Multiple-steady-state growth models explaining twin-peak empirics?

Abstract

The explanation of twin peak empirics through multiple-steady-state growth models has one serious implication: Whenever a model generates twin peaks in GDP per capita it also generates twin peaks in other variables. We check for some multiple steadystate models whether or not they have twin peaks in the other variables besides GDP per capita. It turns out that the required twin peaks do not exist for the textbook version of the population trap model but a modified version cannot be dismissed. Stiglitz’ (1987) endogenous growth model requires twin peaks in savings per capita, which are not found in the data. Models of human capital accumulation and endogenous population growth are closer to the idea of twin peaks because twin peaks in population growth rates do exist, but human capital accumulation is not exactly distributed in a way twin peaks would require but that distribution rather follows a structure of one peak with one or two stairs depending on the human capital variable.