Stock Returns, Inflation, and the Phillips Curve

Abstract

In recent years, the negative relation between stock returns and inflation has been rigorously investigated. Studies by Lintner [25], Bodie [1], Jaffee and Mandelker [21], Nelson [28], Fama and Schwert [11], and Gultekin [17], to name but a few, consistently show that stock returns are negatively associated with inflation, if associated at all, and conclude that the Fisher hypothesis does not hold for stocks. Explanations for this finding are equally abundant. For instance, Fama [10] postulates that the observed negative association is largely a spurious phenomenon, a conclusion he justifies by maintaining that it is consistent with a simple money demand-quantity theory world. Geske and Roll [15] extend this line of thought by developing a more complex macro-economic model that includes such items as unemployment, corporate earnings, and government financing. Following a somewhat different approach, Ram and Spencer [32] reexamine Fama's work by augmenting the Fisher equation to include the Phillips curve hypothesis. They conclude that there is a positive association between inflation and real activity and a negative relationship between real stock returns and real activity, a position that contradicts Fama's findings. A problem common to the above explanations is that a general equilibrium framework is not employed. The purpose of this study, therefore, is to further develop this work by examining the relationships among real stock returns, inflation, and unemployment using a general equilibrium model. A nonspurious negative relation of real interest and inflation is employed that is based on the Levi and Makin studies [23; 24]. It is shown that the natural rate of unemployment phenomenon, which causes the Phillips curve to shift, also simultaneously causes the real interest rate-inflation relation to shift. Using this framework, the hypothesis that the real stock return-inflation relation shifts concurrently with the Phillips curve is empirically tested. The hypothesis is not rejected. This suggests that the usual simple regression of stock returns on inflation is misspecified so that the slope estimate is not as steep as the true one. The plan of the paper is as follows. The concurrency of the shift in the real interestinflation relation and the Phillips curve is demonstrated in the next section. In section III, the shift in the Phillips curve is tested by the Brown, Durbin, and Evans method [3]. Then, using dummy variables to represent the shifting phenomenon of the Phillips curve, empirical