Using the Greek Civil War to Re-estimate the Returns to Schooling
For those affected by the civil war, the returns to schooling are higher than the corresponding traditional estimate
Studies have attempted to estimate the returns to education in Greece using a variety of methods and controls. In an early analysis of the 1977 labor market survey the rate of return to schooling in Greece was estimated at relatively low 5.8 percent, which is lower than estimates for Greece from the 1960s. A variety of empirical studies have used changes in supply as instruments in instrumental variables estimations of the returns to schooling. The main finding is that such estimates of returns to schooling are typically larger than ordinary least squares estimates. This suggests that the marginal returns among the low-education subgroups (which are typically affected by supply changes) tend to be relatively high, reflecting their high marginal costs of schooling rather than low ability.
One event that occurred and which could be useful for understanding the returns to education is the Greek civil war, 1946 to 1949. This led to many deaths, economic losses, and severe reductions in schooling expenditures and attendance. In Returns to Education in Greece: Evidence from the 1977 Labor Market Survey using the Greek Civil War as an Instrument (IZA Discussion Paper No. 15541) Harry Patrinos uses the 1977 survey and the Greek civil war as an instrument to re-estimate the returns to schooling in Greece. This is the local average treatment effect framework. Note that given the instrument, in this case the treatment is defined as a reduction of schooling. It is further conjectured that none of the other groups is likely to be affected by this instrument. The Greek Civil War led to a significant reduction in schooling between 1950 and 1955.
Given that the Greek civil war took place from 1946-1949, and the levels of schooling dropped precipitously in the 1950s, the instrument is operationalized by setting it so that those 22 to 27 years of age, that is, who were 10 years of age during the civil war years, are deemed to be affected.
The returns to schooling are estimated for the marginal student affected by the civil war. The estimate of the returns to schooling in Greece in 1977, using the civil was as an instrument, is higher than the existing non-experimental returns to education estimates at the time. The IV estimate is 8.0 percent, which is 26 percent higher than the corresponding OLS estimator. These results can easily be reconciled with the idea of heterogeneous returns to education. Moreover, the LATE concept gives a meaning to these heterogeneous returns: the LATE measures exactly the returns for the group that changes treatment status because of the war. These are predominantly poor individuals, with returns that are probably higher at the margin.
These results are consistent with others who have used civil war as IV and the many studies that use instruments to estimate the returns to schooling. However, since the instrument means less schooling, then this suggests a high cost of war. The indicator of war constraints to schooling result in somewhat higher returns to education for the affected group.