An empirical analysis of economic development that relies on international cross-section data is bound to grapple with the question (first featured in Chapter 2) that Arrow (1974) raised in his Nobel Prize lecture. Arrow observed that cross-country inequality in economic development is not readily amenable to the neoclassical explanation of differences in physical and human assets per capita. Two stylized facts about the experience of economic growth beg for an explanation (Kaldor, 1961; Kuznets, 1966). First, output per worker has risen secularly, and rates of productivity growth show no systematic tendency to decline. Thus growth in per capita incomes has been sustained in many instances. Second, there is great variance in growth performance across countries. This divergence does not seem to be random. But it is not autoregressive either, since there does not exist a strong correlation between the level of income at the beginning of the period and the growth outcome at the end.

Arrow's observation inspired a thriving literature that focused on issues of “convergence” (initiated by Baumol, 1986). A subset of this literature is the “endogenous growth” of the Romer-Lucas class of models (Romer, 1986, 1990; Lucas, 1988; Grossman and Helpman, 1992; Barro, 1990,1991; Rivera-Batiz and Romer, 1991; Levine and Renelt, 1992).

The empirical analysis in this chapter builds upon the tradition of this literature. The conventional reduced-form equations of endogenous growth models incorporate variables that can account for the exceptions to the static version of the fundamental theorems of welfare economics, such as the existence of economies of scale.