Determining the appropriate deflation of nominal magnitudes in the UK consumption function - a nonnested test approach

This paper is concerned with the deflator(s) used to obtain series measured in real terms in 'the' consumption function. Since the precise deflator(s) to be used in any particular specification is not uniquely determined by theoretical considerations alone we consider ways of approaching the issue empirically. The starting point is to rearrange any given specification of 'the' consumption function so that a nominal consumption variable is explained. This implies that different deflators for consumption, and other variables in the initial specification which are expressed in real terms, define non-nested models. We therefore examine the issue of the appropriate method of deflation used non-nested tests. We consider the Davidson, Hendry, Srba and Yeo specification of the quarterly UK non-durable consumption function and find that the preferred method of deflating consumption and income variables is to use the total (durable and non-durable) consumption deflator. We also examine the quarterly UK consumption function of Carruth and Henley, based on the work Hendry and Von Ungern-Sternberg. In this case a prefered deflator does not emerge from the alternatives considered. We conclude by noting that the approach adopted in this paper should be useful in areas other than the consumption function literature.