Government patronage in Iran and Turkey. "Public employment: the engine of female empowerment?"

Master Thesis: Government patronage in Iran

and Turkey

“Public Employment: The Engine of Female Empowerment?”

Name: Robbert Rademakers

Student number: S4184599

Study: Master Economics

ABSTRACT

The feminization of public employment in the Middle-East has offered substantial

opportunities for women in order to become economically empowered. However, only a

limited group of women seems to have profited from this access to public employment.

Additionally, public employment can make these women economically depended on their

governments for their autonomy if the private sector does offer a viable alternative. This

research examines these limitations of the public sector of public employment in Iran and

Turkey by using household survey data (TDHS, IPUMS, HEIS). Iran and Turkey represents two

contrasting systems of female employment in the Middle-East. The former is a patronage

system with the majority of women working for the public sector, the later a capitalist with a

well-developed private sector. This research shows that public employment is a privilege for

highly educated women who are part of the inside network (called Wasta). Furthermore, the

harsh working conditions in the private sector make the dual responsibility of work and

motherhood of many women incompatible. With regard to her economic autonomy, human

capital seems to be an important factor for women to remain independent of government

employment for her process of empowerment.

C

ONTENTS

I.

Acknowledgements...1

II.

Introduction...2

III.

Working women in Turkey and Iran...8

IV.

Literature Overview and Theoretical framework...14

V.

Data & Methods...22

VA. Model...22

VB. Data...26

VC. Description of variables...29

VI.

Results...33

VIA. Binary statistics...33

VB. Regression...37

VII.

Conclusion and Discussion...45

Appendix A1: Regional Characteristics Iran...49

Appendix A2: Regional Characteristics Turkey...50

Appendix B: Model 3 (gender hierarchy and unemployment interaction)...52

I.

A

CKNOWLEDGEMENTS

Nijmegen, 23-10-2016

After six months of dedication, I can write this thank note as the rounding part of my master

thesis. I would like to express my gratitude to several people in particular. With their support,

I was able to personally develop myself as researcher and reach my potential as master

student. First of all, I would like to thank Zahra Mila Elmi, assistant professor at the university

of Mazandaran, for her personal involvement in my project. She was an excellent guide

during my field trip in Iran and welcomed me as a true friend. Furthermore, I am grateful to

Ayşe Gündüz Hoşgör

professor at ÖDTÜ. Our dialogues provided me with unique insights and

fresh ideas. I would like to thank Rob Arts for his language advice, mental support, and above

all his constructive feedback sessions. Finally, I am highly indebted to Jeroen Smits, my daily

supervisor. During our many discussions he stimulated me to develop two of the most

valuable traits for a future academic career: remain critical and stay creative.

II.

I

NTRODUCTION

Disproportionately, women have not been included in the ‘Wealth of Nations’ in contrast to their masculine counterparts. Their relative lack of economic empowerment is considered as a fundamental cause of their marginalized position, withholding them the opportunity to gain autonomy and escape poverty (Swedish Ministry for Foreign Affairs, 2010; OECD, 2012a; OECD, 2012b UN WOMEN, 2013). In order to reverse this unfortunate destiny, the OECD stresses the dual necessity for women to both “participate in the economic process” and “increase their access to economic resources”, which can be attained by entering the labor force (OECD 2012b; Eyben et al., 2008). However, labor force participation should satisfy certain criteria to be used as mean of empowerment. To illustrate, women employed in the agricultural or manufacturing sector are proletariazed as source of cheap labor (Moghadam, 1988). The ‘Modern Economy’ described by Boserup (1970), where female employment is predominantly located in the service industry, seems the way forward for these women chained in marginalization.

Improving the low levels of female labor force participation could offer great opportunities for the women concerned, their households, and countries in general. At an individual level, labor force participation contributes to economic independence of women by improving her health conditions (Gomis-Porqueras et al., 2011) and status (Jensen, 2012). Moreover, employment lowers the reproduction rate, thereby the women prefers the quality of her offspring over the quantity (Lim, 2002). In the Turkish context, female employment has been found to substantially improve the bargaining position of women within the household (Worldbank, 2009a). This increased bargaining position could have spillover effects on the welfare of the women’s children, since a woman is considered to be more altruistic towards her children compared to her husband (Vyrastekova et al. 2014). To illustrate, authors have found a positive link between the working status of women and the nutritional health (Kumar, 1994) and educational status of their children (Afridi et al. 2016). In developing countries, labor force participation of women could expand the monetary position of households with 25% additional income (Worldbank, 2003). At a macro scale, equalized labor force participation enhances the state of the economy by creating substantial efficiency gains (Woetzel et al., 2015). The Worldbank (2009b) concluded that as much as 15% of poverty could be recuded in Turkey if the proposed target of 29% female labor force participation is reached. Consequentially, developing organizations urge nations to invest in female labor force participation, as is defined by the United Nations in the Beijing declarations (1995):

‘Promote women's economic rights and independence, including access to employment’

Beijing Declaration (strategic objective F1)

For Middle-Eastern countries, the importance of female labor force participation requires additional attention. Declining fertility and mortality rates create a large number of potentially productive female employees. A considerable amount of these women have moved with their families to the metropolitan regions searching for well-paid jobs as is demonstrated during the last decades in Tehran, Istanbul, Cairo and Tunis. As can be seen in table 1, the dependency ratio of Iran is one of the lowest in the world indicating that there are more than two people of productive age for every care-needing person. This ‘Demographic Dividend’ can act as a two-edged sword if the right policy contexts are not in place in order to absorb marginalized groups during this process of growth (Bloom & Finlay, 2009). Joffe (2011) argues that the mismatch between the demographic transition and the economic opportunities, has contributed to the Arab Spring and instability in the region. Marginalized women, being substantially penalized, were one of the leading protest groups in this period of tumult. Country/region Year 1990 2013 Iran 95 39.6 Turkey 68.2 50.3 Arab 87.7 61.7 Upper-Middle Income 58.5 41.6

Table 1: Dependency ratio’s comparison; care requiring population (0-14 & 65>) divided by working age population (15-64) (United Nations’ World Population Prospects, 2016)

In the Middle-East, governments used to be oasis of employment opportunity for women. Moghadams has called this process in MENA: ‘the feminization of government employment’ (p. 126, 2005). The public authorities erected themselves as benefactor of women creating a special and dependent patronage relationship between the government and concerned women. During the Nasar-Sadat-period (1956-1981) public employment was explicitly guaranteed for both men and women in Egypt with the only precondition of a completed secondary diploma (Assaad, 1997). This policy had as result that female labor force participation increased to 21,4% in Egypt in 1984 (Moghadam, 1993) of which more than 65% was absorbed by the public sector (CAPMAS, 1986). The socialist Ba’ath party in Syria and secularists in Tunisia introduced comparable policies to promote the economic inclusion of women (Moghadam, 1993). In Turkey, the Kemalist governments between

1920 and 1980 applied an equivalent statistic approach to the employment issue (Celasun & Roderik, 1989).

