female labor force participation in turkey: the role of conservatism

10
Female labor force participation in Turkey: The role of conservatism İdil Göksel İzmir University of Economics, Department of Economics, Sakarya Cad. No:156, 35330, Balçova, İzmir, Turkey article info synopsis Available online 16 May 2013 The aim of this paper is to investigate the role of conservatism in shaping women's labor force participation decision. Turkey is a particularly important case as, unlike in many other countries, female labor force participation (FLFP) has shown a decreasing trend over the last 50 years. In addition to the main determinants found in previous literature, this paper adds a new variable that influences FLFP in Turkey: conservatism and the role of social norms. Three indices that might influence FLFP are identified: religion, social norms, and conservatism. The results are in accordance with the previous literature, which reveals that urbanization, and education levels play an important role in FLFP. An additional important innovation of this paper is the identification of the impact of social norms and religion on rural and urban areas. While such factors have a negative effect on female employment in urban areas, no significant effect is seen in rural areas. © 2013 Elsevier Ltd. All rights reserved. JEL classification: J16 J21 J71 Introduction This paper aims to investigate the role of conservatism in shaping women's labor force participation decisions in Turkey, which over the last 50 years has witnessed a declining female labor force participation rate. In this regard, Turkey is an interesting case for the investigation of the determinants of labor force participation. There are also arguments with increasing tendencies in Turkey towards more conservatism especially after the Justice and Develop- ment Party's rise to power. Moreover, according to the Global Gender Gap Report 2012 (World Economic Forum, Geneva, Switzerland), Turkey's gender gap index equals 0.601, making Turkey the 124th country among the 135 countries included in the survey. 1 Global gender gap index benchmarks national gender gaps on economic, political, education and health criteria. Even more seriously, Turkey has the second lowest gender gap index ranking (just above Iran) in the upper middle income group that it belongs to. As a result, it is important to elaborate on the reasons behind these facts in order to form future policies both for Turkey and for other developing countries that may face the same challenges. Especially during the EU accession process of Turkey, it is essential to identify such issues, propose possible solutions, and examine the female labor force participation determi- nants potentially responsible for Turkey's extremely low ranking in the gender index. Last but not least, as Ilkkaracan (2012) states, in the case that Turkey still aspires to be a member of the European Union, it has to meet the EU Lisbon criteria, with minimum targets for member nations' employ- ment rates of women and men at 60 and 70%, respectively (Council of European Union, 2000). In the literature, there is a well-established link between education, fertility and female labor force participation (FLFP) (Spierings, Smits, & Verloo, 2010; Van der Lippe & Van Dijk, 2002). Figs. 1 and 2 present the evolution of Turkish fertility, female education and female labor force participa- tion rates, respectively. While the level of education is increasing, the fertility and female labor force participation rates are decreasing over time. For example, the share of women educated to at least high school level, as a proportion of all urbanized working age women increased from 23% in 1988 to 42% in 2008 (Ilkkaracan, 2012). Accordingly, neither education level nor fertility rate can be a major factor in the decreasing trend in female labor force participation. Education and fertility levels show an improving trend, thus, we must seek other factors influencing the stagnant FLFP in Turkey. For this purpose, this paper aims to examine the relationship between social and cultural factors, and FLFP. By investigating the relationship between conservatism and FLFP, this paper adds previously unconsidered variables which have an influence on the FLFP in Turkey, namely: Women's Studies International Forum 41 (2013) 4554 0277-5395/$ see front matter © 2013 Elsevier Ltd. All rights reserved. http://dx.doi.org/10.1016/j.wsif.2013.04.006 Contents lists available at ScienceDirect Women's Studies International Forum journal homepage: www.elsevier.com/locate/wsif

Upload: idil

Post on 27-Jan-2017

215 views

Category:

Documents


2 download

TRANSCRIPT

Women's Studies International Forum 41 (2013) 45–54

Contents lists available at ScienceDirect

Women's Studies International Forum

j ourna l homepage: www.e lsev ie r .com/ locate /ws i f

Female labor force participation in Turkey: The role of conservatism

İdil Gökselİzmir University of Economics, Department of Economics, Sakarya Cad. No:156, 35330, Balçova, İzmir, Turkey

a r t i c l e i n f o

0277-5395/$ – see front matter © 2013 Elsevier Ltd. Ahttp://dx.doi.org/10.1016/j.wsif.2013.04.006

s y n o p s i s

Available online 16 May 2013

The aim of this paper is to investigate the role of conservatism in shaping women's labor forceparticipation decision. Turkey is a particularly important case as, unlike in many othercountries, female labor force participation (FLFP) has shown a decreasing trend over the last50 years. In addition to the main determinants found in previous literature, this paper adds anew variable that influences FLFP in Turkey: conservatism and the role of social norms. Threeindices that might influence FLFP are identified: religion, social norms, and conservatism. Theresults are in accordance with the previous literature, which reveals that urbanization, andeducation levels play an important role in FLFP. An additional important innovation of thispaper is the identification of the impact of social norms and religion on rural and urban areas.While such factors have a negative effect on female employment in urban areas, no significanteffect is seen in rural areas.

© 2013 Elsevier Ltd. All rights reserved.

JEL classification:J16J21J71

Introduction

This paper aims to investigate the role of conservatism inshaping women's labor force participation decisions inTurkey, which over the last 50 years has witnessed adeclining female labor force participation rate. In this regard,Turkey is an interesting case for the investigation of thedeterminants of labor force participation. There are alsoarguments with increasing tendencies in Turkey towardsmore conservatism especially after the Justice and Develop-ment Party's rise to power. Moreover, according to the GlobalGender Gap Report 2012 (World Economic Forum, Geneva,Switzerland), Turkey's gender gap index equals 0.601,making Turkey the 124th country among the 135 countriesincluded in the survey.1 Global gender gap index benchmarksnational gender gaps on economic, political, education andhealth criteria. Even more seriously, Turkey has the secondlowest gender gap index ranking (just above Iran) in theupper middle income group that it belongs to. As a result, it isimportant to elaborate on the reasons behind these facts inorder to form future policies both for Turkey and for otherdeveloping countries that may face the same challenges.Especially during the EU accession process of Turkey, it isessential to identify such issues, propose possible solutions,and examine the female labor force participation determi-nants potentially responsible for Turkey's extremely low

ll rights reserved.

ranking in the gender index. Last but not least, as Ilkkaracan(2012) states, in the case that Turkey still aspires to be amember of the European Union, it has to meet the EU Lisboncriteria, with minimum targets for member nations' employ-ment rates of women and men at 60 and 70%, respectively(Council of European Union, 2000).

