The world elasticity of labor substitution across education levels.

We estimate the world elasticity of substitution between labor types on a very well-known data set. Using enrollment ratios and government expenditures on education as instruments, the plausible elasticities varying from 2.00 to 3.21 for the 'college-completed' definition suggests imperfect substitution between skilled and unskilled labor. Considerably higher values are found for the 'secondary' and 'primary-completed' definitions. Overall, the higher the threshold for defining skilled labor, the lower the elasticity and less likely is the switch between types of labor. These findings complement micro-evidence and support stronger links between productivity and output for the 'college educated' labor definition. [ABSTRACT FROM AUTHOR]