Flexible Inflation Targeting under a Non-Linear Phillipscurve

This Paper analyses the optimal degree of flexibility under a Lucas type convex Phillipscurve. As a benchmark, we first analyse optimal monetary policy with a linear Phillipscurve and persistent cost-push shocks. As in Svensson (1997a), a central banker who possesses private information and who inherits society´s preferences will engage in too much output stabilisation; because of which welfare will be improved by appointing an individual who is less flexible. Moreover, we are able to investigate the determinants of the optimal degree of flexibility and contrast them with results obtained in a situation where the central banker has an output target which exceeds its potential. If the central banker has no private information under a linear Lucas type Phillipscurve, it will be optimal to abstain from output stabilisation entirely. Next, we extend the symmetric information case by assuming the Phillipscurve is convex. In this respect, it is shown that even under strict inflation targeting, the optimal conditional inflation forecast will be state-dependent. This is because the central bank can to some extent predict the slope of the Phillipscurve. Furthermore, if the degree of flexibility is zero, monetary policy will be subject to a deflationary bias which will exceed the bias obtained in a model where the expected slope of the Phillipscurve is constant. We also show that the long run average rate of inflation will be strictly increasing in the degree of flexibility. This is because both demand uncertainty and persistent supply shocks will, on average, cause output to fall below potential. A central bank with an output target equal to potential will try to prevent this but because of the lack of an information asymmetry its attempts will only increase the expected rate of inflation. Therefore some degree of flexibility will be socially optimal in this model because it will render the deflationary bias obtained under strict inflation targeting less severe.