How important is human capital? A quantitative theory assessment of world income inequality

by Andrés Erosa, Tatyana Koreshkova and Diego Restuccia

We develop a quantitative theory of human capital investments in order to evaluate the magnitude of cross-country differences in total factor productivity (TFP) that explains the variation in per-capita incomes across countries. We build a heterogeneous-agent economy with cross-sectional variation in ability, schooling, and expenditures on schooling quality. By embedding our analysis in a growth model with tradable and non-tradable sectors, we model sectorial productivity differences across countries, as documented in Hsieh and Klenow (2007). The parameters governing human capital production and random ability and taste processes are restricted by a set of cross-sectional data moments such as variances and intergenerational correlations of earnings and schooling, as well as slope coefficient and R2 in a Mincer regression. Our main finding is that human capital accumulation strongly amplifies TFP differences across countries: To explain a 20-fold difference in the output per worker the model requires a 5-fold difference in the TFP of the tradable sector, versus an 18-fold difference if human capital is fixed across countries. Moreover, we find that sectorial productivity differences play a prominent role in quantitative implications of the theory.

There is plenty of empirical literature trying to establish the importance of human capital in cross-country income differences, usually neglecting that human capital may be endogenous. Clearly, a more structural approach is warranted and this paper delivers this, along with a very rich model. Will this paper convince the cross-country regression enthousiasts?

2 Responses to How important is human capital? A quantitative theory assessment of world income inequality

  1. Alberto says:

    Cross-country regressions have been abused to the point that I do not believe anything coming out of them. As for VARs, the best they could do is highlight correlations, but there is no way they can reliably convey causation.

    Structural estimation is required, and for that a proper model is needed. This paper has a model that is much more complicated than would be necessary, and there is no data set in the world that could be used to estimate it. Make it simpler!

  2. NotMankiwRomerWeil says:

    I agree wholeheartedly that cross-country regressions have been abused. But I disagree with Alberto that we absolutely need to regress something. What this paper does is take a model seriously, make it fit some dimensions of the data and see how far we can get with that. Given the availability of good data (or the lack thereof), this is the best you can do, in particular for developing economies.

    In this regard, Christian should have mentioned the pioneering work he has done in this regard with Doug Gollin on the impact of malaria. This is an area where there is no reliable data and cross-country regressions have no meaning. Yet he and Doug have been able to come up with very powerful results by using structure and reasonable calibration. Malaria: Disease Impacts and Long-Run Income Differences. I think this is the way of future research in development and growth, until we get good data at the least.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s

<span>%d</span> bloggers like this: