سیاست های پولی در محدود بهره صفر : یک تمرین مقایسه ای مدل

In this paper, we summarize the findings from zero-interest-bound simulation exercises conducted on the policy/forecasting models of the three major central banks. After imposing a fixed-period zero-interest-bound episode on each model, we consider common variations in the monetary-policy reaction function to minimize the macro-economic consequences of such a deflationary regime. Although there is some heterogeneity in the ranking of these remedial policies, reflecting the different properties of the models, we find that more aggressive policy rules and price-targeting rules are potentially candidates for robust monetary strategies. J. Japanese Int. Economies20 (3) (2006) 305–313.

This paper and the three that follow consider the effectiveness of a number of proposals for reducing the consequences of the zero lower bound on nominal interest rates using simulations of the policy models of three major central banks: the Bank of Japan's JEM model (Fujiwara et al., 2005); the European Central Bank's Area-Wide Model (Fagan et al., 2001) and the Federal Reserve's FRB/US model (Brayton et al., 1997). Our goal is to look for policies that work well across a number of models.1 These papers were originally prepared for a conference held in September 2004 as part of the joint NBER/CIRJE/EIJS/CEPR “Japan Project” meeting. The zero bound poses a problem for monetary policy because central banks typically respond to weak aggregate demand conditions by lowering short-term interest rates. Since nominal interest rates cannot be reduced below zero, some modifications in monetary-policy strategy will be needed at low levels of interest rates. Interest in such strategies has increased in recent years because short-term interest rates have been very low in a number of countries: in Japan, the call money rate has been at or below 50 basis points since late 1995; in the United States, the federal funds rate was below 150 basis points from late 2002 to the summer of 2004; and in the euro area, the equivalent rate was held at 200 basis points from summer 2003 until late 2005. Accordingly, a number of proposals have been made to reduce the consequences of the zero lower bound on nominal interest rates. A common element of these proposals is that they work by lowering expected future real short-term interest rates, either lowering expectations of future nominal rates or raising expectations of future inflation. Our use of central-bank models to evaluate the effects of these policies is a key feature of this project. Some earlier studies have examined the effects of changes in monetary policy using highly stylized models—for example, Wolman (2005) and Eggertson and Woodford (2003). We believe, however, that using policy models has some important advantages relative to those relatively more compact frameworks. For instance, in the central-bank policy models, there is a much richer range of economic mechanisms at work, and these additional mechanisms allow a more detailed examination of the drawbacks and advantages of certain policies. Moreover, the central bank models are estimated, and thus may provide a more-realistic platform for policy evaluation. Finally, these models are, for better or for worse, familiar to policymakers, and so provide them with a natural benchmark for variant analysis. In this project, we evaluate various policies looking at the response of the economy to a specific shock under broadly harmonized conditions. In each model, a set of demand shocks is chosen so that the short-term interest rate would be pinned at zero for a period of four to five years—under the assumption that the central bank will follow a baseline “Taylor rule” policy once the zero-bound is no longer binding. We consider such a shock to be a good approximation to the kind of severe demand shock that leads to fears about the zero bound. Indeed, in each model, the zero bound makes the effects of the shock on output and inflation considerably worse than would otherwise be the case. Given this benchmark, we consider whether specific policies can undo some of the consequences of this shock: • Reifschneider and Williams (2000) have argued that a more-aggressive monetary policy can reduce the adverse effects of the zero bound. We model this more-aggressive policy as larger coefficients on output and inflation in a Taylor rule. • A central bank may promise to make up any shortfall in monetary-policy stimulus relative to the Taylor rule that occurs during a zero-bound period ( Reifschneider and Williams, 2000). Under such a policy, future short-term interest rates will be lower than they would be under the Taylor rule, and, in particular, will stay at zero beyond the period suggested by the Taylor rule. • Price-level targeting, in which the inflation component of the Taylor rule is replaced by the deviation of the price-level from a target, has been advocated by Wolman (2005) and Eggertson and Woodford (2003). • Krugman (1998) has suggested that central banks could raise their inflation target permanently, or at least for a very long period, as a way of escaping a zero-bound episode. • State-contingent policies—such as policies that are more aggressive when the economy is approaching or leaving the zero bound—may be a way of achieving some of the benefits of alternative policies when interest rates are very low while returning to more-normal rules for setting policy at other times. The remainder of the paper is organized as follows. In Section 2, we briefly discuss the models used in this exercise and highlight their similarities and points of departure. Thereafter, we describe the model comparison exercise in terms of the policy rules examined and the demand shocks implemented to create a lower bound benchmark. The results from this harmonized experiment are summarized in Section 4. Section 5 concludes.