The private sector considers women as secondary workers (Anker, 1998; Tansel, 2001) and is therefore willing to pay a premium to hire male employees. Women are not only confronted with gender-based discrimination since they are less likely to get hired (Moghadam, 1993) and promoted (Davison & Burke, 2000) but as well because they receive substantially lower wages compared to their male counterparts (Christofides 2002). Governments as public and considerable institutions have less ability to discriminate based on gender (Long, 1975). First of all, because governments, are being checked by the general public and held accountable. Secondly, as sizable institutions they need efficient and aligned human resource policies. To illustrate, in Turkey the selection procedure for civil servants is organized by a central exam (KPSS). Accordingly, most Middle-Eastern states denounce gender discrimination constitutionally (Kanani, Jorfi, 2014), and legally bind themselves to gender quality and incorporate female employment as integrated part of their development plans (e.g. Ninth Development plan of Turkey & Fifth development plan of Iran). This is complementary with the finding that after adjusting for the difference in individual endowments, women and men face a comparable return if they are employed by the government, however, in the private sector the return to both education and experience is significantly lower for women (Tansel, 2004). Also from a supply-side perspective, women in the Middle-East prefer to be employed in the public sector because it offers increased status, security and superior benefits (Barsoum et al., 2014). Furthermore, the working conditions in the private sector, indicated by the relative long working hours, are another limitation of the private sector making it hard to integrate a private sector job with her dual role as a mother (SYPE, 2009).

Apart from the positive aspects of public employment for women, the overreliance on government employment for women to achieve economic empowerment has two major drawbacks. Firstly, governments, especially in market economies, are restricted by their budgets. This makes public employment a restricted tool to incorporate women in the economic arena. Generally, only a favored group of women can profit from these economic opportunities. For example, in Egypt the access to public employment was restricted to the highly educated thereby leaving the most vulnerable group excluded (Assaad, 1999). Similar, exclusively the secular upper-middle class in Tehran profited from the employment policies of Mohammad Reza during his reign as Shah (king) starting from the British intervention in 1941 until the Iranian revolution in 1979 (Moghadam, 1993). Being part of the internal privileged network is of substantial importance to get access to these privileged resources. In a recent Gallup poll, more than 90% of the people living in the Middle-East believe ‘wasta’ is needed to get a scarce government job. In the Arab world this term ‘wasta’ (ةط ِسا َو) is used for group-favored

networks (Barnett et al. 2013) and generally considered to be a form of corruption. Secondly, if public employment is not accompanied by a joint development in the private sector, women could become dependent on their governments and the benevolence of politicians (Dahlerup, 1987). Moghadam describes this public oriented employment system as a mere transformation from patriarchy to patronage, where not the husband but the government is the economic lifeline of the woman. This is contradictory to the notion of empowerment where autonomy and self-determination are important input factor (OECD, 2012).

‘Throughout, the Middle East, the active role of the government in national development has meant that many women no longer rely on a male guardian as provider, but rather the state ‘

(Moghadam, 1993 p. 62). The negative consequences of patronage become visible during periods of government distress, as could be the case during budget cuts and structural adjustment programs. The jobs and wages of civil servants are an easy target to balance the government (Türel, 2001). Egypt implemented structural adjustment programs during the early 90s of the last century, thereby adhering to the privatization and deregulation credo of the Washington consensus. As a result public employment and wages were substantially reduced (Assaad, 1997). Women were unable to mobilize themselves to the (underdeveloped) private sector (Assaad & Arntz, 2005). Their only viable alternative was to re-embrace patriarchy (Bennholdt-Thomsen et al, 1988; UN Women, 2013). Similar trends have been experienced in Turkey after the government restructuring of the 80s (Pacyzynska, 2013). Furthermore, if women are dependent on patronage they can become play-balls of politics. The Rouhani administration recently installed a quota stating that only 10% of the ministerial employees are allowed to be female (IPI, 2015). This policy was a compensation, under pressure of the religious leader of Iran, Ayatollah Khamenei, for the prior implemented ‘liberal’ policies of the government In order to examine the consequences of patronage two distinct points are of particular interest. First, it is needed to examine the kind of women that are in particular attracted to respectively private and public employment. This could teach us which groups are depending on, and above all, are profiting from government opportunities for their economic empowerment. Secondly, by revealing the factors that make women prefer public over private employment, we can examine the working conditions that need to be improved to make private employment complementary. For example, if mothers are more inclined to accept public jobs, the private sector should be adapted in such a way that women can more easilily combine their productive and reproductive roles. Currently, Turkey and Iran represent two contrasting exemplarily systems for the region both shaping the quality and quantity of employment opportunities. Turkey, as a result of the structural adjustment programs

implemented in the 1990s, adopted a western model with a primary role for the private sector as provider of female employment. In 2013 only 15% of the working Turkish women have found formal employment at a government agency, the other 85% are active in the private sector (TDHS, 2013). In contrast, Iran can be described as a modern patronage system where more than 50% of the working women depend on government employment. By comparing the effect of two above-mentioned factors in Iran and Turkey, we can evaluate the advantages and vulnerabilities of both systems. Therefore, the first research question is formulated as followed:

Research Question 1:

“What are the socio-economic characteristics of female civil servants in Turkey and Iran and to what extend do these characteristics differ between both countries?”

