The Term Structure of Interest Rates and the Demand for Money: British Results from a Portfolio Model

Abstract

Monetary economists have long taken the view that more than one interest rate might reasonably be expected to influence the demand for money, but problems of multicollinearity on the empirical side and a general failure to produce a priori results from the standard paradigms on the theoretical side have delayed the implementation of this notion. Recently, though, breakthroughs have occured in both areas such that theoretical results now give some guidance as to how the term structure-demand for money problem might be approached and, further, some recent empirical results on the U.S. data, at present somewhat loosely related to the theory, suggest strongly that a multi-interest rate framework is appropriate for the demand for money problem. The general theoretical paradigms available are M. Friedman's inventory-theoretic [6], J. I. Bernstein and D. Fisher's mean-variance [2] and J. Stiglitz's dynamic consumption [11], although Stiglitz does not address the demand for money problem explicitly. The more general dynamic consumption model, in any event would certainly fail to generate any predictions for interest rates of a sufficiently useful sort, but both the inventory-theoretic and the mean-variance model can; indeed, the former has been subjected to a successful empirical test on U.S. data by H. R. Heller and M. S. Khan [7]. The empirical implementation of the mean variance model is the purpose of this paper. In this paper, after specifying the portfolio model in a term structure-demand for money context and demonstrating its implications for empirical work, we conduct a number of tests on the British data. As we will see, the mean-variance model also generates the sign expectations that M. Friedman derived for the inventory-theoretic model, a fact which suggests that one must construct the theory with rather more attention to detail in order to claim any firm empirical support for one or the other of these paradigms. In this connection, the real virtue of the portfolio model is that a three security problem can be solved-with money, a short term bond, and a long term bond in it-and the solution is