Is the Fisher effect asymmetric? Cointegration analysis and expectations measurement

Abstract

Using U.S. post-war data, we investigate whether the interest rate response to inflation known as the Fisher effect could be asymmetric. The asymmetry considered is that the long-run change in the interest rate is larger when inflation rises than when it falls. The possibility follows from behavioural hypotheses about the relationship of inflation expectations to actual inflation. Using an asymmetric cointegration approach, we find asymmetric cointegration in the Fisher effect for the post-war period through 1979, but not subsequently. We then find that starting in 1980, a breakdown developed in the relationship between inflation expectations from surveys and actual recent inflation rates, a breakdown not accounted for by asymmetry. If the survey results approximate true expectations, then econometric testing using actual recent inflation to compute expected inflation will suffer from mismeasurement, which could explain the finding of no cointegrating Fisher effect post-1979. The paper accounts for breakpoints and uses bootstrapping to conservatively estimate statistical significance.