This paper provides empirical evidence that there is no absolute convergence between the GDP per capita of the developing countries since 1950. Relying upon recent econometric methodologies (nonstationary long-memory models, wavelet models and time-varying factor representation models), we show that the transition paths to long-run growth are very persistent over time and non-stationary, thereby yielding a variety of potential growth steady states (conditional convergence). Our findings do not support the idea according to which the developing countries share a common factor (such as technology) that eliminates growth divergence in the very long run. Instead, we conclude that growth is an idiosyncratic phenomenon that yields different forms of transitional economic performance: growth tragedy (some countries with an initial low level of per capita income diverge from the richest ones), growth resistance (with many countries experiencing a low speed of growth convergence), and rapid convergence.
growth convergence developing countries long memory wavelets time-varying factor models