پاسخ نرخ مدت به عدم اطمینان به سیاست های پولی

This paper shows that greater uncertainty about monetary policy can lead to a decline in nominal interest rates. In the context of a limited participation model, monetary policy uncertainty is modeled as a mean preserving spread in the distribution for the money growth process. This increase in uncertainty lowers the yield on short-term maturity bonds because the household sector responds by increasing liquidity in the banking sector. Long-term maturity bonds also have lower yields but this decrease is a result of the effect that greater uncertainty has on the nominal intertemporal rate of substitution—which is a convex function of money growth. We examine the nature of these relations empirically by introducing the GARCH-SVAR model—a multivariate generalization of the GARCH-M model. The predictions of the model are broadly supported by the data: higher uncertainty in the federal funds rate can lower the yields of the three- and six-month treasury bill rates.

The current generation of quantitative macroeconomic models, such as those based on the real business cycle paradigm, invariably cast the analysis within a stochastic environment in which the first moments of policy variables constitute the almost exclusive object of interest. In this literature, beginning with the Lucas tradition that emphasizedthe distinction between unanticipated and anticipated monetary policy and continuing with modern extensions that introduce various real and nominal rigidities (sticky prices, sticky wages, and limited participation models, for example), there are few examples that study the impact that the second (and higher) moments of policy variables have on economic activity and welfare. This paper broadens the analysis of macroeconomic policy by investigating how monetary policy uncertainty affects one important aspect of the macroeconomy: nominal yields on risk-free bonds. We are not the first to point out the paucity of research that examines the consequences of policy uncertainty. Obstfeld and Rogoff (2000) highlight the scant attention that policy uncertainty receives in open economy, macroeconomic policy analysis. While concerns about uncertainty of monetary policy are reflected in popular discussions of policy transparency and policy risk, the theoretical neglect of these issues is primarily driven by a key technical consideration: the solution of stochastic general equilibrium macroeconomic models typically involves a linear approximation that implies certainty equivalence in equilibrium. Obstfeld and Rogoff (2000) depart from certainty equivalence by assuming that the exogenous variables in the model have lognormal distributions. This particular distributional assumption allows them to obtain closed form solutions. Our analysis also requires that we make distributional assumptions to find exact solutions to the economy but these take the form of a discrete-state Markov process for monetary policy. Moreover, the transition probability matrix of this Markov process is appropriately parameterized to study the effects of time-varying uncertainty.1 Few papers outside the finance literature have successfully explained the variation in the term structure of interest rates with a modern equilibrium macroeconomic model. For example, den Haan’s (1995) analysis predicts a yield curve that is essentially flat. A notable exception is that of Evans and Marshall (1998) who find that a limited participation model of monetary non-neutrality is broadly consistent with empirical regularities in the term structure. A limited participation model is an attractive environment for an investigation of policy uncertainty on term-structure relations because of three important properties:2 (1) the channel of monetary policy transmission is captured through the traditional mechanism of liquidity affecting interest rates which, in turn, affect real activity; (2) agent’s savings decisions, which in part determine the supply of funds in the loan market, are made before the state of the world is known. Consequently, time-varying uncertainty in monetary policy may create an endogenous response in the loan market which will be reflected in interest rates; and (3) nominal interest rates are affected by both Fisherian and liquidity factors.Subsequently, changes in the second moment of monetary policy (which in our model is described by a simple money growth rule) may affect interest rates through one or both factors. The few previous studies that have examined the effects of time-varying uncertainty (e.g., Lee, 1995; Hodrick, 1989; Dellas and Salyer, forthcoming) used a simple cash-in-advance framework so that nominal interest rates are not affected by liquidity considerations. In addition, the environments investigated in these papers were either exchange economies or they insulated production from monetary uncertainty so that the interaction between uncertainty, output, and interest rates could not be analyzed. It is important to also note at the outset that we model a very specific type of monetary policy uncertainty. That is, we characterize policy uncertainty as a time-varying conditional variance of themonetary growth rate. This narrowfocus, in combinationwith the economic environment, leads to testable hypotheses.Of course, there aremany other interesting types and sources of policy uncertainty, e.g., the nature of the central bank’s reaction function (is it a forward- or backward-looking Taylor rule and does it include interest-rate smoothing terms?), time variation in the parameters of the reaction function, uncertainty due to measurement error in real-time data (as stressed by Orphanides, 2001), and uncertainty over the objective function of the monetary authority. Understanding the roles that these and other sources of uncertainty have on economic behavior is a laudable research goal, in our opinion; we view our research as a first step in furthering that goal. The main results in our paper can be summarized as follows. The model predicts that increases in monetary policy uncertainty will produce a generalized decline in interest rates for all maturities. This prediction has different explanations that depend on the maturity of the bond: at the very short end of the maturity spectrum, the endogenous response of savings (i.e., funds placed in the banking sector) to greater uncertainty results in more liquidity in the lending market, thus lowering the nominal yield. At longer maturities, the decline in rates because of greater uncertainty is due to the fact that the marginal utility of a dollar is a convex function of money growth, which causes a fall in the certainty equivalence of a dollar in the future. The predictions on term premia are indeterminate since they depend on risk aversion and the persistence of monetary policy. Empirical investigation of these propositions requires that we introduce appropriate econometric methods to measure the effects of uncertainty (which we take to mean volatility for empirical purposes) on the conditional mean of the variables in a VAR. A natural solution to this problem consists of generalizing the GARCH-M model to a multivariate context. The resulting model, which we label GARCH-SVAR and which is similar to the MGARCH-M VAR in Elder (forthcoming), not only allows direct measurement of volatility effects on the impulse response functions but also delivers interesting new properties for them: different shapes as a function of the magnitude of the shock, and asymmetric responses to positive and negative shocks. We apply the GARCHSVAR model to Evans and Marshall’s (1998) monetary VAR to identify the monetary policy innovation series and find broad support for the predictions of the model. In addition, the behavior of output volatility is consistent with the general decline in volatility reported in McConnell and Pérez-Quirós (2000). The remainder of the paper is organized as follows. Section 2 presents themodel, whose solution is described in Section 3. Section 4 measures the effect of monetary policy uncertainty empirically and Section 5 presents our conclusions and directions for future research.