In the literature, there is a well-established link betweeneducation, fertility and female labor force participation(FLFP) (Spierings, Smits, & Verloo, 2010; Van der Lippe &Van Dijk, 2002). Figs. 1 and 2 present the evolution of Turkishfertility, female education and female labor force participa-tion rates, respectively. While the level of education isincreasing, the fertility and female labor force participationrates are decreasing over time. For example, the share ofwomen educated to at least high school level, as a proportionof all urbanized working age women increased from 23% in1988 to 42% in 2008 (Ilkkaracan, 2012). Accordingly, neithereducation level nor fertility rate can be a major factor in thedecreasing trend in female labor force participation.

Education and fertility levels show an improving trend,thus, we must seek other factors influencing the stagnantFLFP in Turkey. For this purpose, this paper aims to examinethe relationship between social and cultural factors, and FLFP.By investigating the relationship between conservatism andFLFP, this paper adds previously unconsidered variableswhich have an influence on the FLFP in Turkey, namely:

46 İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

conservatism and the role of religion and social norms inTurkey. The idea of explaining economic outcomes by socialnorms, religion and tradition is not new in the literature.Fernandez and Fogli (2009), make use of a 1970 census tocreate a sample which consists of women born in the US butwhose parents were born elsewhere. They use past FLFP andancestral fertility rates as cultural proxies and find a positiveand significant power of this proxy for decisions of membersof the current generation about work and fertility. This paperextends this approach by creating an index for conservatismin order to investigate the degree to which social norms andconservatism are a determinant of married women's FLFPdecisions.

This paper offers the first attempt in the forming of indicesto explain conservatism in Turkey and its influence on FLFP. Inorder to measure this position, an original proxy for conser-vatism is created by using a unique dataset about perceptionsin Turkey. Two indices that might influence conservatism areidentified: religion, and social norms. A combination of theseindices is used to determine the general influence of beingconservative in social terms. Furthermore, in this paper, theroots of conservatism are disentangled to determine whetherreligiosity or traditional social norms have a greater influenceon the female labor force participation decision.

Literature review and discussion on female labor forceparticipation and conservatism in Turkey

In recent years, an increasing number of papers combiningsociology and economic outcomes have enriched the economicliterature. The investigation of the relationship between religionand economic performance has been the focus of a number ofthese (Barro & McCleary, 2006; Becker & Woessmann, 2009;Iannaccone, 1998; Noland, 2005). In spite of contradictingoutcomes in this literature, all agree on the significant influenceof religion on economic development. The literature analyzingthe relationship between social norms and female labor forceparticipation (Burda, Hamermesh, & Weil, 2007; Fernandez,Fogli, & Olivetti, 2004; Vendrik, 2003) show that the influence ofsocial norms is also significant. Both religion and social normsplay a major role in the formation of a country's culture; hence,economists also tried to understand the connection betweenculture and economic outcomes (Guiso, Sapienza, & Zingales,2006; Tabellini, 2005), frequently finding a significant effect ofculture. Moreover, social norms, religion and culture not only

0

1990

1991

1992

1993

1994

1995

1996

1997

1998

0.5

1

1.5

2

2.5

3

3.5

Fig. 1. Fertility level of TurkSource: Turkish Statistics In

influence the economic development, but also have anintergenerational transmission effect (Bisin & Verdier,2000; Hazan & Maoz, 2002). This literature mainly concen-trated on macro data in order to analyze sociological factorson economic development of different countries. Thispaper, in contrast, concentrates on Turkey, and specificallyaims to show the effect of these aspects on FLFP. It alsopresents an innovative conservatism index which consistsof two sub-indices: religion and social norms.

The recent literature categorizes six main reasons for thedeclining trend in Turkish FLFP. These are: urbanization(Baslevent & Onaran, 2002; Ilkkaracan, 1998), lack ofavailability and affordability of childcare institutions (Acar,2008), U-shaped characteristic of labor force participation(Cagatay & Ozler, 1995; Tansel, 2002), unequal division oflabor at home (Dedeoğlu, 2010; Eyüboğlu, Özar, & Tanrıöver,2000; Ilkkaracan, 1998; Moghadam, 1998), low educationlevel of the female population and economic crisis (Dayioglu,2000; Ince & Demir, 2006).

These factors have some influence on FLFP; however, theyare not able to fully explain the decreasing trend in femaleemployment. Firstly, in Turkey, a continuing migration fromrural to urban areas since 1950s is obviously an importantfactor for the decreasing trend in FLFP. When women fromrural areas migrate to big cities, they have difficulties findingemployment both because they lack the necessary skills tointegrate into the urban labor market, and also because ofhusbands' changed attitudes to their wives' employment in thenew environment. Secondly, although the lack of availabilityand affordability of childcare institutions is an important factorin urban areas, this may not be the only explanation for thedecreasing trend in FLFP. Childcare institutions have beenimproving in Turkey and one year of pre-school education for6 year olds has recently beenmade compulsory. Moreover, thefertility rate in Turkey has been decreasing. On the other hand,as the third factor above suggests, Turkey might be on thedownward-sloping part of the U-shaped curve of labor forceparticipation. In relation to the fourth factor, as also empha-sized by Tansel (2002), tradition and culture play an importantrole in determining the degree of FLFP and their importance isdiscussed in this paper. Fifthly, it is true that the average femaleeducation level is still lower than the male, but there is anincreasing trend and it can therefore not be the sole reason forthe decrease in FLFP. Lastly, İzdes (2012) analyzes howwomenunemployment changed after three recent economic crises

1999

2000

2001

2002

2003

2004

2005

2006

ish women by years.stitute.

47İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

(1994, 1999 and 2001) and one economic contraction (1991)in Turkey. She demonstrates how different crises causeddifferent results in the unemployment rate of women. Throughthe 1991 contraction and the 2001 crisis, women employmentwas affected negatively, as women were mainly concentratedin vulnerable sectors and/or positions. On the other hand,during the 1994and 1999 crises, addedworker and substitutioneffects prevailed,meaning that female laborwas substituted formale labor during economic recessions, in linewith theMarxistview, which states that during economic contractions womentend to be more in demand as they constitute a cheaper, lessorganized and more controllable labor force. In short, differentcrises influence women's employment in different mannerstherefore the crisis alone cannot explain the downward trendofFLFP.

These six factors certainly influence FLFP, but it is notclear whether they are the sole reasons for the decreasingtrend in FLFP. This paper argues that conservatism is also asignificant determinant of FLFP, and analyzes the relationshipbetween conservatism and FLFP; hence, it is important tounderstand the nature of conservatism in Turkey first.

It is generally agreed that conservatism is on the rise inTurkey. Because conservatism is a qualitative concept, andtherefore impossible to prove statistically, I will try to givesome examples to demonstrate the upward trend in conserva-tive norms and behaviors in Turkey.

In their book, Carkoglu andKalaycioglu (2009) illustrate therising tide of conservatism in Turkey by focusing on the majorrole played by religiosity. One example is the web page ThePresidency of Religious Affairs of the Republic of Turkey, whichshows growth in the number of Qur'an courses and the numberof students involved. Figs. 3 and 4 show the number of Qur'ancourses and number of students attending, respectively for therecent period.

Carkoglu and Kalaycioglu (2009) claim that the increasingtrend towards conservatism is caused by long-term socio-political modernization, industrialization, increased pace ofsocial mobilization, and contemporary regional turbulencesdue to changes in the international system since the end of theColdWar. They claim that in the post-1980 era, Turkish societybecame increasingly more urban and relatively more affluent.

90

80

70

60

50

40

30

20

10

0

1955

1965

1975

1985

1990

1992

1994

1996

1998

2000

2002

2004

2006

FLP Education

Fig. 2. Percentage of literate women by years and percentage of female laborforce participation rates by year.Source: Turkish Statistics Institute.

At the same time, Turkish people started to be highly sensitiveto the uncertainties of socioeconomic and socio-politicalchanges occurring in and around the country. In the empiricalsection of their book they show that almost 40% of thepopulation of Turkey desires to return to the traditional socialnorms, which existed during “the good old days”. In a surveythey conducted in 2006, 51% of the respondents are clearlycloser to being very conservative, and only about 22% remaincloser to being not conservative at all.

Some items in the World Value Survey (European andWorld Values Surveys Four Wave Integrated Data File, 2006)may also give some clues about the level of conservatism inTurkey. One of the questions asks “Independently of whetheryou go to church (mosque) or not, would you say you are: areligious person, not a religious person, a convinced atheist?”.The results for Turkey are presented in Figs. 5 and 6 for malesand females, respectively. The summary of results announcedin 2011 shows that 81% of the population claims to be religious.When compared with the other countries also known to berelatively religiouswith respect to the rest of theworld, Turkeyseems to be even more religious than many of these. Forexample the proportion of those who believe that religion isimportant is 74% in Italy, 90% in Malta, 75% in Poland, and 35%in Spain. The figure for Turley is particularly high compared tothe Czech Republic, Sweden and Germany, where the percent-age is as low as 20%, 23% and 27%, respectively. There are manyquestions about social norms and conservatism inWVS, but thequestion about religion is the only one that occurs in eachwave. In summary results of the 2011 survey Esmer (2012)states that since 1996 there are two beliefs that have remainedunchanged, that males should be the head of the family, andthat women should be obedient. Three fourths of the Turkishpopulation believes that the household head should be a male;approximately 60% of the population agrees that womenshould obey their husbands. Moreover, there was an increasein the percentage of people that find it acceptable for a man tohavemore thanonewife. The rates increased from10% in 1996,to 19% and 7% for men and women respectively in 2000, and23% (including 19% of women) in 2003. Another issue whichshows a similar rise is the right of husbands to rise violenceagainst their wives. The proportion of those supporting thisincreased from 19% to 30% between 1996 and 2011. Further-more, 70% of all the population and 69% of women agree withthe statement that employment of a mother is harmful for herchildren.

The results of the findings indicate increasing conservativeattitude and norms in Turkish society inwhichwomen are seenas obedient wives and mothers. The aim of this paper is toinvestigate the degree to which conservatism influences FLFP.For this reason, I have given a brief explanation about the trendin conservatism in Turkey in this section and have suggestedreasons for this fact. As the dataset used in this paper is only forone year, it will not be possible to give a detailed explanation ofthe decreasing trend of FLFP, instead the paper aims to showthat conservatism is a significant determinant of female laborforce participation.

Links between conservatism and FLFP

In the literature, labor market participation of women isassumed to depend on their evaluation of the market wage

0100020003000400050006000700080009000

2000-2001

2001-2002

2002-2003

2003-2004

2004-2005

2005-2006

2006-2007

2007-2008

2008-2009

Nu

mb

er

Academic year

Fig. 3. Number of Qur'an courses.Source: The Presidency of Religious Affairs of the Republic of Turkey. http://www.diyanet.gov.tr/turkish/ingilizce/Default.aspx.

48 İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

against their reservation wages. In contrast, in this paper, it isassumed that women do not make decisions alone. Theiremployment decisions are determined and influenced bydifferent factors, such as social norms, husbands' permission,care responsibilities, and legal and institutional factors.2

There are three links through which social norms andconservatism influence the participation decision of women.The first is the education link. Until a certain age, parents makethe decisions about their children's educational attainment.For example, the first 12 years of education are compulsory inTurkey at present.3 If parents decide not to invest in theeducation of their children after this compulsory educationperiod, then it is impossible for these children to go touniversity, regardless of their wishes. Following Tansel (2002),in a previous paper, I show that there is a gender bias againstgirls in educational investment (Göksel, 2010). If the socialnorms are prevalently against female employment, and peoplebelieve that women's role is in the home, then the girls livingin such a society are less likely to have a high level of educationand to find employment in the future, even if they want to.

