The conquest of U.S. inflation: learning and robustness to model uncertainty

Previous studies have interpreted the rise and fall of U.S. inflation after World War II in terms of the Fed's changing views about the natural rate hypothesis but have left an important question unanswered. Why was the Fed so slow to implement the low-inflation policy recommended by a natural rate model even after economists had developed statistical evidence strongly in its favor? Our answer features model uncertainty. Each period a central bank sets the systematic part of the inflation rate in light of updated probabilities that it assigns to three competing models of the Phillips curve. Cautious behavior induced by model uncertainty can explain why the central bank presided over the inflation of the 1970s even after the data had convinced it to place much the highest probability on the natural rate model.