The models presented so far focused on physical and human capital accumulation.

Eco­nomic growth is generated by exogenous technological progress. While such models are useful in thinking about sources of income differences among countries that have (free) access to the same set of technologies, they do not generate sustained long-run growth (of the country or of the world economy) and have relatively little to say about sources of technology dif­ferences.

We have already encountered the AK model in Chapter 2. This model relaxed one of the key assumptions on the aggregate production function of the economy (Assumption 2) and prevented diminishing returns to capital. Consequently, continuous capital accumulation could act as the engine of sustained economic growth. In this chapter we start with a neoclassical version of the AK model, which not only shows the possibility of endogenous growth in the neoclassical growth model, but also provides us with a very tractable model that find applications in many areas. This model is not without shortcomings, however. The most major one is that capital is the only (or essentially the only) factor of production, and asymptotically, the share of national income accruing to capital tends to 1. This, however, is not an essential feature of neoclassical endogenous growth models. We present two different two-sector endogenous growth models, which behave very similarly to the baseline AK model, but avoid this counterfactual prediction.

We conclude this section with a presentation of Paul Romer’s (1986) path breaking article. In many ways, Romer’s paper started the endogenous growth literature and rejuvenated the interest in economic growth among economists. While Romer’s objective was to model “technological change,” he achieved this by introducing technological spillovers—similar to those we encountered in Chapter 10. Consequently, while the competitive equilibrium of Romer’s model is not Pareto optimal and the engine of economic growth can be interpreted as a form “knowledge accumulation,” in many ways the model is still neoclassical in nature. In particular, we will see that in reduced-form it is very similar to the baseline AK model (except its welfare implications).

11.1.