As mentioned earlier, patronage is in particular problematic to attain empowerment for women when women are unable to mobilize themselves to the private sector in times of distress. This ability to relocate towards the private sector in the MENA region is not only affected by cultural factors, created by gender inequalities and hierarchies, but often the simple quantity demanded of labor in the private sector is lacking. Youth unemployment is with 25% the highest in the MENA region. For women this figure is even substantially higher at 30% (ILO, 2013). Some countries offer disastrous prospective for young women with unemployment rates around 65% in Egypt, 40% in Iran and 50% in Jordan. (Mottaghi et al, 2014). The causes for these depressed figures are diverse. Government mismanaging seems to be an important determinant for the underdevelopment of the private sector (Schwedler, 2013). The excessive demand of employment by the government crowds out the ability of the private sector to develop and become competitive (Assaad, 1997 Assaad & Arntz, 2005; IMF, 2013). Additionally, many resource rich nations in the Middle-East often lack the incentive to undertake painful reform programs (Shambayati, 1994). The revenue of these governments is not directly linked to the industrial competitiveness and general state of the economy of the country but rather the abundance of natural resources. In constrast, public employment is often used as political measure to stabilize regions with high unemployment rates. In 2010 the Turkish government started a massive investment of 32 billion dollars to pull the Kurdish region (South East Anatiolia ‘Güneydoğu Anadolu’) out of its misery with public projects in energy, irrigation and education. This project alone should create three million new public jobs for Turks living in the Eastern part of the country (Özhan, 2008). Therefore, the mentioned patronage relationship is amplified in cultural and economic rigid environments, where there is a lack of female employment by the private sector. In these ‘patronage’ counries without well functioning private sector women have no alternative than to work for the government. Authors concerned with the economic empowerment of women established the

educational dominance hypothesis (Spierings et al. 2010). Human capital is thought to be an important factor to increase autonomy in the most rigid environments. For highly educated women, who are more productive and better armed against gender hierarchies, there could still be opportunities in the private sector, as they are to a lesser extend considered as secondary workers. This insight leads to our second research question.

Research Question 2:

“Does human capital ensure employment autonomy, respectively in Iran and Turkey, in economic and cultural situations where the private sector has the least to offer?”

In the following chapter (II), Turkey and Iran are compared and contrasted. It will be demonstrated that the public and private employment system differ significantly between both nations. This explicit comparison could therefore lead to valuable new insights. In chapter III an extended theoretical framework with deduced hypothesis is presented to examine the decision of women to work for a public or private employer. Thereafter, in chapter IV, methodological considerations are discussed. Finally, the results are presented and discussed in respectively chapter V and VI.

This research could add significant insights to the existing literature in three ways. Firstly, female labor force participation has received much attention in the literature in both Turkey (Gündüz-Hoçgar & Smits, 2008; Worldbank, 2009; Dayıoğlu & Kırdar, 2010) and Iran (Elmi & Noroozi 2007; Majbouri, 2015, 2016), but the public-private relation has been neglected and only been considered in the light of wage differentials in both sectors (Chrisofides, 2013; Blank, 1985; Assaad 1997; Tansel, 1999 2004; Jovanovic and Lokshin, 2004) or in ethnic labor markets (Moshe and Semyonov, 1994; Eisinger, 1986). This study will focus specifically on the mechanisms that determine whether women end up as employee in the public or private sector, including both economic and cultural factors. Secondly, a direct comparison between two countries has scarcely been applied in this field of research. Even though, a comparison between Turkey and Iran is not uncommon in the economic literature (Salehi-Isfahani et al., 2008; Shambayati, 1994. Contrasting both countries could give us insights in the effect of the national institutions and conditions, such as, for example, the strict secular laws in Turkey compared to the Iranian theocracy. Finally, the overreliance on government patronage has been discussed in the literature by authors like Moghadam in the Middle-East (1993) and Assaad in Egypt (1997, 2004), however a potential solution to increase economic autonomy has never been examined.

III.

W

ORKING

WOMEN

IN

T

URKEY

AND

I

RAN

Historically, Iran and Turkey have followed comparable development trajectories into the modern era. As great foregone empires, both have never been colonized by western powers. This is in contrast to most countries in the MENA region. Many of these nations were either directly colonized by European nation states (Egypt, Tunis) or included in the Sykes-Pictor agreement (Iraq, Syria). Two illustrious leaders, Attaturk (1923) in Turkey, and Reza Shah in Iran (1925), the father of the notorious last Shah of Iran (Mohammed Reza), seized power at the beginning of the 20th _{century. The major}

objective of their reigns was to modernize their countries both culturally and economically. Both countries applied a similar statist approach, where the state takes a central role in the process of change, to achieve their goals. The Iranians were often inspired by the example of their Turkish neighbor, noticeably with composing their constitution, in which (gender) equality is a core component (Sohrabi, 2011). Nevertheless, Iranian women are still economically excluded, compared to nations in the same cultural region (Arab league), as we can see in table 2 with only 18.3% economically active women in the formal economy. This leads to logical observation that the other 81.7% of the Iranian women is either housewife or included in the informal shadow economy. The figure of female labor force participation in Turkey is substantially higher compared to Iran, however, still below its potential as upper middle-income country.

Puzzling is the fact that women in both nations seem to have received enough means of empowerment as can be concluded from the high education attainment levels and low fertility numbers in table 2. Human capital is considered as main driver of economic empowerment for women in the Middle-East (Tansel 2002; Spierings et al. 2010). Furthermore, the introduction of the nuclear family structures in Iran and Turkey offers the opportunity to women to shift her reproductive responsibility towards productive means (Moghadam, 1993). Finally, at the demand side of labor, there is also no shortage of female friendly jobs that are being located in the service sector as we can conclude from table 2 (Iran: 49; Turkey: 50%) (Boserup, 1970). These deprived figures of female labor force participation indicate the multitude and variety of economic and cultural hurdles that have to be taken by the women in both countries in order to achieve economic empowerment.

Apart from the similarity that Iran and Turkey are facing low levels of female labor force participation, the source of female employment started to diverge since the 50s of the former century and amplified in the 1980s as a direct result of the overall change in economic structure in both countries.

Iran was able to set up a government-depending line of employment, labelled as patronage, as a direct result of its rich stock of national resources. It developed itself into a rentier state, which is not uncommon in the Middle East (Shambayati, 1994) similar to Saudi-Arabia among others. Turkey, underprivileged with a scarce stock of natural resources, developed a competitive and international oriented private sector, under pressure from foreign institutions (IMF, Worldbank). In the Islamic Republic of Iran, the government has become the engine for women’s employment opportunity as 41.91% of the women works for a public employer. For formal white collar employment this figures is even considerably higher and close to 60% of total tertiary employment (HEIS, 2012). Turkish women relied on the private sector like their male counter parts for employment with relatively 13% and 15% public employment. In this chapter we will place these employment and economic systems of Iran and Turkey in their historic and economic context.

