The Long-Run Education Cost of World War II. Example of Local Average Treatment Effect Estimation

An important component of the long-run cost of a war is the loss of human capital suffered by children in schooling age who receive less education because of the war. This paper shows that in the European countries involved in WWII, children who were ten years old during the conflict were significantly less likely to proceed into higher education. On the contrary, we find no effect for individuals in the same cohorts living in countries not involved in the war. Using data for Austria, Germany, Sweden and Switzerland, we estimate the cost of the war in terms of earning losses suffered by those individuals who did not reach higher education because of the conflict and we compute the implied loss of GDP for their countries. In order to identify this cost, we interpret our Instrumental Variable estimates as measures of the Local Average Treatment Effects (Angrist and Imbens, 1994) of education connected to our war instruments. Inasmuch as WWII has caused an increase of liquidity constraints for families with children in schooling age (e.g. the absence of the father), our estimates may be considered as measures of the long-lasting income losses that could be avoided in peaceful times by exogenously increasing the educational attainment of children in families subject to constraints similar to those caused by the war (e.g. families with single mothers).