Bayesian estimation of a non-linear Gaussian state space stochastic frontier model: Assessing effects of external factors on productivity change

Kelly Trinh | The University of Queensland 

We propose a non-linear Gaussian state space stochastic frontier model to jointly estimate the impacts of external factors on production process: the impact on the attainable production set (the shift or the shape of the frontier) and the impact on the level technical inefficiency (the movement away/towards the frontier).

The proposed model does not require a prior specification of the distribution of technical inefficiency, which is often assumed to be a truncated normal or an exponential distribution. Furthermore, the model allows for the effect of external factors on the level technical inefficiency to vary across time and individuals. We use the proposed model to investigate how the external factors (i.e., foreign direct investment (FDI), economic freedom index) might have an influence on production process.

By using a dataset of 21 countries over time period 1995-2011, we find that FDI plays an important role as influencing the shift of the common production frontier and the movement towards the frontier for some countries. The impact on the level technical efficiency does not seem to vary much over time, but seems to vary across the countries. With regard to economic freedom index, its impact on the shift of the frontier is more profound than its impact on the level technical efficiency, which is found to be insignificant for some countries.