Revisiting Risk-Return Relation By The Modern Econometrics Analysis And Its Implications to Monitor A Potential Financial Crisis

Abstract

This paper re-examines the risk-return long-run relationship by the modern econometrics analysis and further investigates the resulting implications for detecting a potential crisis. We propose a new estimation based on the generalized Cochrane-Orcutt transformation via the AR (K) approximation for a regression displaying both properties: imbalanced or spurious, even though the error terms are unknown in practice including the possibility of the GARCH effect. Our procedure provides the consistent estimators for both imbalanced and spurious regressions. More importantly, the proposed estimation resolve the risk-return puzzle in intertemporal capital asset pricing (ICAPM) theory, which postulates the negative and linear relation between the expected return and conditional variance. That implies using our methodologies can yield the positive risk-return relationship and more significant results when compared to the current literature. Finally, based on the analysis of our new estimation, a potential incoming crisis could be monitored which further provides strong evidence on the usefulness of our methods.