Study of exchange rate modelling

Abstract

This paper presents a study in the mainstream of exchange rate modelling.

The literature survey begins with an historical section, tracing the origins of many presently used principles and concepts to debates that started over two centuries ago. Literature on balance of payments determination is. included as relevant to this survey. The elasticities, absorption, and monetary approaches are presented as three basic models which have often served as skeletons on which recent attempts at modelling exchange rates have been built. Various individual components or aspects of exchange rate modelling are then discussed-- components such as expectation formation assumptions, stock and flow effects, and flexibility of prices, which can be manipulated to transform one of the three basic models into one of the other two, or to extend one of the basic models into a new model. The dynamics involved in exchange rate modelling are considered, laying the groundwork for a later chapter. Finally, empirical work on exchange rate models is briefly summarized.

In the three central chapters, empirical versions of the monetary, Dornbusch, and stock-flow type models are developed and econometrically tested using a sterling effective exchange rate and data for the United Kingdom and its major trading partners. In cases for which more than one set of data is arguably appropriate for a given variable used in estimation (for instance, Ml, M3, or sterling M3 might be the appropriate money supply measure to use in regression) the sensitivity of the estimation to the data set used is studied. On the basis of Sargan's test for common factors, all three models show signs of dynamic misspecification (though for the stockflow model that conclusion is seen to depend on the data set used). Both graphical methods and two different F- tests provide evidence of structural breaks in estimations of all three models over the period studied. The log likelihood ratio -and Davidson and MacKinnon's non-nested tests are used to compare the three models and to compare various estimations of each of the models.

The last main chapter of the text presents further discussion of the dynamics involved in exchange rate modelling. In particular, models displaying saddlepoint type stability (a popular aspect of recent modelling, especially rational expectations modelling) are considered in terms of realism of their underlying assumptions. The idea of structural stability or robustness is discussed as relevant to the issue of realism of saddlepoint type models. Possible extensions of the present research are discussed in the conclusion