Symbolic transfer entropy test for causality in longitudinal data

Abstract

In this paper, we use multiple-unit symbolic dynamics and the concept of transfer entropy to develop a non-parametric Granger causality test procedure for longitudinal data. Monte Carlo simulations show that our test displays the correct size and large power in situations where linear panel data causality tests fail such as when the linearity assumption breaks down, when the data generating process is heterogeneous across the cross-section units or presents struc- tural breaks, when there are extreme observations in some of the cross-section units, when the process displays causal dependence in the conditional variance and when the analysis involves qualitative data. We illustrate the usefulness of our proposal with the analysis of the dynamic causal relationships between public expenditure and GDP, between firm productivity and firm size in US manufacturing sectors, and among sovereign credit rating, growth and interest rates.