Education, Growth and Income Inequality
Coen Teulings and Thijs van Rens
Estimates of the effect of education on GDP (the social return to education) have been hard to reconcile with micro evidence on the private return. We present a simple explanation that combines two ideas: imperfect substitution between worker types and endogenous skill biased technological progress. When types of workers are imperfect substitutes, the supply of human capital is negatively related to its return, and a higher education level compresses wage differentials. We use cross-country panel data on income inequality to estimate the private return and GDP data to estimate the social return. The results show that the private return falls by 2 percentage points when the average education level increases by a year, which is consistent with Katz and Murphy's  estimate of the elasticity of substitution between worker types. We find no evidence for dynamics in the private return, and certainly not for a reversal of the negative effect as described in Acemoglu . The short run social return equals the private return.
Published in the Review of Economics and Statistics, 90(1), pp.89-104.
Published version: February 2008 [download pdf]
One-but-last version: July 2006 [download pdf]
First version: January 2001
The previous (February 2003) version is still available as CEPR Discussion Paper 3863 or Tinbergen Institute Discussion Paper 02-001/3 or here. The new version is substantially shorter and we have updated our estimates using new data until 2000.
A non-technical summary is available on the Barcelona GSE website.
Download dataset and Stata codes for this paper:
After publication of this paper, I became aware of the following article that uses cross-country panel data to estimate the social return to education from an aggregate Mincer equation and documents evidence for diminishing returns.
Ram, Rati (1996). Level of Development and Rates of Return to Schooling: Some Estimates from Multicountry data, Economic Development and Cultural Change, 44(4), pp.839-857.
Thijs van Rens | IDEAS/RePEc | Google Scholar | ResearchGate