinces, and the observed period of 6 years (2009-2014).

2.2 Research questions and hypothesis

Therefore, the research questions are as follows:

1. Was the delegation of public goods from the government ethnically, or economically driven?

2. Did Kurdish provinces have different treatment in public goods provision during the AKP mandate?

Based on the research question regarding the factor of delegation of public goods, the main hypothesis is as follows:

H0: Improvement in prospects (public goods provision), is significantly correlated with the ethnic composition in each province. That means Kurdish provinces profited less than the other ones during the AKP mandate.

HA: Improvement in prospects (public goods provision), is not correlated with the ethnic composition in each province. Education and health provision was correlated with economic performances.

The assumption is that the funds from the government are given and used for the specific purpose of stimulating the take-up from the population (more classrooms and free books increase enrollment rates). Therefore, this study does not assume a possible corruption and any misuse of resources transferred to provinces.

2.3 Variables

There is no consensus in the empirical research on how public goods provision should be measured. Federal expenditures and transfers from central government to provinces would be an ideal measure to see on what basis the state delegates these funds, however, it was not possible to gather such data. Following the work of Wimmer (2016), Gennaioli and Rainer (2007), and La Potra et.al. (1999), I used the proxy variables to measure the public goods provision in these two sectors. Firstly, in countries where the population relies on public school systems, such as Turkey, adult illiteracy rates capture the supply of public goods and alphabetization projects (Wimmer, 2016) government might undertake to decrease illiteracy. In addition to the illiteracy rates, other education levels an individual might obtain, such as Bachelor's degree, and Master's diploma can help in explaining the mechanisms of public goods delegation. Because of the lower socioeconomic development scores in the Kurdish provinces, we can assume that the share of the highly educated population will be lower than in homogenous Turkish provinces. Additionally, we can test the variables such as the schools per capita per province, and the teaching staff per capita. These variables capture the direct provision of financial funds for education purposes.

Regarding proxies for public provision in the health sector, infant mortality rates are a commonly used variable. Even though child mortality rates might be affected by the disease prevalence, climate, and other geographical factors, Gennaioli and Rainer (2007), La Potra et.al. (1999) and Wimmer (2016) highlight the impact of basic healthcare provision and government-run immunization projects on infant mortality rates. However, infant mortality rates are not the only proxy for public health provision. Others include the number of medical facilities and medical staff, and the number of hospital beds per thousand people. These proxies measure the direct impact of the funds delegated to each province, where on a per capita level, there should be no significant disparities.

The next chapter explains the process of data collection, variable selection, and an explanation of the model. Following that, I detail the results of the regression, analyze the results, and discuss the findings and limitations of this research.

2.4 Data collection and analysis

As previously mentioned, the Turkish Republic does not recognize any ethnic minorities, therefore assuming the population of Turkey is composed only of the Turkish ethnic group. Questions related to the religious denomination and ethnicity have been left out from the national census since 1965. Since that was almost five decades ago, it would not be wise to use 1965 ethnicity shares. Additionally, government officials demanded that similar questions be excluded from large-scale surveys, such as the Demographic and Health Survey (DHS), and the World Values Survey (WVS). Fortunately, there are methods to estimate the ethnic structure of each province.

Livny (2015) used KONDA's dataset from 2006 on the shares of the Kurdish population in each province. Unfortunately, it was not possible to get that dataset for this research. Instead, DHS from 1993, 1998, and 2003 included a question related to the mother language. Following the works of Эзduygu, Romano, and Sirkeci (1999), and Koз, Hancioрlu and Cavlin (2008), the mother tongue variable seems to be the best alternative for the ethnic shares. All three surveys show almost exact shares for each province, and since there has not been a significant demographic shift in the past decade, I decided to proceed with the analysis using the data from the DHS 2003, except for the Yalova province, for which the data was available from the DHS 1998. These shares, are, therefore, time-invariant and identical for each observed year in the analysis. The map below shows the shares of the Kurdish population in a total population of each province, based on the DHS 2004 mother language variable.