The second link is through marriage. In conservativesocieties, women have less freedom to choose their partners,and therefore, less bargaining power in the household. In asense, they have to obey their husbands after marriage.Fernandez et al. (2004) have shown that the number of menbeing brought up in a family in which the mother worked hasbeen a significant factor in paving the way for an increase offemale labor force participation. In a conservative society, awoman is unlikely to find such a husband, so her husbandwill most likely also have the same norms as her father. In amore recent paper, I show that those men whose mothers'

0

50000

100000

150000

200000

250000

300000

2000-2001

2001-2002

2002-2003

2003-2004

Nu

mb

er o

f st

ud

ents

Acad

Fig. 4. Number of students oSource: The Presidency of Ringilizce/Default.aspx.

role was a housewife tend themselves to prefer wives whoare not in paid employment. In parallel, those daughters ofmothers who were in the role of housewife tend to adopt thesame role themselves after marriage (Göksel, 2012). More-over, having low bargaining power means that women haveless influence on the decision about fertility (Rasul, 2008)and a larger family than what they would prefer also reducestheir opportunities to participate in labor force.

The third link concerns the labor market. In a conservativesociety, social norms opposed to women working meanemployers can set lower wage for females. Tansel (2005)shows that this is indeed the case in the private sector inTurkey. Likewise employers may simply hire men when thechoice is between men and women. As a result, returns toeducation for females are lower than they are for males,which in turn strengthen the first link.

In this paper, married women's labor force participation istaken into consideration as this provides an opportunity toobserve all the above-mentioned links. The education levelsof women are added to the analysis to control for theeducation link. Moreover, each item in the conservatismindex is constructed in such a way that they cover at least oneof the links above. Although the last link is more difficult toanalyze with the existing data, it is controlled implicitly asmale employers who are against the employment of theirown wives are likely to be more reluctant to hire females.

Data and descriptive statistics

In this paper, I mainly use the 2006 Household StructureSurvey (HSS) conducted by Turkish Statistical Institute

2004-2005

2005-2006

2006-2007

2007-2008

2008-2009

emic year

f the Qur'an courses.eligious Affairs of the Republic of Turkey. http://www.diyanet.gov.tr/turkish/

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

1990 1996 2000 2005

Religious Not Religious Atheist

Fig. 5. Response of Turkish men to the question “Independently of whetheryou go to church (mosque) or not, would you say you are: a religious person,not a religious person, a convinced atheist?” in respective four waves ofWVS.

Table 1Descriptive statistics.

Mean Standard deviation

Age 43.57 13.48Age2 0.11 0.47Education 4.81 3.85Grandmother 0.06 0.24Sons 1.01 1.07Daughters 0.84 1.04LogHusbandIncome 6.31 0.65Share Income 0.12 0.28Husband's conservatism 0.06 2.92Urban 0.55 0.50

Source: Author's own calculations using 2006 HSS.

49İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

(TurkStat). It is a unique dataset about perceptions and socialnorms. This survey (Turkish Statistical Institute, 2006) is aproduct of joint research by the TurkStat and the GeneralDirectorate of Family and Social Studies on the householdstructure of Turkish families. TurkStat has a standard method-ology to conduct its surveys, which are representative for all ofTurkey. This survey consists of more than a hundred questionsabout the household structure, perceptions and habits ofTurkish families. The details of the questions of this survey areexplained later in this paper.

Table 1 reports the descriptive statistics. In the table,education presents the number of years spent on educationwhile daughters and sons are the number of girls and boys inthe household, respectively. Grandmother is a dummy variablethat takes the value one for the presence of a grandmother inthe household. Loghusbandincome is the logarithm of thehusband's income. Urban is a dummy variable that takes thevalue one if the population of the location is higher than 20,000.

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

1990 1996 2000 2005

Religious Not Religious Atheist

Fig. 6. Response of Turkish women to the question “Independently of whetheryou go to church (mosque) or not, would you say you are: a religious person,not a religious person, a convinced atheist?” in respective four waves of WVS.

In the 2006 HSS, the individuals are asked directly whetheror not they think that it is appropriate that women work. Thisvariable is not used in the analysis as it is totally endogenous,but Fig. 7 presents theproportion ofmenwhodonot approve ofwomen working. When the men opposed to women workingwere asked to justify their opinion, 63% of them replied that“The woman's main duty is to take care of the children and dothe domestic work”.

Fig. 7 shows that, in general, the employment of women isfound more inappropriate in urban areas than in rural areas.Moreover, the ratio of those finding it inappropriate is higher inthe eastern regions of Turkey. As the eastern part of Turkey isconsidered to be more traditional, the dataset reveals that thereis a negative relationship between conservatism and FLFP.

In Turkey, there has been a continuing migration fromrural to urban areas since the 1950s. Discussing the manyreasons for this phenomenon is outside the scope of thispaper, but as noted before, urbanization is an importantreason behind the decreasing trend in FLFP. Fig. 7 once moreshows that attitudes of men towards women's employmentdiffer between urban and rural areas.

Methodology

In order to make use of the questions in the 2006 HSS, theprinciple component analysis4 is used to form three differentindices and the answers of the husbands are always taken intoconsideration. The main index used throughout the paper iscalled the conservatism index, which is a combination of twosub-indices: the social norm index and the religion index. Of themany questions about social attitudes in the 2006 survey, alimited number has been selected to form the social norm indexaccording to their correlation with the question mentioned inthe data section: “whether they think it is appropriate thatwomen work or not”. As explained before, due to endogeneity,this question was not used in the analysis. However, otherquestions which would highly correlate to this question werechosen to form the social norm index.5 Dummies are formedusing the answers to the following questions: “Would theexistence of unmarried cohabiting couples in your neighbor-hood disturb you?”, “Would the existence of children ofunmarried couple in your neighborhood disturb you?”, “Doyou agree with the statement: The continuation of a generationis guaranteed only by a son?”, “Doyou agreewith the statement:Having a son increases the respectability of a mother”, “Do youagree with the statement: The main duty of a mother is

0,6

0,5

0,4

0,3

0,2

0,1

0

Istan

bul

Wes

t Mar

mar

a

East M

arm

ara

Aegea

n

Med

iterra

nean

Wes

t Ana

tolia

Midd

le Ana

tolia

Wes

t Blac

k Sea

East B

lack S

ea

North

Eas

t Ana

tolia

Midd

le Eas

t Ana

tolia

South

Eas

t Ana

tolia

urban rural

Fig. 7. Percentage of men that answered negatively to the question “whether it is appropriate that women work or not” in 2006 HSS.

