Abstract
A flexible procedure is proposed whereby both the random and systematic errors in measurement of survey data on inflation expectations can be appropriately handled when these surveys are used in the estimation of macroeconomic models. During the period 1952–1980, Livingston's two expectations series and the SRC series systematically underestimated inflation expectations by nearly 25%. The ASA–NBER survey, however, does not exhibit such underestimation. Thus, the use of the Livingston or SRC data as an unbiased measure of the true expectations of inflation would result in overestimation of the Fisher effect and the ex ante real interest rate.