با استفاده از نرخ بهره بلند مدت به عنوان ابزار سیاست پولی

Using a short-term interest rate as the monetary policy instrument can be problematic near its zero bound constraint. An alternative strategy is to use a long-term interest rate as the policy instrument. We find when Taylor-type policy rules are used by the central bank to set the long rate in a standard New Keynesian model, indeterminacy—that is, multiple rational expectations equilibria—may often result. However, a policy rule with a long-rate policy instrument that responds in a “forward-looking” fashion to inflation expectations can avoid the problem of indeterminacy.

Over the past decade, inflation rates in many countries have fallen to levels not much above zero. This long-sought return to price stability should provide significant benefits in terms of enhanced economic efficiency and performance; however, as noted early on by Summers (1991), low inflation rates may also make it harder for central banks to achieve their macroeconomic stabilization goals. Specifically, as inflation has declined, short-term nominal interest rates, which are the usual instrument of monetary policy, also have fallen and have closed in on, or even run up against, their lower bound of zero, likely limiting the extent to which real interest rates can be lowered. This constraint may diminish the ability of a central bank to stimulate the economy and offset adverse macroeconomic shocks through the usual policy transmission mechanism of lower real rates (see Reifschneider and Williams, 2000 and Clouse et al., 2003). This long-standing theoretical concern about the so-called liquidity trap has been given a visceral immediacy by the example of Japan, which spent more than a decade mired in economic stagnation and deflation. In trying to stimulate the Japanese economy, the Bank of Japan came up against the zero bound when it lowered its policy rate (the overnight call rate) to zero in February 1999, and further policy stimulus via lower short rates was clearly impossible. Japan's predicament generated much discussion about the nature of the zero bound constraint and its importance in hindering macroeconomic stabilization. In response, many researchers have proposed using a variety of alternative monetary policy strategies and policy instruments other than the short rate—such as the monetary base, a long-term interest rate, or the exchange rate—to provide increased stimulus in such a situation (notably, Krugman, 1998, Meltzer, 2001, Svensson, 2001, McCallum, 2002 and Okina and Shiratsuka, 2004). These alternative monetary policy proposals have also been discussed for the U.S. economy. As illustrated by the press coverage of Chairman Greenspan's testimony above, faced with sluggish real growth in 2003 and inflation and short-term interest rates at historic lows, the Federal Reserve also studied various strategies to stimulate the economy if short rates fell to their lower bound (e.g., Bernanke and Reinhart, 2004). A common thread in these alternative policy strategies to stimulate the economy is their reliance on influencing the public's expectations of future policy actions—often by committing to a clear target path for some future economic variable, such as the short-term policy rate, the price level, or the currency exchange rate.1 The forward-looking nature of these alternative policy proposals is perhaps most transparent for the case of influencing expected future short-term interest rates, and thereby long-term bond rates. Accordingly, the first line of defense at the zero bound appears to be flattening the entire yield curve. Although flattening the yield curve is the clear objective of policy when confronting the zero bound, it is not obvious that market expectations of future short rates will automatically align themselves with those desired by the central bank. The research mentioned above assumes that the central bank can credibly commit to future policy actions and that expectations are rational. Under these assumptions, the central bank simply needs to announce an appropriate policy rule, sit back, and be confident that markets will coordinate on the desired equilibrium. In such an environment with credibility and appropriate policy, the zero bound has little effect. In practice, however, one may not have full confidence in the seamless communication of future policy actions to markets imagined in theoretical models. Indeed, this concern that markets may misinterpret the central bank's intentions is particularly acute at the zero bound when it cannot use the short rate to “back up its words with actions.” Instead, it may be useful, or even necessary, to signal to markets the desired path of expected short rates through concrete actions. In particular, the central bank can transmit its intent by using the long-term bond rate as the policy instrument. The goal of such an alternative policy approach is to effectively communicate to markets a path for future interest rates, as summarized by the targeted long rate, that replicates as closely as possible the desired equilibrium implied by the optimal policy under commitment. This approach to “targeting” long rates is a type of “targeting rule” in the sense of Lars Svensson, by which the central bank is committed to achieving a relationship between long rates and other endogenous variables.2 Therefore, consistent with Chairman Greenspan's views, after the short rate is pushed to zero, an obvious next step for monetary stimulus is to push interest rates of progressively longer maturities to zero as well. Despite being a natural alternative to a short-rate policy instrument, there has been remarkably little analysis of the properties of a long-rate policy rule.3 This is somewhat surprising because affecting the economy through long rates has typically been the acknowledged channel for monetary policy transmission—the overnight rate on its own is not an important factor on the economic decisions of most consumers and businesses. That is, a long-rate policy may be useful strategy even in cases where the zero bound did not constrain policy. Therefore, in Sections 2 and 3, we illustrate how minor the step is in moving from a short-rate to a long-rate policy instrument—at least on a formal modeling level. Section 2 analyzes the standard formulation of a New Keynesian macroeconomic model coupled with a Taylor-type rule for setting the short-term interest rate as the monetary policy instrument, and in Section 3, the one small modification made is that the rate on a longer-maturity bond becomes the policy instrument. However “obvious” it may seem to shift from using a short rate to a long rate as the monetary policy instrument, we find that doing so raises the important issue of indeterminacy. Indeterminacy arises because many possible future paths for the short rate may be consistent with a given setting of the long rate. Thus, using the long rate as a monetary policy instrument increases the likelihood of indeterminacy even for policy rules that appear prima facie to be reasonable. One solution to the potential indeterminacy is for the central back to have a clear signaling and communications strategy about the intended path for the short rate that is associated with the long-rate monetary policy rule. An important complementary factor is the stability of the rational expectations equilibrium (REE) under learning, which indicates whether atomistic economic agents could coordinate expectations on a candidate REE (Evans and Honkapohja, 2001). Clearly, if an equilibrium is stable under learning, such learning could help reinforce a central bank's communication strategy. In addition, in the case of multiple equilibria, stability under learning can be used as an equilibrium selection criterion. Section 4 extends the analysis to explore an alternative specification of the monetary policy rule that follows McCallum's recommendation that the determinants of an operational policy must be observable by the policymaker in real time. We find that policies that respond to observable lagged data are also subject to the problem of indeterminacy and that this problem is even more profound owing to the possibility of multiple minimum state variable (MSV) solutions. Given the risk of indeterminacy with standard rule formulations, Section 5 examines an alternative specification of the long-rate policy rule that yields a determinate outcome. Specifically, we show that “forward-looking” versions of a policy rule that uses the bond rate as the policy instrument entirely overcome the problems of indeterminacy that plague rules responding only to contemporaneous or lagged variables. Our finding that properly specified forecast-based long-rate policies guarantee a determinate REE while backward-looking variants do not stands in stark contrast to the literature on short-rate policy instruments, where the opposite conclusion is often found to be true (e.g., Bernanke and Woodford, 1997). Finally, as noted in Section 6, although our analysis focuses only on the long rate, our results appear relevant for other proposals for avoiding deflation that involve shaping the path of expectations about the future.

