Hayek versus Harvard: The Case Against Industrial Policy

21 Pages Posted: 21 Jan 2020

Date Written: December 28, 2019


Taleb (2015) shows that Ricardo's law of comparative advantage assumes perfect foresight and therefore zero risk (or a 100% probability of occurring). Ricardo's law is essentially a static rather than a dynamic take on the division of labor. A branch of modern economists (e.g., Rodrik, 2008) has hijacked Ricardo's law under the umbrella of 'industrial policy'; markets do not reach their 'optimum' levels of specialization on their own (that is, they specialize and have comparative advantages in the 'wrong' lines of production), which implies a 'market failure'. This market failure must be corrected by top-down incentives that diversify and grow the 'right' comparative advantages. Nonetheless, this paper shows that top-down industrial policy leads to overspecialization and over-optimizing, which increase the odds and impact of adverse events. I conclude that there is a delicate line between overspecialization and diversification: the optimum, which constantly shifts, can only be discovered bottom-up by individual risk-takers rather than top-down by technocrats. As a result, most variants of industrial policy (e.g., Hidalgo & Hausmann, 2009; Rodrik, 2008) are bound to fragilize emerging economies.

Keywords: industrial policy, comparative advantage, overoptimization, overspecialization, fragility, economic growth, gross domestic product, crisis

JEL Classification: L50, L52, F43, F63, H81, L26, O21, O24, O25, O57

Suggested Citation

Dirkmaat, Olav, Hayek versus Harvard: The Case Against Industrial Policy (December 28, 2019). Available at SSRN: https://ssrn.com/abstract=3510719 or http://dx.doi.org/10.2139/ssrn.3510719

Olav Dirkmaat (Contact Author)

Francisco Marroquín University ( email )

Calle Manuel F. Ayau (6a calle final), zona 10
Guatemala City, 01010

HOME PAGE: http://www.ufm.edu

Do you have negative results from your research you’d like to share?

Paper statistics

Abstract Views
PlumX Metrics