IZA DP No. 14735: Does Education Really Cause Domestic Violence? Revisiting the Turkish Data
published as 'Compulsory Schooling Reform and Intimate Partner Violence in Turkey' in: European Economic Review, 2022, 150, 104313.
Using the 2008 Turkish National Survey of Domestic Violence against Women (NSDVW) and the 1997 compulsory schooling policy as an instrument for schooling, Erten and Keskin (2018, henceforth EK), published in AEJâ€“Applied Economics, find that women's education increases the psychological violence and financial control behavior that they face from their partners. The authors also claim that the incidence of financial control behavior rises because women become more likely to be employedâ€”supporting the instrumental violence hypothesis. They present this evidence only for women who live in what they call â€œrural areas during childhoodâ€. We first show that the evidence EK provideâ€”which exists only for childhood rural areasâ€”is a result of their misclassification of the rural areas variable. We show that once this variable is defined properly, the evidence for their findings vanishes. Second, ignoring the misclassification of the rural status variable, we demonstrate a number of serious flaws in their empirical analysis: (i) selection bias resulting from the policy altering the composition of women in their sample, (ii) failure of the main identification assumption of RDD for some key outcomes, (iii) failure of the exclusion restriction assumption, (iv) inconsistency in the definition of employment variable across men and women (and a problematic definition of employment of women), (v) elementary mistakes in data cleaning, RDD estimation, and interpretation of the estimates. In addition, the evidence for urban areas contradicts the hypothesis they claim to hold for rural areas. Then, we examine the policy effect on domestic violence outcomes using both 2008 and 2014 TNSDVW datasets. We find null policy effects on psychological violence and almost null effects on women's employment, and positive but statistically insignificant effects on partners' financial control behavior. Hence, our findings do not support the instrumental violence hypothesis, and this holds true for the rural sample as well. The only robust evidence the data provide is that the policy reduces physical violence for women with rural childhood residence.