Localized Technological Change and Path-Dependent Growth

In recent years the theory of macroeconomic growth has seen an expanding literature building upon the idea that technological change is localized (technology-specific) to investigate various phenomena such as leapfrogging, take-off, and social mobility. In this paper I explore the relationship between localized technological change and dependence on history of long-run aggregate output growth. The growth model I set forth show that, subject to mild assumptions on the stochastic process representing exogenous environment, path-dependence of aggregate output growth is a robust property of an economic system with localized learning-by-doing and diversity of technological opportunity. In general there are multiple steady states (with fast vs. slow growth properties) whose selection depends on the sequence of historical events: "a priori" all steady states are attainable but only one of them emerges as the results of history; two economies with identical fundamentals can thus evolve towards different steady states just because of the sequence of historical (possibly small) events. Superior technologies might not be adopted: This happens because the opportunity cost of abandoning an adopted technology is higher ( and the probability of switching even to a dynamically superior path is lower) the higher the accumulation of experience in the adopted technology. However growth turns out to be ergodic if agents are sufficiently heterogeneous in terms of technological capabilities. The same applies if technological opportunity is very different across technologies, but this is hardly the case when there are many "young" technologies.