50 İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

childcare and domestic work?”, and “Do you agree with thestatement: Awoman that has a child has a higher status than theonewithout?”. Each dummy takes the value one if the answer tothe relevant question is positive and takes the value zero if it isnegative. Using principle component analysis6 with thesedummies, a social norm index is formed, in which higher valuesindicate individuals that are more gender biased and moreconservative.

The second sub-index is about the importance of religion inthe decision process. In the survey, there is a question that askshow important religion is when deciding on the followingconcepts: choice ofwife, choice of a friend, attitudes to clothing,voting, relationshipswith neighbors, and nutrition. The dummytakes value one if the respondent answers positively, and zero ifotherwise.

The index of conservatism is formed using all the variablesmentioned above to assess the overall influence of havingconservative norms. This index provides us with a general ideaof the effect of conservatism on FLFP, while we can disentanglewhich issues play more important roles in this process byanalyzing the sub-indices.

Although the 2006 Household Structure Survey (HSS)provides a unique dataset about the perceptions of Turkishpeople, it unfortunately lacks one of the basic pieces ofinformation needed in this paper: the employment status of

Table 2OLS regression results.

Urban

Coefficient Standard

Age 0.0020 ⁎⁎⁎ 0.0004Age2 −0.0263 ⁎⁎⁎ 0.0066Education 0.0191 ⁎⁎⁎ 0.0013Grandmother 0.0179 0.0205LogHusbandIncome −0.0275 ⁎⁎⁎ 0.0071# of boys −0.0013 0.0039# of girls −0.0035 0.0037Conservatism −0.0062 ⁎⁎⁎ 0.0022Constant 0.0927 ⁎ 0.0491Region dummies YesN 3245R2 0.1194

Note: ⁎, ⁎⁎, and ⁎⁎⁎ indicate statistical significance at 10, 5 and 1%, respectively. Ind

the woman. Fortunately, however, it contains informationabout the incomes of individuals and the total householdincome. I am aware of the fact that the presence of an incomedoes not necessarily mean that the woman actually works.She might have inherited land or real estate from which shereceives rent. Still, I believe the proportion of such cases isrelatively low in Turkey. Using the information available inthe dataset I form the variable share income, which I use as aproxy for female labor force participation. This variable alsogives the opportunity to control indirectly for the wagedifferences between genders. It might be the case that thoughboth husband and wife have the similar jobs, the wife earnsless. For this reason, I use the share of the woman's income inthe household as the dependent variable for the analysisof this dataset. Accordingly, the following OLS regressionis run.

share female incomei ¼ α0 þα1Xi þα2Ri þ α3Ii þ εi ð1Þ

where X is a vector of individual and household character-istics, R is the region dummies and I represents the indices.

In the dataset, the age is given between intervals so themedian of the interval is taken when determining age. Sons

Rural

Error Coefficient Standard error

0.0005 0.0006−0.0538 ⁎⁎⁎ 0.01030.0048 ⁎ 0.00260.0132 0.0258−0.0667 ⁎⁎⁎ 0.01040.0014 0.00640.0079 0.00660.0034 0.00450.4518 ⁎⁎⁎ 0.0781Yes23900.0711

icated standard errors are robust.

Table 3Correlation Matrix for the Variables Conservatism, Number of Sons andDaughters and Grandmother (Urban).

Conservatism # of sons # ofdaughters

Grandmother

Conservatism 1.0000# of sons 0.0529 1.0000# of daughters 0.0266 0.0334 1.0000Grandmother 0.0184 00636 0.0487 1.0000

Source: Author's own calculations using 2006 HSS.

51İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

and daughters are the number of the sons and daughters ofthe mother, respectively. Grandmother is a dummy that takesthe value one in the case of a grandmother living within thehousehold. In the dataset, income is also given in intervals.For all intervals except the last one, the median is taken; butthe last interval is given as higher than 2501 YTL. In order tofind an appropriate representative for this interval, a quantilemethod is used, as suggested in Ligon (1989) and the upperbound is found to be 2953 YTL. Urban is the dummy thattakes the value one if the individual is living in the city.Loghusbandincome is the logarithm of the husband's income.For brevity's sake only the results for the conservatism indexare shown in the tables and the results are discussed in thenext section.7

Results

The results are in accordance with previous literature,stating that urbanization and education play an important rolein FLFP. However, the specific aim of this paper is to show theinfluence of conservatism on FLFP, and the results also provethat conservatism plays a negative role in women's employ-ment in urban areas, while having no significant effect in therural areas.

The previous literature supports the view that urbanizationplays an important role in the decrease of female labor forceparticipation, which is also one of the conclusions of this paper.In order to observe the differences between the urban and ruralparts of Turkey, the regressions are run separately. I expectdifferent results for urban and rural areas due to the differencein lifestyles. In rural areas, women traditionally work, while

Table 4OLS regression results for the couples that have the same education level.

Urban

Coefficient Standard

Age 0.0020 ⁎⁎⁎ 0.0006Age2 −0.0373 ⁎⁎⁎ 0.0099Education 0.0268 ⁎⁎⁎ 0.0022Grandmother −0.0060 0.0277LogHusbandIncome −0.0378 ⁎⁎⁎ 0.0114# of boys −0.0101 ⁎ 0.0061# of girls −0.0081 0.0070Conservatism −0.0076 ⁎⁎ 0.0037Constant 0.1363 ⁎ 0.0807Region dummies YesN 1507R2 0.1817

Note: ⁎, ⁎⁎, and ⁎⁎⁎ indicate statistical significance at 10, 5 and 1%, respectively. Ind

this is not the case in cities. Table 2 presents the OLS regressionresults for both urban and rural areas.

