Towards an empirically adequate model of economic growth
The current literature is only slowly moving towards an empirically adequate theory (or collection of models) that is adequate to explain the entire range of facts about developing country growth. That is, neither the traditional Solow nor the augmented neo-classical model nor the "endogenous" growth have proved to be particularly fruitful in explaining the variety of growth experiences (except perhaps in the most mechanical of decompositions of proximate causes of growth).
In the World Bank publication headed by Roberto Zagha on the Lessons of the 1990s I contributed two chapters:
- Chapter 2: Grist and the Mill for the Lessons of the 1990s. This reviews the basic facts of the 1990s, both disappointments and pleasant surprises. This chapter attempts to document the experiences of growth that have been puzzling--for instance:
- why has all of the reform in Latin America produced so little growth (figure)?
- why was the "transition recession" in the FSU/EE countries so deep and so long?
- why has India grown so fast for so long (and in particular what accelerated its growth)?
- Chapter 8: Policy Reform and Growth: What have we learned? This is the overview chapter that draws lessons across trade, financial sector, privatization.
- A Toy Collection, A Socialist Star and a Democratic Dud? (Growth Theory, Vietnam and the Philippines). (June 2002). This paper was written for a volume of analytic studies of country growth by Dani Rodrik. This paper stress the entire range of empirical facts that an empirically adequate model of developing country growth would have to encompass (including (1) long-run divergence on average, (2) steady, non-accelerating growth of the leading countries, (3) some experiences of extremely rapid growth, (4) large within country volatility in medium run growth rates over time.
- I suggest a "states" and "transitions" model of growth as an encompassing framework. That is, a single linear model is obviously incapable of explaining the entire set of stylized facts, but a model in which there are "states" of growth--with perhaps linear dynamics of growth with respect to the usual correlates within "states"--and "transitions" across states--so that there can be sudden accelerations or decelerations that are not themselves determined by the same linear dynamics. The usual linear coefficients on long-period growth regressions could then be some entangled mix of within state and between state effects--which explains the existence of robust partial correlations but also their failings.
- What's the Big Idea?
with David Lindauer. (June 2002). This paper, in Economia (the new "Brookings Papers for Latin America") charts the changing strategies for growth and how they have grown naturally out of the obvious "lessons" of experience--but failed.
- A Conclusion to Cross National Growth Research: A Foreward "To the Countries Themselves"
(June 2001). This paper, written as the conclusion to a volume of regional studies of economic growth by the Global Development Network surveys the progress, and the difficulties, with existing approaches modeling and investigating economic growth. In particular it stresses that growth regressions can almost certainly never reveal policy impacts--because a policy is a mapping from states of the world to policy actions so any empirical measure of "policy" outcomes (e.g. the budget deficit) combines across realizations of states of the world and policy actions. This is now published in the book Explaining Economic Growth edited by Lyn Squire and Gary McMahon, which is itself an interesting book.