Implications of conceptual and data complexities on time-series econometric applications in market integration analysis

Abstract

This article demonstrates and highlights the conceptual limits of current empirical market integration (MI) time series models (threshold models) and their implications on market effi ciency and competitive equilibrium conclusions. The complexities and diversities that characterise the analysis of the concept of market integration are evaluated within the framework of EnkeSamuelson-Takayama-Judge (ESTJ) spatial equilibrium theory. The effi ciency and competiveness implications drawn from MI models are limited by how the data generation process (DGP) is infl uenced by equilibrium conditions, by the tradability restrictions of the inter-markets relationships and by the presence of unobserved transactions costs. However, empirical applications scarcely address these limitations. Two sets of synthesized data with varying levels of non-linear complexity implied by alternating equilibrium conditions are generated to demonstrate conceptual limits of current threshold models in market integration analysis. Inconsistent conclusions that linear representations imply for threshold propagated DGP will also apply for conclusions derived from threshold models if markets are characterised by switching equilibria conditions.