As expected, age and education have positive and highlysignificant effect, while husband's income and husband'sconservatism have highly significant negative influence inurban areas. Interestingly enough, the number of childrenand the presence of a grandmother at home do not havesignificant effects. One might expect a high correlationbetween these variables and conservatism, and that it isthe reason for their insignificant values in this case. Thecorrelation matrix in Table 3 shows that this is not the case.As quantifying conservatism is a difficult issue, also calcu-lating the impact of one standard deviation change of thisquality variable on the dependent variable does not giveenough information. For this reason, the differences be-tween the extreme values are calculated. The difference inthe impact of conservatism on FLFP between the urbanfamily with highest conservatism in Turkey and the one withthe lowest conservatism is 0.05 percentage points, which isequivalent to the influence of approximately 2.5 years ofeducation.

In contrast, in rural areas only husband's income has asignificant and negative influence. It is not surprising that inrural areas we do not see any significant effect of conserva-tism, because in rural areas women traditionally have alwaysworked. Even if the rural area is conservative it would notaffect the probability that women work.

Another question might be whether or not the educationcombination of couples is important. In order to understandthis issue, couples are divided into three groups: Those thathave the same education level, those in which husband has ahigher education level, and those in which wife is moreeducated. The results are shown in Tables 4, 5 and 6.

The husband's conservatism index continues to show itsnegative significant effect when the couples have the sameeducation level, but it loses its significance in the case that oneof them has a higher level of education. It might be concludedthat higher education weakens the influence of conservatism.The same test could be also done by interacting educationvariable with conservatism variable, but in the regression wecontrol for the wife's education level, and in order to interact,we should also control for husband's education level whichwould be highly correlated with husband's income.8

Rural

error Coefficient Standard error

0.0001 0.0009−0.0955 ⁎⁎⁎ 0.01460.0158 ⁎⁎⁎ 0.00450.0023 0.0344−0.0693 ⁎⁎⁎ 0.0162−0.0011 0.0102−0.0018 0.01100.0093 0.00670.4331 ⁎⁎⁎ 0.1122Yes11270.0714

icated standard errors are robust.

Table 5OLS regression results for the couples in which husband is more educated.

Urban Rural

Coefficient Standard error Coefficient Standard error

Age 0.0015 ⁎⁎⁎ 0.0004 0.0010 0.0009Age2 0.0058 0.0084 −0.0193 0.0163Education 0.0100 ⁎⁎⁎ 0.0020 −0.0054 0.0038Grandmother 0.0417 0.0310 0.0144 0.0410LogHusbandIncome −0.0219 ⁎⁎ 0.0098 −0.0659 ⁎⁎⁎ 0.0154# of boys 0.0016 0.0048 0.0019 0.0088# of girls −0.0023 0.0037 0.0143 0.0090Conservatism −0.0032 0.0027 −0.0010 0.0051Constant 0.1174 0.0618 0.3141 ⁎⁎⁎ 0.1065Region dummies Yes YesN 1461 1140R2 0.0417 0.0947

Note: ⁎, ⁎⁎, and ⁎⁎⁎ indicate statistical significance at 10, 5 and 1%, respectively. Indicated standard errors are robust.

52 İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

Furthermore, when I analyze the disentangled version ofthe indices for urban areas, both social norm and religionindices take negative and significant values and the influenceof social norms is higher than the one of religion. In ruralareas neither is significant, as conservatism index itself is notsignificant. The results are presented in Table 7.

These results reveal that conservatism is one of thedeterminants of FLFP and social norms play a more importantrole than religion in the formation of the general attitudetowards women employment in the society.

Robustness

An additional test was carried out in order to make theresults more convincing. Throughout the paper, it is claimedthat urbanization plays an important role in the decreasingtrend of female labor force participation, and the mainreason, this paper argues, is the fact that people carry theirsocial attitudes with them when they migrate. In rural areas,men do not oppose of the tradition of womenworking. Whenthese menmigrate to big cities, however, they start to changetheir ideas about the participation decision of their wives.Women's work in the rural area is on the land with otherwomen, but in a city, it means working in an office or afactory with both men and women. This is one of the factors

Table 6OLS regression results for the couples in which wife is more educated.

Urban

Coefficient Standa

Age 0.0019 0.0015Age2 −0.0255 0.0259Education 0.0194 ⁎⁎⁎ 0.0062Grandmother −0.0784 0.0504LogHusbandIncome −0.0359 0.0275# of boys 0.0310 0.0214# of girls 0.0170 0.0181Conservatism −0.0127 0.0077Constant −0.0267 0.2027Region dummies YesN 277R2 0.1235

Note: ⁎, ⁎⁎, and ⁎⁎⁎ indicate statistical significance at 10, 5 and 1%, respectively. Ind

that influence their views about their wives' participation inthe labor force. In order to test this theory, the same analysiswas carried out for the three big cities of Turkey: Istanbul,Ankara and İzmir, which receives most of the migration. Inorder to confirm the argument put forward in this paper, theinfluence of conservatism index for these cities should behigher than the previous analysis. Table 8 presents the resultfor the OLS regression. In the first column whether or not thewomen are living in an urban area is not controlled as thesethree big cities consist mainly of urban regions.9 In thesecond column living in an urban area is also controlled. Inthe third and fourth columns, the results are given only forthe individuals that live in an urban area and rural area,respectively. In these regressions, city dummies are usedinstead of region dummies. The coefficient of the conserva-tism variable in this analysis (−0.0126) is greater than theone of the whole sample (−0.0062) in absolute terms. Asexpected, the influence of conservatism is greater for the bigcities, and there is a positive relationship between urbaniza-tion and conservatism.

Conclusion

This paper explored the role of conservatism on FLFP usingthe 2006 Household Structure Survey in order to determine

Rural

rd error Coefficient Standard error

0.0045 ⁎ 0.0024−0.0178 0.03330.0125 0.01020.1990 0.1368−0.0703 ⁎ 0.03820.0124 0.02670.0058 0.0323−0.0145 0.01600.2860 0.2729Yes1230.2008

icated standard errors are robust.

Table 7Summary OLS regression results for other indices.

