Measurement errors in survey forecasts of expected inflation and the rationality of inflation expectations

Studies that use survey data to test the rationality of inflationary expectations usually test for unbiasedness by regressing actual inflation on expected inflation and testing the joint hypothesis that the intercept is zero and the slope coefficient is one. Such studies are fatally flawed because they incorrectly assume that expected inflation is measured without error. A procedure that allows for the influence of observation errors in expected inflation is developed and applied to the University of Michigan Survey Research Center data. Contrary to some recent studies, unbiasedness is rejected.