The term structure as a predictor of economic activity

Abstract

The principal aim of this paper is to assess the relationship between the term structure and economic activity and correspondingly to evaluate the forecast strength of the term structure on economic activity for the following countries; South Africa, Poland, China, Chile and Brazil. The motivation is due to the existent of numerous studies conducted to investigate the predictive strength in developed countries but not nearly as much for developing countries. The simple probit model that estimates probability of an economic activity downswing directly suggested by Estrella and Mishkin (1996) is used. The results from this model are compared with a modified probit model Dueker {1997) proposed. The modified version of the model strives to eliminate the serial correlation which may possibly be present among error terms by an addition of an independent variable equivalent to a lagged dependent variable in the simple probit model. Furthermore the study employs the Generalised Method Moments as an estimation technique on the basis of it being viewed as superior from an efficiency perspective when compared to the Ordinary Least Squares (OLS) methods. The findings revealed that the term structure prediction of economic activity is still reliable for both policymakers and private investors, though the predictive strength is not as robust when compared to the developed countries. Similarly the GMM performs relevantly well in comparison with the OLS methods utilised.