Which Explains Income Differences across Advanced Economies: Solow Model or MRW Specifications? Fresh Evidence From Bayesian Monte Carlo Simulations

Previous frequentist research exploring the Solow growth model and its MRW specification has grappled with persistent challenges such as multicollinearity and reverse causality inherent in neoclassical growth models. This study aims to showcase the Bayesian approach’s advantages in determining whether the Solow or MRW model better explains income variations across advanced countries. To this end, we apply a hybrid Metropolis-Hastings algorithm within the Bayesian non-linear framework to a panel of 38 advanced countries from 1970 to 2019. The results yield significant insights: Firstly, by integrating informative priors that capture prior beliefs on variable interactions, Bayesian inference effectively addresses potential multicollinearity and reverse causation. Secondly, Bayesian estimation introduces valuable shrinkage effects that enhance estimation accuracy. When subject to Bayesian shrinkage, broad capital elasticities more closely align with true values, highlighting the influence of technological advancement and latent factors often overlooked by traditional observation. Lastly, our analysis reveals that the augmented MRW model, accounting for heterogeneous technology growth and depreciation rates, is the best fit. This study provides a reliable empirical foundation for shaping comprehensive policies aimed at fostering sustained economic growth through technological progress alongside the accumulation of physical and human capital.