Country/Region

Iran Turkey Arab League

Upper- middle Income countries Employment position

% Women working in public sector 41.91% 12.9% % Working women in service sector publicly

employed

60% 25%

Female labor force participation (2014) 18.3% 30.5% 22.3% 42.5% Means of Empowerment

Gross enrollment female secondary education 86% 113% 69% 93%

Fertility per woman (2014) 1.7 2.1 3.4 1.8

Tertiary Jobs (%) (2010) 49% 50% NA 52%

Table 2: Turkey and Iran in perspective (Retreived from Worldbank, 2016)

Turkey: A capitalist model

The establishment of the secular Turkish republic in 1923 by Kemal Attatürk introduced a new chapter in Turkish history after centuries of regional hegemony during the Ottoman-era (1299-1923). Partly prompted by its nationalist ideology, but predominantly as a result of economic necessity, the new government adopted an import substitution strategy untill the 1970s to promote its manufacturing industries and reduce its foreign dependency on imports. Jobs in the capital intensive import substation sector are generally considered to be too ragged and masculine for women (Moghadam, 1998). This had as result that in 1985 79% of the employed Turkish women were active in the agricultural sector (Worldbank, 2016). Although, to put this figure into perspective, the group

of economic active women was still remarkably low. Furthermore, the disproportionate inability for women to profit from the new opportunities did not put the archaic male-breadwinner’s system under stress. Without the ability to get a job, women committed themselves as housewives, as their only viable alternative, to their husbands in order to survive (Bennholdt-Thomsen, 1988).

The public sector was the main engine of formal women employment during the social-democratic governments in the 60s and 70s. During this period the amount of civil servants was annually growing by 8-11% thereby outperforming the private sector opportunities substantially (Türel, 1999). This era has been called ‘etatism’ (English: statism) where the Kemalists introduced central planning policies, mixed with market elements, to achieve self-sufficiency (Celasun, 1989).

Economic pressure in 1977 made the import substitution strategy untenable. The costly machinery needed to mimic western economies, had to be imported from developed countries. Therefore, Turkey became dependent on short term foreign credit to pay its current account deficit and government debt (Celasun, 1989). When global markets contracted in the late 70s, Turkey was led into a full-fledged currency and debt crisis (1979). After the military coupe d’état of 1980, which was partly legitimized by the deprived economic situation, prime-minister Özel, head of a conservative government, transformed this untenable economic system to an export oriented program. In order to improve the economic competitiveness, this strategy had as main goal to utilize Turkish main comparative advantage: cheap labor. In accordance with the IMF and Worldbank, Turkey developed a stabilization and structural adjustment program implementing the credo of the Washington consensus: liberalization, privatization and deregulation (Senses, 1991). Ever since, the Islamic parties (like the AKP) have united their religious agendas with market-friendly policies (Narli, 2007). The rise of the labor oriented manufacturing industries at the end of the 20th century offered increased opportunities for Turkish women in the weaving and clothing factories. The reduction of government employment blocked an important lifeline of employment for women. As seen earlier in table 2, only 12,9% of employment is generated by the public sector, which is in contrast with the average of 21,3% in other OECD countries (OECD, 2013). Also from the supply side, government jobs became less attractive since real wages were affected by inflation, as they were fixed on the zero-line to cut back on government expenditures (Yavuzaslan, 2015).

The painful and profound structural adjustment policies paid itself off as Turkey is currently ranked as 45th _{out of 148 countries in 2015 on the ranking of the World Economic Forum measuring the}

efficiency of financial, good and labor markets. On this ranking Turkey outdistances European countries such as Italy (49), Hungary (60) and regional competitors Egypt (119), Lebanon (113), Morocco (72). It made Turkish private sector more internationally competitive (Yeldan, 1987), less

discriminatory towards women (Black & Brainerd, 2004; Cagaty & Ozler, 1995), and greatly improved the salary conditions of private jobs compared to civil servants in the tertiary sector (Önarnan & Stockhammer, 2005). In short, women in Turkey have a viable alternative for patronage since the economic opportunities have become more widespread as is the case in other modern economies in the west. Both from the demand as well as the supply side women became increasingly absorbed by private employment. However, the transformation is not yet completed as many substantial reforms in the capital and labor market still have to be taken (Demir & Erdem, 2010). To illustrate, the Turkish labor markets are still sticky and favor nepotistic relationships.

Iran: Modern Patronage

After the era of convergence during the reign of Attatürk and Reza Shah, Iran started to diverge economically from Turkey since the 50s. Mohammad Reza Shah (1941-1979), son and successor of the former shah, overthrew the nationalist prime-minster Mossadeq with the support of the Americans. Ever since, the power became increasingly concentrated in the position of the Shah with the abolishment of the democracy in 1975 as the final step of this process. Just after the coupe, Mohammad Reza renegotiated a lucrative oil deal for Iran with the British, as compensation of being a western satellite state during the Cold War. This has led to extraordinary profits of the Iranian government since the resource stock was directly controlled by the government (Abrahamian, 2013). Ever since, Iran has developed itself as rentier state, where the government takes a dominant role in the economic arena financed by the profits of its national resources (Shambayati, 1994).

During the oil boom of the mid-70s, government expenditures became as substantial as 50% of total GDP with the result of increased investments in health-care and education. In this same period, 84% of government revenues directly stemmed from the rents of oil. The government became the engine of import-substitution based industrialization, as 70% of the fixed capital formation was invested by public money. The emphasis on heavy industry, like in Turkey, nor the central role of oil industry, provided economic opportunities for women. Most of the working women, although still a limited group, were absorbed by increased demand of civil servants in conjunction with the expanding bureaucracies. In the 1970s the Iranian government employed more than one million civil servants, which is an increase of 600% compared to two decades earlier. Although, the Shah actively stimulated women to adopt western values by discarding the hijab, introducing modern family laws, and increasing employment possibilities, only a select urban elite profited from his female friendly policies. The female participation rate of 11% was still far below the regional average. Another source of formal employment for women in Iran were the multinational corporations, predominantly in the blue-collar carpet weaving and clothing sector (Moghadam, 1993).

In 1979, increasing social inequality, the dilapidation of the democracy, and corruption were the fertile ground for a successful popular revolt of liberals, nationalist, feminists and Islamists against Shah Mohammed Reza with the latter group of conservative Muslims gaining control (Buchan, 2012). The new political and religious leader, Ayatollah Khomeini, came to power with both a welfare and an Islamic agenda (Bahramitash, 2003). During the revolution, the majority of women and feminists supported Khomeini, as the policies of the shah were only profitable for the Tehran elite. The accumulated inequalities in advance of the revolution made moral redistribution and reverse of the “Westoxication” a priority for the new theocratic government (Moghadam, 2005). The social progress with regard to gender equality after the Shah’s modernizing ‘white’ revolution (1963), consisting of equal inheritance laws, improved autonomy, and mobility of women, was reversed by the introduction of the Sharia. Once again women required the formal approval of their husbands to work or even leave the house. The compulsory hijab became symbolic for this conservative and Islamic vision on the role and position of women within society. The group of upper middle-class women favored by the Shah returned to their homes either because of theirrelations with the ancient regime or their unwillingness to adhere to the strict sharia rules introduced on the working floor (Bahramitash, 2007). In contrast to the change in government gender norms after the revolution (1979), which promoted a domestic jihad, the overall rate of female labor force participation remained stable but low at a rate of 11% as we can see in table 3. This stable figure after the revolt was mainly the result of the Iran-Iraq war (1980-1988), where women needed to occupy the vacant position of the men who were fighting at the front. After the revolution an increased amount of economic opportunities were generated in the public sector. In table 3 we can observe that before the revolution only 19% of the working women was employed for the government, while after the revolution this figure increased to 61%. In contrast is the sharp decrease in female blue-collar work as a result of the exclusion of the manufacturing industry from the global markets and outflow of multinational corporations after the introduction of economic sanctions.

