Computational Economics 15: 223–226, 2000. © 2000 Kluwer Academic Publishers. Printed in the Netherlands. 223 The Power of Tests for Non-Linearity: The Escribano–Pfann Model ⋆ STEVEN COOK 1 , SEAN HOLLY 2 and PAUL TURNER 3 1 Coventry Business School, University of Coventry, Priory Street, Coventry, CV1 5FB, U.K., E-mail: steven.cook@coventry.ac.uk; 2 Department of Applied Economics, University of Cambridge, U.K.; 3 Department of Economics, University of Sheffield, U.K. (Accepted 10 August 1999) Abstract. The power of the recently proposed Escribano–Pfann (1998) model of asymmetry is examined via Monte Carlo simulation. As found previously for an alternative asymmetric model, the power of this model is seen to be low. Key words: Monte Carlo, asymmetry 1. Introduction Since the start of the 1980s applied econometrics has witnessed an explosion of interest in both cointegrated time series and the modelling of non-linearities and asymmetries. The combination of these two research programmes has resulted in the asymmetric error correction model, first introduced by Granger and Lee (1989) (GL). This model allows for asymmetric behaviour by partitioning the error cor- rection term (ECT) about its mean, thus permitting differing speeds of adjustment on either side of the cointegrating vector, or attractor. However, the failure of this model to capture asymmetric behaviour in practice led Cook, Holly and Turner (1999a) (CHT) to examine its power via Monte Carlo simulation. It was found that the GL model has low power. An alternative approach to modelling asymmetry within the error correction framework is provided by Escribano and Pfann (1998) (EP), which partitions the ECT using the difference operator, . Therefore the EP model allows asymmetric adjustment to be captured in ‘. . . situations where the growth rate of the decision variable exceeds the growth rate of the target’ (Escribano and Pfann, 1998, p. 207). As this approach has met with some success in practice (see Cook, Holly and Turner, 1999b) the power of this model will be examined here. This paper will proceed as follows. In Section 2 the EP model will be described and a Monte Carlo analysis of it conducted. Section 3 concludes. ⋆ This research has been financed by ESRC Award No. L116251017.