Limited participation models are perhaps the only class of dynamic equilibrium models of monetary economies whose predictions of term-structure relations match the data reasonably well. Because they are capable of generating a liquidity effect, these models are particularly well suited to investigate the transmission of monetary policy on the term structure. The modeling tradition that characterizes these models (as well as most dynamic equilibrium models) essentially devotes undivided attention to the analysis of relations based on first moments of the stochastic processes that characterize the behavior of policy variables. As we discuss, this restrictive analysis is largely motivated by the technical difficulties entailed in solving these models rather than by an intrinsic disinterest in higher moment effects. One contribution of this paper is to open new ground in this modeling tradition by exploring the effects of a particularly relevant second moment effect: that of time-varying monetary policy uncertainty on term rates. Contrary to cursory intuition, we show that term rates tend to decline when monetary policy becomes more uncertain. At the short end, this increase in uncertainty results in increased liquidity in the lending market whereas at the long end, the convexity of consumption to money growth modifies the certainty equivalence of a dollar in the future. Another contribution of this paper is to introduce new empirical methods designed to measure second moment effects on the conditional mean of a dynamic system. The GARCH-SVAR delivers broad empirical validation to the predictions of the theoretical model but we expect that its interest widely transcends the application in this paper. For example, the GARCH-SVAR can be used to obtain direct measures of the effects of output and inflation volatility on the response of monetary policy and term rates. Generalizations of the GARCH-SVAR are immediately apparent and we reserve for future research theinvestigation of these issues. One specific advantage of the GARCH-SVAR that we want to consider is the possibility of identifying the structure of reduced form VARs by finding the space of linear combinations that will insure GARCH effects are restricted to the diagonal terms of the variance covariance matrix. The analysis in this paper adds to a growing literature that examines the effects that second moments have on the conduct of monetary policy. For instance, Dupor (forthcoming), shows that randomizing the monetary growth rate can increase utility in an economy with nominal price rigidities (due to monopolistic competition) since money growth surprises can help to eliminate welfare losses due to monopoly. In our model, greater uncertainty results in amore elastic response of labor due to a given monetary shock because of the greater fall in interest rates—hence it is possible that greater uncertainty in monetary policy is welfare improving. However, the simple environment studied here (in particular, the assumption of no-productive assets) makes any welfare claims tentative at best. In another related paper, Dotsey and Sarte (2000) demonstrate in a cash-in-advance economy that increased variability in money growth can increase the economy’s growth rate. Like the results in our paper, the mechanism is that increased uncertainty results in a greater precautionary savings which, in turn, leads to greater capital and growth. While these results improve our understanding of the impact of uncertainty, clearly, more research is needed—critically, the role of money and liquidity needs to be given greater attention, we believe, if these efforts are to be successful.