The weight of circumstances in the inequality of opportunity in Mexico: an estimation over a wide set based on machine learning

In this research I calculate the weight of each circumstance (from a wide set) on the ex-ante inequality of opportunity, at both national and regional levels. Conditional inference trees are employed to identify de different types determined by these circumstances, and Shapley-value decomposition technique is used for estimating the weights. The results confirm the wealth at origin as the main circumstance (55%). Nevertheless, is important to note the significant weight found in variables such as skin tone (10%), as well as the participation of a variable so far not considered in previous research works, that is the job position of the main economic provider at home (3%). It is highlighted that at a regional level, there are substantial differences in the weight of each circumstance, aside from the wealth at origin or the maximum parental attainment.