Inflation targeting, price-level targeting and the zero lower bound

We study the impact of inflation targeting (IT) versus price-level targeting (PT) in a simulation model that takes into account the zero lower bound constraint on the central bank’s policy rate. We show that the economy is much less likely to hit the zero bound under PT for a given trend inflation rate. Trend inflation is optimally positive because of the zero lower bound under both IT and PT, but lower under PT. Economic welfare is significantly enhanced in a PT regime.

We solve the model using higher-order perturbation techniques that preserve its essential nonlinearities in order to make valid welfare comparisons. We model monetary policy using simple rules. Under IT, the central bank’s desired interest rate follows a Taylor rule. Under PT, its desired interest rate follows a modified Taylor rule that depends on deviations of the price level from its target path. The actual (net) short-term interest rate is the maximum of the desired rate and zero. We use a smooth function that approximates the kink in the interest rate reaction function at zero arbitrarily well.

Previous studies have analyzed the impact of the zero lower bound on monetary policy in models that are linear except for the zero bound constraint itself. Examples include Adam and Billi (2006, 2007), Aksoy et al. (2006), Billi (2011), Coenen, Orphanides and Wieland (2004), Coibion, Gorodnichenko and Wieland (2012), Jung, Teranishi and Watanabe (2005), Levin, López-Salido, Nelson and Yun (2009), Kato and Nishiyama (2005), Nakov (2006) and Reifschneider and Williams (2000). Exceptions to this rule are the papers of Eggertsson and Woodford (2003), Fernandez-Villaverde et al. (2012), and Wolman (2005). Eggertsson and Woodford’s model is simple enough to solve analytically. Wolman’s model is the closest in spirit to our own. He solves a simple New Keynesian model with two-period nominal price rigidity using projection methods.