A key conclusion of our joint study is that, in each of the three central bank models that we examined, there are indeed strategies that can significantly mitigate the adverse effects of the zero bound. One policy that worked well in all of the models was a more-aggressive monetary policy. A promise to make up any shortfall in monetary-policy stimulus because of the zero bound was also beneficial in all three models, although the degree of benefit varied across models. Two other policies—price-leveling targeting and a higher inflation target—led to mixed results. These mixed results suggest a note of caution may be warranted with respect to these policies—even when they work well in specific models. An important question for policymakers is how to implement policy once the zero-bound has been reached. As Fujiwara et al. emphasize, once the interest rate is zero, it can be hard for policy to be “tangible,” because short-term interest rates are the way that central banks typically implement policy. Such considerations point to the advantages of policies such as a more-aggressive Taylor rule or price-level targeting in that they can be adopted in normal times, when interest rates are positive. By adopting such a policy before the economy is confronted with the zero bound, the central bank can hope to alter the “typical” expectations-formation mechanism. In that way, the central bank will not have to rely on the effects of a policy announcement on expectations formation. Of course, central banks already confronted with the zero bound must operate with the expectations and reputation they have inherited. The papers consider some of the issues raised by this prospect. For example, Reifschneider and Roberts consider the possibility that expectations of financial markets may respond to policy announcements while expectations of other agents do not. They found that the benefits of most policies were reduced in this case, although many policies continued to provide a substantial improvement in the economy's performance, notably the aggressive Taylor rule. Fujiwara et al. note that certain state-contingent policies can have tangible effects: they show that a preemptive policy, in which interest rates are reduced more rapidly once the zero-bound becomes a risk, can reduce the costs associated with the zero bound. They also show that in the period near the exit from the zero bound, a policy that is particularly aggressive in returning inflation to its target level—and thus keeps interest rates lower for longer—can also be beneficial. The three models examined here differ in a number of ways: with respect to their size; degree of aggregation and non-linearity; relevance of forward-looking behavioral elements and adherence to micro-foundations. Hence, accounting for the responses of the different models to harmonized policy experiments is a complicated task—a problem that has plagued other model-comparison exercises, such as the well-known study by Bryant et al. (1993). We nonetheless feel some general conclusions can be reached. First, consider the degree of forward-lookingness. In that regard, JEM and the FRB/US model are closer in the model spectrum than they are to the AWM, which has fewer forward-looking features. Consequently, policies which explicitly rely on “policy duration” effects, such as the Reifschneider–Williams proposal, are relatively less effective in the AWM. Second, asset markets: the stronger and faster is the wealth channel, the more likely will be the recovery from a liquidity trap. Similarly, the more prevalent are long–term interest rates (real or nominal) relative to short rates in determining demand, the more successful a credible zero interest bound policy will be given its flattening of the yield curve. Finally, micro-foundations: models predicated on inter-temporally optimizing behavior tend to generate smoother, less contractionary outcomes from deflationary shocks given that agents can smooth consumption and hours worked. Finally, we might note that although the Bank of Japan has, since early 2006, committed itself to ending “quantitative easing,” normalization of policy and inflation performance may be an extended process. Likewise, many countries, particularly those undergoing rapid structural change, may experience periodic deflation. The topic therefore is far from dead, and neither is the research agenda. Interesting extensions of our work might include the following: examining economic performance under imperfect policy credibility; the international repercussions of zero-interest-bound strategies; and more systematic analysis of the benefits of state-contingent policies. We leave these open for possible future work.