A Simple Evolutionary Model of Invention and Growth Takeoffs

Abstract

This thesis studies a simple evolutionary model of invention to understand how different behavioral limitations would have affected the transition to modern growth. In neoclassical models of invention, “entrepreneur-inventor”s are fully informed, fully rational, and able to solve mathematical optimization problems. In contrast, this thesis assumes that they are not fully informed, they are subject to bounded rationality constraints, and they cannot solve mathematical optimization problems. Within this structure, the thesis uses a genetic algorithm to model how a society—where the norm (or the status quo) is initially not to spend valuable resources to invention—can learn that invention is actually optimal. In other words, this thesis studies how a technologically stagnant society can converge to a growth equilibrium. Three exogenous factors are mutation (how tolerant the society is to deviant entrepreneurs), elite persistence (how effective the knowledge transmission is across generations), and the size of population. These potentially affect two model outcomes, i.e., how long the transition is and whether the society can converge to the neoclassical benchmark exactly. Results show the following: (i) Mutation is not very strongly correlated with the model outcomes, but higher mutation rates are observed along with faster convergence in some specifications. (ii) Elite persistence does not have a monotone effect on the duration of convergence. (iii) Societies generally exhibit variation around the neoclassical optimum in terms of equilibrium values. (iv) There is a very strong scale effect of population size; larger populations converge significantly faster, and the degree of variation around the neoclassical equilibrium is much smaller for larger populations. Even in the best-case scenario—i.e., high mutation rates, high elite counts, and large populations—the evolutionary model converges to the neoclassical equilibrium in 51 generations.