Adaptive estimation of cointegrated models: Simulation evidence and an application to the forward exchange market

The paper reports simulation and empirical evidence on the finite-sample pe rformance of adaptive estimators in cointegrated systems. Adaptive estimato rs are asymptotically efficient, even when the shape of the likelihood func tion is unknown. We consider two representations of cointegrated systems - triangular cointegrating regressions and error correction models. The motiv ation for and advantages of adaptive estimators in such systems are discuss ed and their construction is described. We report results from the estimati on of a forward exchange market unbiasedness regression using the adaptive and competing estimators, and provide related Monte Carlo simulation eviden ce on the performance of the estimators. Copyright (C) 1999 John Wiley & So ns, Ltd.