Fair (2002) argues that New Keynesian models are wrong in predicting that an inflation shock has contractionary effects only if it raises the real interest rate, and that a coefficient on inflation higher than one in the Taylor rule is a necessary condition for stability. While Fair uses his macroeconometric model as a benchmark to evaluate the predictions of the standard New Keynesian framework, we adopt a VAR supported by models in that framework, and the model of Rudebusch and Svensson (1999). The findings are broadly in line with Fair's.