Term structure and forward guidance as instruments of monetary policy (Q2447163)

The authors study a simple monetary model with a Ricardian fiscal policy in which equilibria are indeterminate if the monetary policy consists solely of a rule for fixing the short-term interest rate. \textit{T. J. Sergent} and \textit{N. Wallace} [``Rational expectations, the optimal monetary instrument, and the optimal money supply rule'', J. Polit. Econ. 83, 241--254 (1975; \url{doi:10.1086/260321})] first pointed possible indeterminacy of inflation expectations in forward-looking rational expectations models over an open-ended future when a monetary authority uses an interest rate rule. If the fiscal policy adapts itself to the level of the government's debt to ensure that the debt does not grow faster than the interest rate (\textit{T. J. Sargent} [``Beyond demand and supply curves in macroeconomics'', Am. Econ. Rev. Papers Proc. 72, No. 2, 382--389 (1982); \url{http://www.jstor.org/stable/1802363}] coined this as a Ricardian policy), then the transversality condition does not force a unique path since every path automatically satisfies the transversality condition and there is a continuum of equilibria. In standard New Keynesian models, an active monetary policy, by which the short-term nominal interest rate is raised by more than the increase in inflation, leads to a unique path of the linearized system which stays close to the steady state and this approximate equilibrium is selected as the basis for policy analysis. A series of papers have pointed out that the local determinacy result with an active monetary policy is sensitive to the way preferences and technology are modeled, and the nonlinear system of equations describing equilibrium can give rise to a continuum of equilibria although there is local determinacy around a particular steady state. The authors model explicitly the expectations of inflation which can be self-fulfilling and ask whether a monetary authority can determine a unique equilibrium by choosing a specific expectation process, and make this process the only possible expectations process compatible with the equilibrium (the monetary authority can anchor agents' expectation of inflation). The authors analyze two types of policy instruments for anchoring agents' expectations: (1) a generalized interest rate rule or a term structure rule by fixing the short-term interest rate to a policy of fixing the interest rates on government bonds of several maturities; (2) a forward guidance rule or an expected future interest rate rule associated with each possible current inflation rate, the short-term interest rate and the future short-term interest rate which is expected to prevail for a sequence of \(T\) periods into the future. The two theoretical approaches are closely related to recent innovations in monetary policy by the Federal Reserve and the other central banks: quantitative easing seeking to influence the long-term bond price and forward guidance on the expected path of the future short-term policy rate (see [\textit{B. S. Bernanke}, ``The effects of the great recession on central bank doctrine and practice'', The B. E. Journal of macroeconomics 12, No. 3, 1--10 (2012; \url{doi:10.1515/1935-1690.120})]). The model has the advantage that the monetary policy can be made of a function of a simple observable variable (the realized inflation) rather than a function of the myriad contingencies both fundamental and ``sunspot'' to serve the realized inflation explanation. Section 2 starts with a simple model of an exchange economy with a representative agent and a monetary-fiscal authority financing an exogenously given debt inherited from the past and examines whether the monetary authority can link the price level. The agent is to maximize a utility function given by a sequence of consumption-portfolio choices in (1) subject to the agent's transactions and money holdings in period \(t\) in (2), under the assumptions that an agent cannot directly consume his/her endowment but must spend money to buy his/her consumption from another agent and that financial markets open at the beginning of each period and agents pay taxes. Then authors break to two cases, (i) taxes do not adjust to the current nominal government debt (a tax that takes the infinite stream of future real taxes as exogenously given is referred to as a non-Ricardian tax policy); (ii) taxes are adjusted to accommodate the current nominal government debt (the tax policy always ensures that the government's debt does not grow indefinitely like a Ponzi scheme as the fiscal policy referred to as a Ricardian tax policy). The Ricardian tax policy case is what the paper under review is about to study. In Section 3 the authors replace the