1 post • Page 1 of 1
- Posts: 8
- Joined: Tue Jun 05, 2018 8:28 pm
Hi, I just want to let you know that I enjoyed very much your presentation. I think that you control quite well for demand-side considerations in your choice of controls to do the propensity-score matching. Your variables are mainly household characteristics that capture demand-side effects regarding children' education. The hope then is to isolate the effect of migration by a parent on the level of education of children. However, households may face different supply-side constraints regarding the provision of schooling services. One would think that distance between household location and the location of relevant schools, given the age of children, would be an important factor to consider. For example, distance between a high school and household location is pertinent if the household has a child old enough to attend high school. For such households, proximity to a primary school is not pertinent, but proximity to a high school would help the child achieve a higher level of education. The implication for the matching is that two households that are identical on all demand-side factors, like household size and education of the parents, may desire the same level of education for their children. However, the years of schooling of their children may differ in the end because one household is located close to a school and the other is not. Parents migrate because the household is located far from employment opportunities. Could it be that schools or good schools are also distant from the location of households with migrating parents? Can you treat distance to school as a pure random variable with the same distribution for households with no migrating parents and households with at least a migrating parent? If you could account for this sort of effect, your analysis would be strengthened.
Users browsing this forum: No registered users and 1 guest