A Note on Endogenous Growth and Scale Effects : Reconciling Recent and (Very) Long-Run Evidence

The endogenous growth models of the early 1990's predicted scale effects: larger population - more resources devoted to R&D - faster productivity growth. As pointed out by Jones (1995) this prediction is not consistent with the experience of industrialized countries like the U.S. over the post-WWII period. However, as argued persuasively by Kremer (1993), there is evidence for the existence of scale effects in technology creation over the very long-run, i.e. centuries or thousands of years. This paper reconciles this recent vs. long-run evidence by extending Howitt's (1999) model of endogenous growth without scale effects to include a subsistence constraint in consumption. When an economy is poor and the subsistence constraint is binding, innovation tends to be concentrated in a limited number of 'subsistence' sectors like basic food stuffs, producing scale effects initially. As the economy grows richer and the constraint is relaxed, consumers begin to value more other types of goods and services, R&D becomes spread over an increasing number of sectors, and long-run scale effects vanish