Time Period

1975-76* _1982-1983* ₂₀₀₆** Government employment

% working men in public sector 6% 19%

% working women in public sector 19% 61% 50%

Overall Labor Market

% labor force participation women 11.2% 11.1% 18% % women in blue collar jobs 20-27% 6%

*Figures based on data from Moghadam (1993) ** Source Ipums 2006

In the 1990s, president Rafsanjani (1989-1997) committed himself to reform and diversify the Iranian economy to become more competitive and to eradicate poverty. This attempt failed as in 2005 only 9% of the Iranian export was resulting from non-oil revenues (Moghadam, 2005). To date, this has not led to complementary opportunities Turkish women experienced since the 80s who were being absorbed by the labor-intensive export sector. Furthermore, Iran still has its advantage as resource state with currently 15% of the global gas resources and 10% of the oil being located in the country (US Energy department, 2011). Iran, therefore, has the ability to continue its position as rentier state even in the modern era. However, the private sector is less efficient and unable to live up its demographic dividend with youth unemployment around 30%, especially for young women who face an unemployment rate of 50% (Worldbank, 2016). The underdevelopment of this sector has been attributed to two processes. Specifically, for Iran, Amirahmadi (1990) argues that the combination of the Iranian revolution, after which economic sanctions were imposed, and the Iran-Iraq war, the industrialization process has been deprived. According to Karshenas (1990) rentier states, like Iran, did not feel the need to reform their private sectors. In contrast, many resource scarce countries like Turkey had to strictly adjust their economy to remain internationally competitive. It seems as if the resource advantage of Iran has turned into a liability for their performance in the globalized economy (Moghadam, 2005).

The feminization of government employment became the only viable opportunity for educated women (Moghadam, 2005). These women are not experiencing much competition from men in the public sector since most qualified males have found jobs in ‘masculine’ industries directly or indirectly related to the oil branch were the wages are considerably higher. The disproportionate wage distribution towards men and lack of employment opportunities for young women led to the re-embracement of patriarchy by many Iranian women (Moghadam, 2003). For example, Moghadam (2009) concluded that the amount of arranged ‘Mahr’ marriages increased recently. In these marriages the husband and wife agree upon a monthly allowance paid by the husband in order to generate a stable source of income and to fulfill her duty as housewife.

To put into perspective, the dependency of Iranian women on the government can make them vulnerable, as public employment opportunities are instable in rentier states. The IMF calculated that the oil revenues as most important income for the Iranian government are under stress considering the stressed market, discovery of shale gas in The United States, and depleting resources (IMF, 2015). The potential government revenues from oil will decrease from almost 50% now to 33% in 2019 leading to necessary government cuts of at least 10%. Farzanegan (2011) has found that government consumption, mainly consisting of government employment, is highly elastic to the oil price. This

trend is most likely to continue in the upcoming decades making public employment an unsustainable source of empowerment and showing the sensitivity to patronage.

IV.

L

ITERATURE

O

VERVIEW

AND

T

HEORETICAL

FRAMEWORK

To data, the majority of the literature concerned with the discrepancies between private and public employment is primary focused on the wage differentials between both sectors (Chrisofides, 2002; Blank, 1985; Assaad 1997; Tansel, 1999 2004; Jovanovic and Lokshin, 2004). These authors estimate the likelihood of ending up in either sector in order to use it as correction in the wage-setting

equation for the public and private sector. In these studies, the emphasis is not put on the underlying mechanism of the decision itself, but merely to identify the differences in wages between the public and private sector. In the framework presented in figure 1, our model is presented in order to analyze the decision of a woman to work for the public or private sector. In order to construct this model, the existing literature on public and private employment is supplemented with publications on female labor force participation. As is assumed in Tansel (1999, 2004), the labor force decision and the subsequent sector selection are made simultaneously, which makes the sector selection an extension of the decision to join the work floor.

Based on this literature we can construct a framework, which is portrayed in figure 1 and will be explained throughout this chapter.

To start with, four aspects of this framework are important to consider. Firstly, women are considered as the central agent in the decision framework with the final say over their economic decisions (Spierings et al., 2010; Moghadam, 2004). The woman can peruse her own preferences, although she is constrained by external actors. It is of interest to examine to what extend the situation of a woman as autonomous and empowered individual can be improved. This Is in line with the contemporary feminist’s view that sees women as:

“ agents of change rather than passive recipients” (p.2 Tasli, 2007)

Secondly, the labor decision of the woman is nested in the market for female employment where both demand and supply conditions for female labor have to be met (Klasen en Pieters, 2013; Morikawa, 2015). In this study we are primarily interested in supply side factors of female

employment. These factors can be considered as the characteristics and motivations of the employee that lead to the decision of a woman to work in the public or private sector.

Thirdly, the supply side of the sector selection should not be reduced or restricted to the woman as individual. These women are nested in her family as spouse and mother, and inhabitants of their region and country. When deciding where to work, she will take the consequences of these multiple societal roles into considerations. Especially in the Middle-East, the roles for men and women in society are strictly separated with the former mentioned having a productive responsibility and the latter a reproductive. The traditional patriarchal family has experienced dramatic changes with the reduced importance of the extended household. Nevertheless, male-dominance is still an important aspect of Middle-Eastern cultures (Spierings et al. 2010; Sharabi, 1988). Therefore, women are still subordinates compared to their men making her decision interrelated with the opinion of her husband (Yamin, 2011; Göksel, 2013; World Bank, 2009). To illustrate, women in Iran still need the formal approval of their husbands to start working (Majbouri, 2016). Furthermore, recent policies are exemplary for the emphasize most Middle-East countries put on the reproductive role. The concept of a ‘domestic jihad’ for women was raised by Rafsanjani (President of Iran). Similarly, in Turkey the recent (re)introduction of the three child policy by president Erdogan AKP emphasizes this role. Therefore, the supply decision of women could be directly shaped by these multiple contrasting roles. Even more so since public and private employment are to a different degree compatible with being

an individual, mother and wife. In the following paragraphs these three roles of the woman are directly linked to her supply as employee in the public or private sector.

