Optimal Monetary Policy in a Model with Habit Formation and Explicit Tax Distortions

A number of recent papers have explored monetary policy options, including inflation targeting and inflation forecast targeting (notably Svensson (1999a, 1999b, 2000)) and price level targeting (Wolman 2000, Batini and Yates 1999, Blinder 1999). Most papers explore “optimal” monetary policy in the context of a single model. However, a number of conclusions made in the literature depend strongly on the model specification used. In addition, most papers have used the efficient policy frontier concept to define optimal monetary policy. This paper investigates the behavior of a variety of small structural macro models under a variety of targeting rules. The paper examines both minimum variance policy frontiers and utility-maximizing policy. In the latter case, an explicit model of consumer behavior with inflation-induced tax distortions is explored. The paper examines the improvement in utility from an optimal price-level target and re-examines the improvement in utility in moving from a positive to a zero target inflation rate. (JEL D12, E52, E43) Preliminary and Incomplete Not for Quotation Senior Vice President and Director of Research, Research Department, Federal Reserve Bank of Boston, Boston, MA, 02106. e-mail: [email protected]. I thank Richard Kopcke, Scott Schuh, Geoffrey Tootell and seminar participants at the Bank of England, the Sveriges Riksbank, Stockholm University, Brandeis University, and Middlebury College for their helpful comments, and Ann Ferris for excellent research assistance. Thanks to Dan Feenberg for providing tax rate data from TAXSIM. The views expressed in this paper are the author’s and do not necessarily reflect those of the Federal Reserve Bank of Boston or the Federal Reserve System. “Nothing breeds failure like success.” The success of monetary policy in controlling inflation in the developed economies over the past 20 years has led many to wonder how to retain that success, or even to improve upon it. Certainly it is reasonable to be concerned about the extent to which favorable monetary policy has arisen because of the idiosyncratic traits of particular central bankers. Going forward, it would be preferable to design monetary institutions so that successful practice would be incorporated in central bank charters, rather than embodied in individuals. As a result, a large literature has emerged to propose optimal monetary policy strategies. The work of Svensson (1999a, 1999b, 2000) and Bernanke, Laubach, Mishkin, and Posen (1999) has focused debate on the primacy of inflation targeting as an organizing principle. For many years, the idea that central banks might directly target the price level was dismissed. The reasoning was roughly as follows: Because stabilizing the price level requires moving the inflation rate above and below its target, and because doing so requires corresponding movements in output (according to most models), price level targeting would entail unacceptable variability in output. But more recent work has called into question the generality of this proposition. Wolman (2000), for example, shows that when the nominal interest rate hits its zero lower bound, a price level target may help to lower ex ante real interest rates, because it will imply a future rise in expected inflation. The more general proposition that price-level targeting might provide a “free lunch,” in which inflation and output variability are reduced without consideration of the zero lower bound, has been advanced by Svensson (1999b). This result, however, is modeldependent, and as I will show below, does not hold up in data-consistent macro models. Most of the literature relies on an ad hoc monetary policy loss function that is a weighted average of variances of inflation (about its target), output (about