There has been a shift in conversation over the last years and at least in US they are claiming that the highest variance in student performance can be attributed to socioeconomic status and developmental environment. So it would seem like selecting based on school performance could make things worse. One could make the case that this would really hurt very smart/ high-performing individuals from low socio-economic status families but most people (and rational policy) care about helping/ making an impact at the population level.