Urban Rural

Coefficient Standard error Coefficient Standard error

Religion −0.0045 ⁎⁎ 0.0021 −0.0052 0.0032Social norm −0.0064 ⁎⁎ 0.0030 −0.0049 0.0051

Note: ⁎, ⁎⁎, and ⁎⁎⁎ indicate statistical significance at 10, 5 and 1%,respectively. Indicated standard errors are robust. All the variables that arecontrolled for in the previous analysis are also controlled in this one.

53İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

the main influencing factors. Using this dataset and principlecomponent analysis, two indices, and one index that combinesthe two are formed. The social norm index is calculated bytaking into account the answers to selected questions aboutperceptions in the dataset. The religion index measures theextent to which religious beliefs are determinant in people'semployment decisions. The conservatism index is formed byusing principle component analysis, this time taking intoaccount all the variables used to form the previous indices.

This paper shows that social beliefs and behaviors also playan important role in women's decision to seek employment.Women do not make their decisions in isolation, but areinfluenced by their environment. In conservative and tradi-tional areas, in which men have a greater decision-makingpower, and in which there is greater gender inequality,women tend to remain in the home, as expected by society,and become housewives.

Education plays an important role in women's decisiontowork through two channels. Firstly, a higher level of educationmakes it easier to find a job, and secondly, as shown in thispaper, education weakens the influence of conservatism.

Urbanization is one of the most important reasons for thedecline in female labor force participation, not only because of

Table 8Istanbul, Ankara and İzmir.

I II

Age 0.0014(0.0006)⁎⁎

0.001(0.000

Age2 −0.0314(0.0140)⁎⁎

−0.0(0.014

Education 0.0215(0.0022)⁎⁎⁎

0.021(0.002

Grandmother 0.0279(0.0427)

0.029(0.042

LogHusbandIncome −0.0210(0.0136)

−0.0(0.013

# of boys −0.0000(0.0065)

−0.0(0.006

# of girls −0.0120(0.0076)

−0.0(0.007

Conservatism −0.0126(0.0039)⁎⁎⁎

−0.0(0.003

Urban 0.019(0.025

Constant 0.0670(0.0933)

0.052(0.094

City dummies Yes YesN 1107 1107R2 0.1506 0.151

Note: ⁎, ⁎⁎, and ⁎⁎⁎ indicate statistical significance at 10, 5 and 1%, respectively. Fig

the lack of job opportunities forwomen in cities but also becauseconservative men's preferences differ between urban and ruralareas. In their book, Carkoglu and Kalaycioglu (2009) claim thaturbanization is one of the reasons of the increasing trend inconservatism. This paper shows another link: the link betweenconservatism and female labor force participation. Higherurbanization causes higher conservatism, which leads to lowerfemale labor force participation.

If Turkey prefers to encourage women to increase femalelabor force participation, it first has to give more importanceto education, not only for children but for adults as well.Education does not only mean literacy here; the educationsystem in Turkey should be reformed in such a way that willrecognize women also as people that have equal powers andrights in the family.

Last but not the least, this paper shows that conservatismand social norms play a very important role in FLFP. The onlyway to improve this, again, is education. A nationwideeducation campaign for all inhabitants of the country mightprovide a good background to form a society purged of anybackward ideas about women. This is possible only with theinteraction of the legal authorities, institutions, and NGOswith a multidimensional approach. Though I am aware that itis a kind of utopia, it cannot hurt to try.

Acknowledgments

I would like to express my great appreciation to Assoc.Prof. Saniye Dedeoğlu for her help to improve this paper.Moreover, I would like to offer my special thanks to Dr. Olaf J.De Groot for his comments. I remain indebted to İrfanÇöllüoğlu and Refet Şaban for their help in providing thenecessary datasets. Last but not least, advice given by the two

Urban Rural

46)⁎⁎

0.0015(0.0006)⁎⁎

0.0002(0.0021)

3150)⁎⁎

−0.0291(0.0146)⁎⁎

−0.0630(0.0470)

52)⁎⁎⁎

0.0207(0.0023)⁎⁎⁎

0.0375(0.0103)⁎⁎⁎

33)

0.0020(0.0326)

0.1803(0.0552)

2156)

−0.0194(0.0139)

−0.0711(0.0552)

0007)

−0.0012(0.006)

−0.0106(0.0227)

1227)

−0.0150(0.0077)⁎

0.0345(0.0339)

1239)⁎⁎⁎

−0.0127(0.0041)⁎⁎⁎

−0.0117(0.0100)

64)49)

0.0615(0.0943)

0.3254(0.4051)

Yes Dropped1042 65

0 0.1455 0.1804

ures in parentheses are robust standard errors.

54 İ. Göksel / Women's Studies International Forum 41 (2013) 45–54

anonymous referees has been a great help in forming the finalversion of this paper.

Endnotes

1 Lower gender gap index is associated with larger gender gap in thecountry. When gender gap index equals to one it means equality betweenmales and females. In 2012 Iceland has the highest ranking with an index of86.40.

2 For legal and institutional frameworks in Turkey please check Ilkkaracan(2012).

3 Initially the compulsory education was for 5 years, then in 1999 it isextended to 8 years and in 2012 it became 12 years.

4 Using factor analysis also gives similar results.5 Choosing other combinations of the variables do not change the results

significantly.6 Using factor analysis or just adding up the dummy values and forming

the index do not change the significance of the results.7 The results for the other indices are available from the author.8 Tables A1, A2, A3, A4, A5 and A6 give information about the education

combination of the couples for urban and rural areas, respectively.9 Only 100 women live in rural areas.

References

Acar, Sevil (2008). An analysis of female labor force participation in Turkeyfrom a gender perspective. Paper presented in 4th IUE InternationalStudent Conference, Cooperation, Coordination and Conflict, Izmir, Turkey,April, 2008.

Bisin, Alberto, & Verdier, Thierry (2000). Beyond the melting pot: Culturaltransmission, marriage and the evolution of ethnic and religious traits.Quarterly Journal of Economics, 115(3), 955–988.

Barro, Robert J., & McCleary, Rache M. (2006). Religion and economy. Journalof Economic Perspectives, 20(2), 49–72.