One proposed solution to the zero bound problem is for the central bank to use a longer-term bond rate as the policy instrument. In this paper, we show that such a policy can easily lead to indeterminacy; furthermore, in some cases, more than one of the resulting multiple equilibria may be stable under learning. We show that the problem of indeterminacy can be avoided if policy is explicitly forward-looking. Specifically, a forward-looking policy rule in which the bond rate is determined by expected inflation over the maturity of the bond produces a determinate and stable equilibrium. The bottom line of this stylized analysis is that, although “moving out the yield curve” in response to the zero bound involves more considerations than might be apparent at first, a careful central bank conceivably could pursue an effective long-rate policy. However, it should be stressed that our austere theoretical analysis is silent on many important practical issues. For example, our framework cannot address questions such as the tradeoff in the choice of maturity to use as the policy instrument. Moreover, we have not considered the characteristics of long-rate policy rules that are able to best mimic the optimal equilibrium under the zero bound. Such an analysis requires an empirical model and explicit incorporation of the zero bound on interest rates and is a fruitful direction for future research. Of course, lower future short nominal rates and lower current and future long nominal rates could accomplish the goal of trying to reduce real interest rates enough in order to stimulate the economy. However, higher expectations of future inflation can also lower real interest rates, and there are many proposals to exploit such expectational policy levers on the economy at the zero bound (e.g., Krugman, 1998, Svensson, 2001 and Svensson, 2003; and McCallum, 2000 and McCallum, 2002).20 Our analysis appears to provide a cautionary note as well for this general class of proposals that involve expectational policy instruments like the long rate. In particular, trying to manipulate the average expected inflation rate over the next, say, 5 years by means of an inflation target may not be sufficient, even if credible, to determine a single outcome, because there may be many possible paths for expected prices that are consistent with a long-run average inflation rate. Finally, our analysis of the manipulation of the long rate should not be construed as suggesting that we think this is the only or even the most effective policy that could be pursued at the zero bound when economic stimulus needs to be applied. But we do believe that given the current central bank unanimity regarding the short rate as the instrument of monetary policy, manipulation of the long rate is the most likely first step in overcoming the zero bound—again, as suggested by Chairman Greenspan's views in the epigraph.