Redistributive Policy and Endogenous TFP Disparity Across Countries with a Special Focus on the Us, Japan and Australia

Abstract

The traditional emphasis on breaking down output growth according to growth in inputs and growth in TFP is misplaced. More appropriately, we should break it down according to changes in and technology. These two are equilibrium concepts, rather than purely technical properties of production functions. The influence of redistributive taxation (on growth) is nonlinear and depends crucially on the current state of the economy. How an economy responds to changes in the average marginal tax rate of income reveals its efficiency state. We establish the above three key results in a neoclassical, dynamic general equilibrium model in which the institutional and technological factors, as well as redistributive policy, endogenously determine the total factor productivity (TFP) of a country. Economists point towards cross-country differences in TFP to explain wide differences in income per capita across countries. However, econometric exercise proves to be challenging in the absences of a theoretical model which could map institutional and policy parameters into TFP and accumulated inputs as an equilibrium outcome. In this paper, we take up this challenge. The model allows us to break down TFP into two key components: technology - which identifies the production possibility frontier and - which determines the location of the economy inside that frontier. We calibrate that model to three separate economies: the US, Japan, and Australia. The numerical analysis of the data from those three calibrated economies highlight that TFP differences play a significant role in explaining the differences in output per capita across U.S., Japan, and Australia. We then conduct simulation experiments based on random selections of a set of parameters with suitably specified distributions to match the per capita income distribution among the countries around the world and generate data for per capita output and its various components. We do a variance decomposition of analysis of that simulated data to support our key findings.