Finally, the demand side only becomes important for our analysis as it differentiates the opportunities for women. Women become disproportionally depending on public employment in regions where the private sector has limited amount of demand as a result of economic (unemployment) or cultural (gender hierarchy) reasons. The degree to which a woman becomes vulnerable to these demand side restrictions depends on her ability to defend herself in these rigid situations. Especially, human capital seems to be important, as expressed in the educational dominance hypothesis (Spierings et al., 2002).

Woman as individual

The woman as individual has her own set of preferences and aspirations when choosing a job. Most economists concerned with the public/private decision consider the woman as a homo economicus selecting the job with the highest return in the form of wages (Tansel, 1999, 2004; Assaad, 1997; Blank 1985). This is in line with earlier economist such as Mincer who states that the labor decision is a trade-off primarily between monetary inputs and opportunity costs of working (Mincer, 1962). Contrasting is the view on woman as being a homo sociologicus. From this socological perspective, the women is not concerned with mere monetary outcomes, but considering a broad set of factors to select a job. To illustrate, Maczulskij (2016) found that risk-averse women in Finland are willing to accept lower wages in the public sector, since she is offered more job security as a civil servant. In addition, the OECD has found that many public jobs in the Middle-East offer substantial non-monetary benefits over the private sector. Examples are the substantially longer maternity leaves and job tenure among other (OECD, 2008). Two individual characteristics are important determinants for the public private decision as they shape the monetary returns as well as the non-monetary possibilities: experience indicated by age and human capital embodied in education

To start with the first mentioned, the monetary return structure seems to differ between the public and private sector for women of different age categories. Salaries in the Turkish and Iranian public sector are directly linked to seniority as a result of the highly formalized wage setting procedures (Akarçay-Gürbüz & Polat, 2016). In contrast, private sector wages are connected to the marginal productivity of an employee, which is supposed to increase with age (Mincer, 1962). Since the wages are bound to the marginal productivity, the return to age starts to depreciate after a certain threshold. For example, the know-how you built up at the beginning of your career could have become obsolete. Therefore, Tansel (1999, 2004) in Turkey and Assaad (1997) in Egypt has found,

correcting for a variety of factors including self-selection, that young women initially have a wage advantage in the private sector. However, for middle-aged and elderly women public employment offers a higher return. Complementary, in the same research both authors concluded that young women are actually most likely to end up in the private sector and older women are working for the government ascribing it to the sector’s wage differences.

With regard to education we can discover similar diverging wage patterns for women in the public and private sector. In Turkey the return to human capital is higher in the private sector for women with low levels of education and women with a university degree (Tansel, 2004). The group of women with intermediate levels of human capital seems to be best of as employee of the government (Tansel, 2004; Assaad, 1997). Complementary with these findings are the work of Psacharopoulos & Patrinos (2004) in MENA , ILO (2010) in Jordan and Assaad (1997) in Egypt who suggest that the private sector offers an advantage over the public sector especially for tertiary educated employees. They argue that the maximum wages are not constrained to an upper limit in the private sector. However, education is not only advantageous in terms of increased wages. As mentioned earlier, the hiring procedures in the public sector, as a large bureaucratic institution are formalized. In Turkey a centrally organized HR department is responsible for all the job applications (OECD, 2008). This department has to use select based on simple indicators such as education to sort from the vast inflow of applications. Therefore, a completed secondary education could be considered as an entry ticket for public employment. Although this is not as extreme as in Egypt during president Nassar where both men and women, who finished a secondary education, were guaranteed of a job as civil servant. Since education is an important condition for public employment, women with more human capital have increased autonomy to peruse the monetary advantages in the private or the non-monetary advantage in the public sector.

Woman as mother

Having children could be a considered as an important opportunity cost of working, since women have to combine their productive and reproductive roles. In a recent survey study conducted by Moghadam (2009) in Northern Iran, almost 30% of the women indicated that the responsibility for her children is the most important reason to stay at home. Dedeoglu (2010) concluded that family friendly working conditions are valued as important for Turkish women who are currently looking for a job. With the introduction of the new family laws under the Khomeini regime in Iran, public sector jobs have adapted to fit this dual responsibility of women at home and in society (discussed in chapter II) by providing part time jobs. In Iran, an average working week of a white-collar job for a woman consists of 35 hours in the public sector and 40 in the private (HEIS, 2012). In Turkey the

average working week is much longer in both sectors, respectively 40 and 45 hours in the public and private sector. The proportionate advantage of the public sector in terms of working conditions is similar (Tansel, 2004) in both countries. So the relative low working hours in the public sector of both countries decreases the opportunity cost of working mothers, thereby making a job as civil servant preferable, especially in Iran.

Woman as wife Husband as resource

The role of the husband in the woman’s decision between public or private employment could be substantial. Empirically, Tansel (2004) and Assaad (1997) have found a strong link between the public/private decision of the husband and the wife with the direction predominantly running from the former to the latter. A husband can be an asset for a woman in order to advance in life. Bernasco (1998) argued that the economic and social resources and accomplishments of men could be deployed to support their wives to advance in the public arena. Tansel (2004) and Spierings et al. (2010) attribute this to the network effect where the husband can utilize his contacts in order to arrange a job for his spouse. This fits in well with the overall importance of ‘wasta’ (ةط ِسا َو ; tr: Nepotism/Networks from Arab) in the Middle-East, where it is hard to get hired when you are not part of the inner circle (Barnett et al., 2013). Women in Iran still have limited geographical and social mobility, as the approval of the husband (or other senior male) is still required to move the house or get a job. Therefore, Iranian women are expected to depend more on the network of her husband, as they are restricted to build a social network on their one.

Two additional explanations could be important for the finding of Tansel and Assaad. Firstly, contradictory to the abovementioned network argument, it could be well the case that people working in the same sector have a higher probability of ending up married. Offices are a fertile environment for socialization between men and women, in particular since gender segregation in public spaces is still common. So, working in the same sector could be a cause of marriage not a result. Secondly, it could be the case that it is the status derived from a job that creates this positive correlation between the sector of both spouses (Smits et al. 1996). Turkish women admit that status considerations in general are an important factor for the simultaneous decision to participate and the content of the job (World bank, 2009). The Status similarity hypothesis predicts that both wife and husband take into consideration the relative status derived from their work. The status linked to the occupation of the one partner creates the boundaries of the other partner to operate in (Hout, 1982). Although this hypothesis has been used to explain the occupational similarities between both

spouses (Smits et al. 1996), it could be extended to the public/private decision. It is a common phenomenon in the Middle East that being employed for the government offers you status premium (Yavuzaslan, 2015).

