Endogeneity in Estimating the Child Labor Model with Censored Data: An IV-TOBIT Approach

In the decent work literature abolishing child labor is one of the main elements of rights at work. According to the convention 138 of the International Labor Organization any form of work that harms physical mental or moral wellbeing of the child or threatens the child’s morality is considered child labor and must not be done by any children less than 18 years of age. However in many developing countries due to extreme poverty and unfavorable labor market conditions a considerable number of households are obligated to be content with their children’s participation in the labor market. In this paper the effects of the household’s characteristics on the probability and the amount of child labor is studied. In order to do so the characteristics of the child mother head of the household and the household are considered as the explaining variables. The model applied in this study is an IV-TOBIT regression model. The child’s participation in income yielding activities for the household results in an endogeneity bias in the TOBIT model. Therefore the head of the household’s net income is used as an Instrumental Variable. The data in this study are extracted from the Households’ Income–Expenditure Survey of the year 2013. The group under study are the children between 6 and 18 years of age. Near 25 thousand children were studied. According to the results being male has a significant positive effect on the increase in the hours of work done by a child and its probability. Furthermore the head’s activity status and employment status and higher education levels of the mother decrease child labor.