deterministic model nominal interest rate by the real interest rate under the Ricardian tax policy. They use the Fisher equation for an economy in which the market clearing equations hold for goods at every date and agents are uncertain about the purchasing power of money in the next period. The study of the stochastic process of beliefs can be reduced to agents' expectations if the central bank can control a sufficiently large number of nominal bonds with different maturities. A Markov process of beliefs is represented by a Markov matrix with transition matrix of probabilities that inflation is within a certain level in next period under the current inflation. Proposition 1 shows that a Markov matrix represents expectations which can be anchored if the bond prices of maturities \((1, 2, \dots, S-1)\) which are compatible with the Markov matrix are such that the matrix of bond payoffs satisfies \[ \text{rank}\, [I, q^1, \dots, q^{S-1}] = S.\tag{R1} \] A necessary condition for an expectation matrix to yield bond prices satisfying (R1) is that the matrix is invertible. Section 4 formalizes the policy of announcing the path of the expected future (short-term) interest rate by a rule which associates with each inflation rate a sequence of future expected short-term bond prices. Bernanke [loc. cit.] specified ``forward guidance about the future path of policy rates'' with monetary policy based on a term structure rule like in Section 3, and the Federal Reserve has begun to make announcements of the path of the expected future short-term rate and is considering making such announcements a systematic policy. Proposition 3 shows that the Markov matrix is uniquely determined by a forward guidance rule \((q^1, E^1, \dots, E^T)\) for the expected future short-term interest rates if \(T=S-1\) and \[ \text{rank}\, [\frac{\delta}{1+\pi}, q^1, E^1, \dots, E^{T-1}] = S.\tag{R2} \] Note that i.i.d. expectations are never compatible with (R1). However i.i.d. expectations are compatible with (R2) in the special case \(S=2\), and Proposition 4 shows that this is the only case where i.i.d. expectations satisfy condition (R2). Section 5 extends previous theoretical approaches to more realistic models of economies in which goods are produced, the technology is affected by real shocks, and production decisions are affected by the nominal interest rate. A New Keynesian model has the most realistic description of the feedback between nominal and real variables, adhering instead to the simpler flexible price model of production introduced by \textit{R. E. Lucas jun.} and \textit{N. L. Stokey} [Econometrica 55, 491--513 (1987; Zbl 0613.90007)]. The authors focus on the case where the monetary policy involves fixing the term structure of interest rates in this section by assuming that productivity of the shocks are independent of inflation and agents' expectations of inflation-productivity next period are given by the joint probability distribution. The transition matrix for real shocks is exogenously given and independent of the realized inflation. The role of the monetary policy is to anchor agents' expectations of inflation and partially to undo the effect of the productivity shocks on production through an interest rate policy. Proposition 5 provides that the family of Markov matrices \((B^g)_{g\in G}\) represents inflation expectations which can be anchored by a term structure policy if the compatible bond prices averaged over the productivity shocks satisfy \[ \text{rank}\, [{\mathbf 1}, E^g(q^1), \dots, E^g(q^{S-1})]=S, \;\;\;\text{for each \(g\in G\)}.\tag{R3} \] Condition (R3) is essentially the same as (R1) for an exchange economy. Section 6 devotes to an example of inflation targeting as rank conditions (R1), (R2) and (R3) are not practical to enter for the choice of an inflation process. The authors consider an economy in which a central bank has a target rate and seeks to induce the expectation that inflation reverts to the target rate whenever current inflation deviation from the target in order to set the previous section into operation. They take \(G=1\) (no real shock) in Section 5 and a specified interval of inflation rates. The authors show how the analysis can be applied for selecting a mean-reverting process around a target. However, the target rate and the minimum variance criterion were not derived from first principles. In this sense, this paper is the first step toward a richer analysis that simultaneously studies not only the determinacy but also the optimality of the monetary policy. It will be interesting to see the authors' more complete analysis along the price rigidities studied in the New Keynesian models, and how to practice the the term-structure rule and forward guidance rule in real world.