Baslevent, Cem, & Onaran, Ozlem (2002). Are married women in Turkey morelikely to become added or discouraged workers? Labour, 17(3), 439–458.

Becker, Sascha O., & Woessmann, Ludger (2009). Was Weber wrong? Ahuman capital theory of protestant economic history. Quarterly Journalof Economics, 124(2), 531–596.

Burda, Michael, Hamermesh, Daniel S., & Weil, Philippe (2007). Total work,gender and social norms.Working Paper 13000. National Bureau of EconomicResearch.

Cagatay, Nilufer, & Ozler, Sule (1995). Feminization of the labor force: The effectsof long-term development and structural adjustment. World Development,23(11), 1883–1894.

Carkoglu, Ali, & Kalaycioglu, Ersin (2009). The Rising Tide of Conservatism inTurkey. The United States: Palgrave Macmillan.

Council of European Union (2000). Lisbon European Council PresidencyConclusions. http://www.europarl.europa.eu/summits/lis1_en.htm.

Dayioglu, Meltem (2000). Labor market participation of women in Turkey. InFerida Acar, & Aysa Gunes-Ayata (Eds.), Gender and Identity Construction:Women of Central Asia, the Caucasus and Turkey. A. Leiden: Brill.

Dedeoğlu, Saniye (2010). Visible hands–invisible women: Garment production inTurkey. Feminist Economics, 16(4), 1–32.

Esmer, Yılmaz (2012). 2011 Türkiye Değerler Araştırması: Özet Bulgular [2011Turkey's World Value Surveys: Summary Results]. (July 21, 2011).

European and World Values Surveys Four Wave Integrated Data File,1981–2004, v.20060423, 2006. The European Values Study Foundation(www.europeanvalues.nl) and World Values Survey Association (www.worldvaluessurvey.org).

Eyüboğlu, Ayşe, Özar, Şemsa, & Tanrıöver, Hülya T. (2000). Kentlerde Kadınınİş Yaşamına Katılım Sorunlarının Sosyo-Ekonomik ve Kültürel Boyutları[The Socioeconomic and Cultural Dimension of Women's ParticipationDimension of Women's Participation in Work Life in Urban Areas]. Ankara:KSSGM.

Fernandez, Raquel, & Fogli, Alessandra (2009). Culture: An empirical investi-gation of beliefs, work and fertility. American Economic Journal: Macroeco-nomics, 1(1), 146–177.

Fernandez, Raquel, Fogli, Alessandra, & Olivetti, Claudia (2004). Mothers andsons: Preferences formation and female labor force dynamics. QuarterlyJournal of Economics, 119(4), 1249–1299.

Göksel, İdil (2010). Determinants of school attainment in Turkey and theimpact of the extension of compulsory education. In Egypt NationalChild Rights Observatory (ENCRO) (Ed.), Child Poverty and Disparities:Public Policies for Social Justice, Egypt National Child Rights Observatory.

Göksel, İdil (2012). The influence of maternal employment on children's longrun achievement: The case of Izmir. EcoMod 2012 Conference Proceedings.

Guiso, Luigi, Sapienza, Paola, & Zingales, Luigi (2006). Does culture affecteconomic outcomes? Journal of Economic Perspectives, 20(2), 23–48.

Hazan, Moshe, & Maoz, Yishay D. (2002). Women's labor force participationand the dynamics of tradition. Economic Letters, 75(2), 193–198.

Iannaccone, Laurance R. (1998). Introduction to the economics of religion.Journal of Economic Literature, 36, 1465–1496.

Ilkkaracan, Ipek (1998). Göç, Kadının Ekonomik Konumu, Hareket Özgürlüğüve Aile İçi Güç Dinamikleri [Migration, women's economic status, freedomof mobility, and power dynamics in the family]. İktisat, 377(March), 62–67.

Ilkkaracan, Ipek (2012). Why so few women in the labor market in Turkey.Feminist Economics, 18(1), 1–37.

Ince, Meltem, & Demir, Mehmet Hulusi (2006). The determinants of femalelabor force: Empirical evidence for Turkey. Eskisehir Osmangazi Univer-sity Journal of Faculty of Economics and Administrative Sciences, 1, 71–90.

Izdes, Ozge (2012). Economic cycles and gendered employment patterns inTurkey. In Ashwini Deshpande, & Keith Nurse (Eds.), The Global EconomicCrisis and the Developing World: Implications and Prospects for Recovery andGrowth. Routledge Studies in Development Economics.

Ligon, Ethan (1989). The development and the use of a consistent incomemeasure for the general social survey. GSS Methodological, Report No. 64.

Moghadam, Valentine M. (1998). Turkey: From housewife to worker? InValentine M. Moghadam (Ed.),Women, Work and Economic Reform in theMiddle East and North Africa. Boulder: Lynne Reinner Publishers.

Noland, Marcus (2005). Religion and economic performance. World Develop-ment, 33(8), 1215–1232.

Rasul, Imran (2008). Household bargaining over fertility: Theory andevidence from Malaysia. Journal of Development Economics, 86, 215–241.

Spierings, Niels, Smits, Jeroen, & Verloo, Mieke (2010). Micro and macro-level determinants of women's empowerment in 6 Arab countries.Journal of Marriage and Family, 72(5), 1391–1407.

Tabellini, Guido (2005). Culture and institutions: Economic development inthe regions of Europe. Working Paper Series, 1492, : Center for EconomicStudies and Ifo Institute for Economic Research.

Tansel, Aysit (2002). Economic development and female labor force partici-pation in Turkey: Time-series evidence and cross-province estimates.Economic Research Center Working Papers in Economics.

Tansel, Aysit (2005). Public-private employment choice, wage differentials,and gender in Turkey. Economic Development and Cultural Change, 53(2),453–477.

Turkish Statistical Institute (TSI) (2006). Household Structure Survey. http://www.tuik.gov.tr/Veribilgi.do?tb_id=41&ust_id=11

Van der Lippe, Tanja, & Van Dijk, Liset (2002). Comparative research onwomen's empowerment. Annual Review of Sociology, 28, 221–241.

Vendrik, Maarten C. M. (2003). Dynamics of a household norm in femalelabor supply. Journal of Economic Dynamics and Control, 27, 823–841.