The Long-Run Fisher Effect: Can It Be Tested?

In this paper, I provide a plausible explanation as to why past studies have been unable to find support for the long-run Fisher effect. My argument is that exogenous shocks to the inflation rates in industrialized economies have not produced the permanent change to inflation necessary for testing the Fisher effect. Instead of finding a nonstationary, unit-root process for inflation like previous Fisher effect studies, here each country's inflation rate is found to follow a mean-reverting, fractionally integrated, long-memory process. Applying a bivariate, maximum likelihood estimator to a multivariate, fractionally integrated model of inflation and nominal interest, I find that the estimated inflation rates in 17 developed countries are highly persistent, fractionally integrated, mean-reverting processes with order of integration parameters significantly less than one. Since a permanent change to inflation has not occurred, a test of whether a permanent change to inflation affects the nominal interest rate one-for-one will be uninformative as to the truth or fallacy of the Fisher effect hypothesis.