Picture 1: Data source: Demographic and Health Survey 2004 Turkey, Mother tongue variable

One of the major limitations of this research is that the ethnic structure is captured by the mother language variable from the DHS. Nevertheless, it represents the best alternative to the actual ethnic structure. The province with the highest proportion of Kurdish population is Agri with 98.3 percent, followed by Sirnak, and Diyarbakir with more than 90 percent of Kurdish people in the total population. Central and northwestern regions of Turkey have a mostly homogenous Turkish population structure.

2.4.2 Economic performance of the Turkish provinces

Economic factors that might affect government delegation of public goods are labor force indicators and gross regional product. Data for this dimension is taken from the TURKSTAT labor force statistics and national accounts dataset (TURKSTAT, 2016). Intuitively, when gross regional product increases, the province is richer, and therefore public goods provision should be higher if the delegation of funds reflected economic activity. Additionally, unemployment rates should have a negative correlation with public goods provision. The reason both GRP and unemployment rates need to be included in the analysis is the previously mentioned shadow economy, specifically, in Kurdish provinces. If smuggling and trading with Kurdish areas in another country (such as Syria and Iran) exist, unemployment rates will not represent the real picture of the economic activity. However, they are an important factor in the economic performance.

According to the Turkish National Statistical Institute (2015), the highest unemployment rate was in the sub-region Mardin, including the provinces Mardin, Batman, Юэrnak and Siirt, with 24%. All the above-mentioned provinces are majorly Kurdish populated. The lowest unemployment rate was in the sub-region Agri, including Agri, Kars, Igdэr and Ardahan provinces, with 3.4%. Thus, the highest and lowest unemployment rates are in the sub-regions characterized by a high concentration of Kurdish population.

The gross regional product is a measurement of economic growth of territorial units inside a country. A perfect indicator of the economic performance would be the economic sectorization, or the number of people employed in the agriculture, industry, and services sectors. Unfortunately, data on economic sectorization on the province level does not exist, or it has not been available for academic use. Instead, I used data on the actual GRP value in Turkish Liras (TRY), Base year - 2009 and recalculated the shares of each sector in total GRP. These two variables provide different information. Firstly, using the GRP value, we can see the absolute change. The highest agricultural product value was in the province Konya, which is a province rich with arable land and minerals. It has been called one of the Anatolian Tigers because it experienced fast economic growth in the 1980s. Besides being the biggest agricultural producer, the city of Konya is also industrially developed. On the other side, Karabuk province has the smallest agricultural product value. Karabuk is in the northern-central area of Turkey, and it is one of the newest provinces. Before 1995, it was a district of the province Zonguldak. In the industrial and service sectors, the highest value was in the Istanbul province, and the lowest province, Bayburt, which is in north-eastern Turkey. Like Karabuk, it just recently became a province, after being a district of Gumushane.

Secondly, the share of agriculture, industry, and services in total GRP will provide an insight into how regions perform and whether they are agricultural, industrial or service centers. The highest share of agriculture in total GRP belongs to the province Ardahan, followed by Karaman and Sanliurfa, with the lowest in Istanbul, Kocaeli, and Ankara. The local economy of Ardahan depends on farming and raising livestock, but the construction of the Baku-Tbilisi-Ceyhan pipeline gave the local economy a brief boost from 2000 onwards. The UNDP and Turkish government sponsored Ardahan-Kars-Artvin Development Project is expected to boost agricultural, livestock and horticultural production in the easternmost provinces of Turkey, where the focus is to alleviate rural poverty and improve the livelihoods of poor rural people living in these provinces (IFAD, 2017). The project was carried out in the period of 2009-2016. One might argue that due to the agrarian characteristics of lower developed societies and the tribal way of life of the Kurdish population (Kaya M. S., 2011), the share of agriculture in total GRP would be higher for Kurdish provinces. However, out of the top 10 provinces with the highest share of agricultural product in 2014, only five are majorly Kurdish populated. The other five, are homogenous Turkish provinces, and according to DHS 2003, they have no Kurdish population. These provinces are Karaman, Nigde, Bayburt, Yozgat and Afyon.

