Neoclassical, semi-endogenous or endogenous growth theory? Evidence based on new structural change tests

One of the prevalent topics in the economic growth literature is the debate be tween neoclassical, semi-endogenous and endogenous growth theory for the model that best describes the data. An important part of this discussion can be summa rized in three mutually exclusive hypotheses: the “constant trend”, “level shift” and “slope shift” hypothesis. The objective of this paper is to classify countries according to each of these hypotheses and to analyze which of the growth theories seems to be favored. We approach this problem in two-steps: first, the number and the timing of trend breaks are determined using the approach in Nunes and Sobreira (2010); and second, conditional on the estimated number of breaks, break dates and coefficients identified, a statistical framework is introduced to test for general linear restrictions on the coefficients of the linear disjoint broken trend model. Here, we prove a general result that, under certain conditions, a standard F-statistic to test for additional parameter restrictions, given the first step esti mated partition, converges asymptotically in distribution to the usual chi-square distribution. We further show how the aforementioned hypotheses can be formu lated as linear restrictions on the parameters of the breaking trend model and apply the methodology to per capita output of an extensive list of countries. All of our tests are robust as to whether the data are I(0) or I(1) surpassing technical and methodological concerns on previous empirical evidence.