Knowledge obsolescence, human capital inequality, and growth: A network perspective in an automated knowledge society

This paper suggests a micro-founded growth theory of human capital that incorporates three important ingredients: i) learning in a knowledge network, ii) possible skill-down-grading due to knowledge obsolescence, and, iii) fear of technological unemployment due to automation. Heterogeneous agents (optimally) split their time between learning-by-exchanging knowledge or working in the final goods sector. On the aggregate level, our benchmark model shows that learning and the degree of connectivity within the knowledge network directly impact the growth rate of the economy. Moreover, we show the existence of a poverty trap in which society stagnates due to an insufficient level of human capital that is in particular governed by the degree of knowledge obsolescence. In an extension, we control for the fact that learning is a cognitively demanding task associated with learning errors due to cognitive constraints. Therefore, two groups of agents are distinguished: Cognitively constrained and rational optimizers, where both can switch endogenously between a low and high-skilled state. We use this extension to numerically quantify the effects of cognitive constraints on human capital inequality. Inter alia, we show that a knowledge obsolescence shock has transitional as well as long-run negative effects on human capital inequality, where in relative terms, cognitively constrained agents are more affected than their rational counterparts.