Econometric tests of the expectations theory of the term structure in New Zealand

Abstract

This paper attempts to account for the empirical failure of the expectations theory of the term structure when it is tested using a variety of methods based on single-equation and vector autoregressive (VAR) models. It is argued that the failure of the spread to forecast future short-term rate changes is due to the omission from the regression of a time-varying term premium that is correlated with the spread. The inclusion of a white-noise error term in spread regressions is thought to take account of any random components in the term premium, and thus enable better judgement to be made about the expectations hypothesis of the term structure.

This investigation finds there is strong empirical support for the long-run implications of the expectations theory. However, the empirical evidence does not support the short-run predictions of the expectations theory when these predictions are tested by imposing restrictions on the parameters of single-equation and VAR models. These results are inconsistent with the view that the inclusion of a white-noise error term in spread regressions is enough to reconcile the expectations theory with the data.