Husband as restriction

In the MENA region at large, traditional values held by husbands have been found to be important catalysts for the working decision of women. In six Arab countries Spierings et al. (2010) found that indicators for traditionalism and patriarchy, such as the age difference between spouses, have a strong negative influence on the odds of women being active in the economy. Göksel (2010) has made a similar conclusion for Turkish women. She found that traditional men, measured by a survey based conservativeness index, are less likely to allow their spouses to work outside the house. These patriarchal values are held by the family including to woman herself, but predominantly by the male household head. Moreover, conservatism is likely to alter the perceived suitability of working conditions and job content for women. To illustrate, the modern Turkish state was founded as secularist republic. Consequently, there is a strict separation between religion and state. This is also translated to the working conditions in the public sector, since public employees in universities and government agencies are, for example, prohibited to be veiled. Therefore, traditional women could be unwilling, and their husbands probably will not allow them, to work for the government as the working conditions are in contradictions with their conservative values and practices. In 2013, the Erdogan administration made a first attempt to alter these strict secular regulations.

In sharp contrast is the situation in Iran were all women, as a result of sharia law, are obligated to wear a hijab apart from their religious convictions. These strict moral regulations could stronger enforced in the public sector because of two reasons. Firstly, the government can more easily enforce its regulation in the public than in the private sector as the latter cannot be continually examined by public law enforcement (Majbouri, 2016). It is costly to investigate whether all private companies and institutions are adhering to the strict sharia regulation. Furthermore, the working conditions for women in the public sector are subjected to additional moral prescriptions. These guidelines are regulated ‘the good conduct of behavior in public sector’ law (IRN-1993-L-35670). To illustrate, in public universities women are still physically separated from men during lunchtime. So the expectation is that conservative husbands would lead women towards private employment in Turkey,

while in Iran the opposite is true certainly in the light that women still needs the approval of their husband or father to start to work.

IV B. Women and governments

In situations when the private sector is unable to absorb female employment, women become dependent on government employment as the only provider of economic opportunities. This depending, subordinate, and vulnerable economic position for women with regard to their governments has been called patronage. It could either happen when the private sector is unwilling to hire women as they consider females as secondary workers (Ancer, 1998), or because they are unable to do so in case of high unemployment. Although the importance of government employment differs between both countries as we have demonstrated in chapter II, gender-related working hierarchy and high unemployment rates and are both common in Iran and Turkey (Tansel, 2002; Moghadam, 2009).

Gender discrimination and hierarchy can partly be explained by the discontinuity of the labor process due to maternity leaves (Halpert and Burg, 1997). Most of it results from gender stereotypes, especially with regard to ‘statistical discrimination’ where employees are selected on certain simply interpretable attributes such as sex and age (Heilman, 2001; Tomaskovic-Devey, 1993). The fact that the private sector is more affected by gender discrimination is, among others, shown by Tansel (1999). In Turkey, after adjusting for the difference in endowment, women and men face a comparable return if they are employed by the government, in the private sector the return to both education and experience is significantly lower for women. Moshe and Semyonov (1994) have found similar results for Arab women in Israel who are faced with both gender and ethnic discrimination. A first possibility is the ‘glass-ceiling’ hypothesis where highly educated women increasingly join public jobs as their relative opportunities as civil servant compared to the private sector grows in rigid situations. This hypothesis finds strong support in research related to ethnic labor markets. Race and gender discrimination on the working floor are comparable as they are both based on the same stereotyping mechanisms (Sidanius & Veniagas, 2000). Moshe and Semyonov (1994) concludes that Arab women in Israel have lower wages and career opportunities in the private sector when they are working in an area with a Jewish ethnic majority compared to Israeli territories with Arab majorities. On the other hand, in areas either with an ethnic majority or minority, the public sector offers comparable employment conditions. Especially for highly educated Arab women the opportunities in the public sector become better in those areas where she is confronted with most ethnic discrimination. In a similar manner applied to government employment in Iran and Turkey, women

who have received a degree could strategically choose for public employment in regions where the private sector does not have much to offer. When the private sector is underdeveloped, the glass ceiling seems easier to break as civil servant. This could also be considered as patronage, as soon as the inflow of highly educated women as civil servants crowd out low educated women from public employment and the labor market in general.

Secondly, education has been praised as a weapon against marginalization in Iran and Turkey by increasing the ability of women to join the labor force. (Gündüz-Hoşgör & Smits, 2006; Elmi, 2008;

Majbouri, 2016; Taymaz, 2009; Moghadam, 2003). From this general observation three contrasting hypotheses have been derived to describe the value of education in rigorous environments. Firstly, Spierings et al (2010) favor the ‘education-dominance’ theory where the relative benefit of education in the employment process increases when the barriers are most rigid. To illustrate, Spierings et al. discovered that in Middle-Eastern areas with high degrees of cultural traditionalism, tertiary education seems to have a more profound effect on the odds of being employed for women than in more liberal areas. Corresponding is the finding of Assaad (1999) who finds that highly educated women in Egypt are better able to mobilize themselves in the labor market after the structural adjustment programs. Applied to government employment, low educated people could become more depending on government employment in situations when the private sector is underdeveloped, while highly educated women, better armed against gender hierarchies and more mobile to replace themselves in the labor market, keep their economic autonomy to choose either sector.

Contrasting is the ‘situational-dominance’ hypothesis introduced by Spierings et al. (2012). The authors found evidence that the added value of education decreases for women when the labor opportunities are undersized. Translated to patronage, it could indicate that all women regardless of their human capital become evenly dependent on government employment.

V.

D

ATA

& M

ETHODS

VA. Model

To study the research questions ( presented in Chapter I), an empirical model will be constructed based on the presented theoretical framework of chapter III. The following equations (1-5) offer a step-to-step approach to demonstrate multiple theoretical and methodological dimensions of the model.

In Equation 1, we attempt to predict the probability that a woman works in the public sector (

Y

i

¿

instead of the private sector based on a set of individual characteristics ( Xij¿ such as education

and age.

Y

_i

=

β

₀

+

∑

i , j=1 j

β

_j

(

X

_ij

)+

ε

_i

With i = Set of working women in Iran and Turkey j: individual characteristic

Equation 1: Starting model

The presented model in equation 1 would return unbiased but inefficient results if it would be estimated with an ordinary least squared model. The Gaus-Markov assumption of uncorrelated error terms (equation 2) between individuals will be violated. Women living in the same region could both have a higher probability of public employment.

