The model presented in the previous chapter predicts that an initial position of income inequality reduces access to education and, as a consequence, future income distribution. It will be interesting to establish whether this prediction has any empirical validation in the real world. In the present chapter we will review some macro-evidence providing support for this proposition. Starting from an optimal demand for education, where the years of education depend on family income among other things, we derive two testable predictions in the analysis of aggregate data on school enrolments: a negative (linear) dependence of enrolment rates on the Gini inequality measure on income distribution; and a positive dependence on public resources invested in education and/or on the skill premium in the labour market. On the one hand, optimal demand for schooling could be positively associated with unobservable ability; if ability is intergenerationally transmitted, we would get a positive correlation between family income and children's schooling. On the other hand, if liquidity constraints prevented school attendance by children from poor families, we would observe a correlation between educational choices and family incomes, at least in the lower tail of income distribution. We prove that, under the given assumptions, enrolment rates and measures of income inequality (namely the Gini index) should be negatively correlated. However, following the liquidity constraint explanation, the Gini coefficient should be negatively correlated whenever missing financial markets create a barrier preventing access to education by the poorest families.