The effect of fertility and son preference on intrahousehold resource allocation in türkiye

The aim of this study is to examine the mutual causality relationships between son preference, fertility, and intra-household resource allocation. The study uses the Household Budget Statistics, Group B Micro Data Set of the Turkish Statistical Institute. The primary objective of the study is to verify the existence of son preference, as documented in relevant literature in Türkiye, and to investigate this phenomenon further using a more recent and different data set (2013-2019). Additionally, the study examines how agricultural land ownership (fields, vineyards, and greenhouses) influences son preference and fertility decisions. The effect of son preference on fertility is first analyzed using the Linear Probability Model and the Probit Model with the Parity Progression Method, depending on the sex composition of children. Then, this analysis is expanded to include the Heckman Correction Model to prevent sample selection bias. The second aim of the study is to examine in what direction and to what extent the exogenous change in fertility caused by the first-born child of the household being a girl affects the share of education expenditures for children in total expenditures. For this reason, the study examines how human capital investments in children are influenced by exogenous changes in the number of children and how parents adjust the allocation of resources within the household as a result. This analysis is estimated by employing the Instrumental Variable Method for the relevant data set for the years 2015-2019. According to the findings, having a first-born girl increases the likelihood of families having a second child by 5%. In families where the first two children are girls, the possibility of having a third child increases by 25%. It has been observed that in households without any sons, the tendency to prefer a son intensifies as the number of daughters increases. In this manner, our results confirm the findings that families in Türkiye exhibit son-biased fertility behaviors, parallel to previous studies in the literature. In addition to these findings, having the same sex as the first two children increases the likelihood of families having a third child by 20%. This result indicates that households in Türkiye do not only have a son-biased tendency but also desire a sex mix of children. The study also examined how the effect of son preference on fertility differs between households with and without agricultural land ownership, and it was found that the effects are significantly higher in households with agricultural land ownership. This phenomenon stems from parents' desire to secure intergenerational land transmission through male heirs, thereby preserving patrimony within the patrilineal line when they allocate resources within the household. In this case, it is seen that families in Türkiye exhibit a son-biased attitude in their inheritance preferences. The study also examined whether excluding households without children from the sample caused any sample selection bias by running the Heckman Correction Model, and it was found that there was a sample selection bias. According to the Heckman OLS estimation results, the standard OLS results are subject to a downward bias. Therefore, having a first-born girl increases fertility by 6%. In the second empirical analysis, having a first-born girl increases fertility by 6%, as in the first empirical analysis. This exogenous increase in fertility, resulting from son-biased fertility behavior, appears to have increased the share of children's education expenditures in total household expenditures by 8%. It is observed that an exogenous increase in the number of children decreases the share of expenditures on alcohol and cigarettes by 5% and the share of expenditures on fathers' clothing by 15%. However, it increases the share of mothers' clothing expenditures by 21%. The reason behind the increase in mothers' clothing expenditures suggests that many women may increase their clothing expenditures due to postnatal weight gain. It can also be said that the elasticity of women's clothing is more rigid than that of men's clothing. In light of these results, it appears that an increase in the number of children is subsidized by the father's clothing expenditures and household alcohol and cigarette expenditures, but is constrained by the mother's clothing expenditures, in order not to compromise the educational investments made for the children.