Cov

₍

ε_{1 p}, ε_{2 p}

₎

=0

With p = Turkish and Iranian Provinces

Equation 2: Gaus Markov condition of uncorrelated errors

For example, in the capital province Ankara, the bureaucratic hub of Turkey with all the government institutions, the relative supply of public jobs is considerably larger compared to the Mediterranean provinces (Antalya, Izmir). Therefore, women in Ankara have a higher probability of ending up as civil servant compared to women in other regions. In Iran a similar relative concentration of public jobs can be found in the urbanized and developed areas (Tehran, Esfahan & Shiraz). This relative supply factor is presented in equation 3 as the cp term. So women living in the same province (n) share a

common component in the total error term (

c

p

+

ε

¿

¿

leading to the violation of equation 2.

Y

_i

=

β

₀

+

∑

i=1 j

β

_j

(

X

_ij

)+

c

_p

+

ε

_i

With p = Turkish and Iranian Provinces i = Set of working women in Iran and Turkey j: individual character

Equation 3: Biased OLS model

The transformation of the OLS model into a least squares dummy variable model (more commonly known as the Fixed effects model) would absorb all between-province variation (Wooldridge, 2009). As presented in equation 4 we could add dummies for all Iranian and Turkish provinces with the exclusion of one (dummy) province. The direct cross-country differentials are also transferred and absorbed by the included provincial dummies. For the fixed effect model to provide efficient outcomes, we will only include provinces with more than 15 units of observation (women) in the data set.

Y

_ip

=

β

₀

+

∑

i=1 j

β

_j

(

X

_i

)+

∑

p=1 110

B

_p

∂

_p

+

ε

_i

With p = set of Turkish and Iranian Provinces i = Set of working women in Iran and Turkey

j= individual characteristics

Equation 4: Fixed effect dummy model

The major disadvantage of this model is that the included dummies do not only correct for the unobserved heterogeneity between regions, but as well for factors we would like to to measure. Including regional factors, such as unemployment rates, to predict the public/private decision of women would simply be perfectly linear with the regional dummies. It is still possible to include indirect effects between regional and individual characteristics as interaction terms. In equation 5 this

situation is illustrated, where we test the three presented education hypothesis. The individual level of education (

X

iEducation

¿

is interacted with the regional degree of unemployment and gender

hierarchy ( Dp, Up¿ to discover how the effect of education on the public/private decision is

altered by these regional indicators of economic and cultural rigidness.

∂

(

¿¿

p)+

∑

i , j , p=1 k

γ

_k

₍

X

_iEducation

∗

D

_p

₎

+

∑

i , j , p=1 l

γ

_l

₍

X

_iEducation

∗

U

_p

₎

+

ε

_i

Y

_i

=

β

₀

+

∑

i , j=1 j

β

_j

₍

X

_ij

₎

+

∑

p =1 110

B

_p

¿

With p = set of Turkish and Iranian Provinces i = Set of working women in Iran and Turkey j: individual characteristics

Equation 5: Fixed effects model with regional interactions

In order to make a comparison between Turkey and Iran for both the individual and interaction effects, we include an additional interaction terms both with the vector of individual characteristics (

X

_ij ) as well with the regional interaction terms of unemployment and gender hierarchy

(

Xedu ,i∗Dp∗Countryt

)

and

(

Xedu ,i∗Up∗Countryt

)

.

∂

(

¿¿

p)+

∑

i , j , p=1 k

γ

_k

₍

X

_iEducation

∗

D

_p

₎

+

∑

i , j , p=1 l

γ

_l

₍

X

_iEducation

∗

U

_p

₎

+

∑

i , j ,t =1 v

θ

_n

₍

X

_ij

∗

Country

_t

₎

+

∑

edu ,i , p , t=1 w

φ

_m

₍

X

_{edu , i}

∗

U

_p

∗

Country

_t

₎

+

∑

edu , i , p , t=1 w

φ

_m

₍

X

_{edu ,i}

∗

D

_p

∗

Country

_t

₎

+

ε

_i

Y

_i

=

β

₀

+

∑

i , j=1 j

β

_j

(

X

ij

)

+

∑

p =1 110

B

_p

¿

With p = set of Turkish and Iranian Provinces i = Set of working women in Iran and Turkey j: individual character and t: country

Equation 6: Fixed effect dummy model with regional and country interactions

Binary dependent variable

The dichotomous character of our dependent variable, a woman is either active in the public or private sector, makes it necessary to transform the model from an ordinary least squares model to a logit variant. The logistic estimation procedure will have two advantages over the OLS model: I) it

makes sure the estimated values fit within the range of 0 and 1, 2) It takes care for heteroscedasticity since the variance depends and error term depend on the value of the predictor (Demaris, 1995). The coefficients computed in this model do not present the direct change of the concerned variable on the dependent factor, but rather the change of the log of the odds of being employed publically are privately with every step increase of the independent variable.

Effect coding

Similarly, multiple independent variables have a categorical character including education (4 categories), marriage (2) and the sector of the partner (2). In this analysis, the categorical variables are effects-coded creating an efficient reference point to interpret the effect of these categorical variables: the mean of means . If we divide all Turkish women into four categories: no education, completed primary, completed secondary, and completed tertiary we could calculate the division of public and private employment for each education group indicated with

PP

´

zero−Turkey as portrayed

below in table 4.

Turkey Education

Zero Education Primary Education Secondary Education Tertiary Education

% Average public

employment

PP

´

zero−Turkey

PP

´

Primary −Turkey

PP

´

Secondary−Turkey

PP

´

Tertiary−Turkey

Table 4: Constructing mean of means education (Turkey)

The reference point to evaluate the effect of having completed tertiary education is the ‘mean of means’ of these four education groups:

Mean of Means educationTurkey=

PP

´

zero−Turkey

+ ´

PP

Primary −Turkey

+ ´

PP

Secondary−Turkey

+ ´

PP

Tertiary−Turkey

4

Equation 7: Calculating Mean of Means education

This has two advantages over regular dichotomous dummy coding (Binary: one or zero). 1) The outcome is not sensitive to the excluded reference category, which is especially useful when analyzing interactions effects. With binary coded dummies you observe how the categorical variables deviates from the (excluded) reference category along the interaction term. For example, the degree to which women with secondary education (dummy) deviates from tertiary educated women (reference) in provinces with changing levels of unemployment (interaction term). With effect coded dummy’s you analyze this deviation of secondary education in comparison to the earlier explained mean of means