Correlation between gross regional product components and the Kurdishness of a province is low, therefore we can safely include them in the regression without the fear of colinearity. However, industry and services sectors are strongly correlated, and therefore I will have to test them in separate regressions. Correlation table is in the Appendix 1

2.4.3 Measuring Public Funding for Education

Following the research studies of Livny (2015), and Gerring et.al (2015), I obtained data from the Turkish National Statistical Institute on the provincial level for highest educational attainment (TURKSTAT, 2016).The dataset includes nine educational levels ranging from illiterate, literate without a diploma, to doctorate level from the address based population registry. Variables are presented as the proportion of each educational attainment, in a total population of the province, in the period 2009-2014.

Graph no.1 plots the mean illiteracy rate in the observed period. In total, illiteracy rates had a sharp decrease from 2009 to 2012, meaning that the higher economic growth positively affected literacy of the population.

Graph 1 Average Illiteracy Rate

Data source: TURKSTAT (2016)

To see the relationship between changes in the public goods provision with the ethnic characteristics, and economic performances, I used the share of illiterate population for the year 2000, from the population census in that year, downloaded from TURKSTAT website. Illiteracy rates in 2000 are subtracted from each year in the observed period 2009-2014. This way, we are not accounting for the historical disparities in the public goods between provinces, but focusing only on changes in illiteracy levels, and how AKP- run government delegated funds for education.

When talking about high education, variables share of Bachelor degree holders, Master's, and Doctorate diploma, have a negative correlation with the proportion of the Kurdish population. Graph 2 shows the relationship between the proportion of Bachelor diploma holders, and the share of the Kurdish population in the province, for the year 2010. In Ankara lives the highest percentage of people holding Bachelor's degree, because Ankara has the highest number of universities. The lowest share of Bachelor's diploma holders is in the province Agri, with the 98% of the Kurdish population in the total population.

Graph 2 Shares of Bachelor's degree holders and the share of Kurdish population Data source: TURKSTAT (2016)

As previously mentioned, I collected data on the number of teachers and schools per province from TURKSTAT (2016). This dataset includes the number of schools and teachers for education levels: pre-primary, primary school, general, and vocational secondary education, and the number of employed teachers in each. By using actual per capita number of schools and teachers, we account for the direct provision of public funds from the government. However, there are many other effects which can influence the number of educational institutions in a province. One of them might be the lower interest of the population in sending children to school. This might especially be the case in rural areas of Kurdish provinces because students can only learn in the Turkish language. Therefore, using the Exploratory factor analysis (EFA), it will be possible to capture the latent variable - delegation of funds to provinces, which affects all previously mentioned education variables.

2.4.4 Measuring Public Funding for Health sector

All variables capturing the public goods provision in health are collected from the TURKSTAT (2016) database. The first set of variables in the health statistics dataset includes the number of hospitals, number of hospital beds, and the medical staff including physicians and nurses, per thousand people. These variables will have a similar relationship with the public goods provision, as the number of schools and teachers.

However, the correlation between hospitals per capita and the share of Kurdish population is low -0.23. Hospital beds and the number of physicians have a higher negative correlation with the proportion of Kurdish people of -0.40, and -0.37, respectfully.

Lastly, a number of doctors and nurses per capita reflect the provision of the medical staff in health institutions. The size of the nursing staff is especially interesting variable because in general, the nurses are more volatile. The logic is that when there's a lack of financial support, cost reducing measures usually mean firing nursing staff. Therefore, cuts in the number of nurses per capita will reflect the provision of public goods in health because they are likely to vary more. The mean number of physicians per capita in 2011 in Turkey was 1.71. (World Bank, 2017) An interesting fact about the number of doctors is that homogenous Turkish provinces have more than one physician on a thousand people, but several provinces with the significant share of Kurdish population do go below 1. Graph 3 plots the average number of hospitals per 100,000 people.