Essays on non-market returns to education
: evidence from a natural experiment in Turkey

  • Mustafa Ozer

    Student thesis: Doctoral Thesis


    As a result of the political events in the mid-1990s, the compulsory schooling in Turkey was extended to 8 from 5 years. Cohorts who were at the fourth grade or lower had to continue 8 years of compulsory education. This reform generated a natural experiment, and therefore, enables researchers to obtain endogeneity-robust estimates of market and non-market returns of education. In this thesis, I used this reform to investigate some nonmarket returns of education.
    Chapter 2 is the first study to investigate the causal impact of maternal schooling on
    childhood immunisation rates. I used the exposure to the Compulsory Education Law (CEL), enacted in Turkey in 1997, by date of birth and the differentiation in its implementation across regions as an instrument for the schooling of young mothers. CEL generated an exogenous change in compulsory education from 5 to 8 years for women born after 1986. Using the CEL as an instrument, this chapter finds that an increase in the education of mothers significantly increases the coverage of the third (full) dose of DPT and Hepatitis B for their children. Furthermore, the chapter investigates the causality between maternal education and a range of channels affecting immunisation rates suggested in the literature. I find that education increases the age of first marriage and birth, changes women`s and their spouse`s labour market status and significantly effects women`s attitude towards spousal violence against women and gender discrimination in a manner that empowers women.
    Chapter 3 provides empirical evidence regarding the causality between education and mental health in the context of a large emerging economy, Turkey. I exploit differentiation in the exposure to the Compulsory Education Law (CEL) change by the date of birth as an instrument for schooling of young women. When the CEL is used as an instrument for schooling of women, I find that schooling increases the incidence of some mental health disorders. These findings seem quite robust to alterations in the regression definitions and to the inclusion of additional individual-level control variables such as having one`s own income, employment status, marital status, occupation, domestic abuse history in the family, and rural/urban status. I also find a heterogeneous effect of the CEL on educational attainment of women depending on residence status and whether their birth mother had been abused by her partner. Moreover, heterogeneous effect tests also indicate that the effect of education differs by personal income and labour market status of women.
    Chapter 4 is the first study to investigate the causal effect of the husband`s education on violent and abusive behaviour against their wife. To do this, I use a natural experiment, an education reform increasing compulsory schooling from five to eight years in Turkey, as an instrument to obtain endogeneity-robust estimates. I find that the husband`s education lowers the probability of suffering physical, emotional and economic violence as well as experiencing socially unacceptable behaviour. The only aspect of violence not affected by the spouse`s education is sexual violence. I also find that women whose mothers or whose husbands’ mothers experienced domestic violence are more likely to suffer violence themselves.
    Date of AwardOct 2016
    Original languageEnglish
    Awarding Institution
    • University of Portsmouth
    SupervisorAnsgar Wohlschlegel (Supervisor), Andy Thorpe (Supervisor) & Gianpiero Torrisi (Supervisor)

    Cite this