Asymmetric Adjustment and Bias in Estimation of an Equilibrium Relationship from a Cointegrating Regression

This paper uses Monte Carlo methods to investigate the effects of asymmetric adjustment on estimates of the parameters of the equilibrium relationship between a set of variables. We demonstrate that simple least squares estimates and the implicit estimates from a symmetric error correction model both lead to biases in the constant term. This bias increases with the size of the asymmetry and shows no tendency to decline with the sample size. We also show that if the biased estimates of the equilibrium relationship are then used to devide the sample into different regimes to test for assymmetric adjustment,then the resulting test has low power. The power of tests for asymmetry can be increased significantly by using simultaneous estimation of the parameters of the equilibrium relationship and the asymmetric adjustment process. © 2003 Kluwer Academic Publishers.