Discussion of
“Downside Risk at the Zero Lower Bound” by Susanto Basu and Brent Bundick
Taisuke Nakata Federal Reserve Board of Governors
November 2014
Questions
This paper asks two related questions: I
Positive: How does an increase in uncertainty affect the economy when the policy rate is at the zero lower bound (ZLB)?
I
Normative: How should the central bank respond to fluctuations in uncertainty?
Methodology The paper tries to answer these questions in the context of a stylized New Keynesian model. I
To answer the positive question, they calibrate the model to match the unconditional volatility, and the variation over time of conditional volatilities, of key variables, and study how the uncertainty shock propagates.
I
To answer the normative question, they characterize optimal monetary policy with commitment in response to uncertainty shocks.
Results
I
Positive: Uncertainty is more destabilizing at the ZLB than away from the ZLB. A 50 percent increase in shock uncertainty leads to... I I
I
About 0.2 and 0.4 percent reductions in π and y when at the ZLB. Less than 0.1 percent reductions in π and y when away the ZLB.
Normative: The central bank promises to keep the policy rate at the ZLB for longer and promises to generate larger overshooting of inflation and output gap.
Intuition I
A key force: an increase in uncertainty reduces the household demand via precautionary motives.
I
The effects of uncertainty are larger at the ZLB because reductions in the household demand due to precautionary motives are not offset by an increase in the policy rate. I
I
I
Reminiscent of the mechanism behind the “larger government spending multipliers at the ZLB” result. The authors describe this intuition elegantly using a third-order Taylor expansion of the consumption Euler equation.
In response to an increase in uncertainty, the central bank promises larger overheating and a longer ZLB episode, similar to how the central bank responds to a negative demand shock.
I
The paper asks policy-relevant questions.
I
The paper is well-written and technically competent. I
I
solves a fully nonlinear stochastic model using projection methods.
Two minor comments/suggestions: I I
Mechanism Empirical relevance of the price-level targeting.
I
The authors argue that the adverse effects of uncertainty at the ZLB in the paper is due to a “precautionary motive” channel, as opposed to what the authors call a “contractionary bias” channel (Section 5.2).
I
“Contractionary bias”: Since the policy rate is a truncated variable, an increase in uncertainty increases the expectations of policy rates in the future (discussed in Johannsen (2014), Nakata (2013), and Nakov (2008)).
I
The paper assumes that this contractionary bias channel is eliminated under the price-level targeting rule and concludes that the adverse effects of uncertainty are entirely due to “precautionary motives.” I
“Adding a small weight on the price-level in our simple policy rule automatically removes the contractionary bias by offsetting any deflation with equivalent inflation in the future.” (from page 20)
I
We can check the validity of this assumption by studying the effects of uncertainty in a semi-loglinear model where precautionary motives are absent by construction. I
Step 1: Solve the perfect foresight (PF) version of the semi-loglinear economy with the PLT.
I
Step 2: Solve a stochastic version of the semi-loglinear economy with the PLT.
I
Step 3: The difference between the PF and stochastic economies are by construction due to “contractionary bias channel.”
Figure: IRFs: Price-level Targeting Rule, Stochastic vs. Deterministic
Π (φp=1.0) 1
2
0
Annualized %
Annualized %
R (φp=1.0) 2.5
1.5 1 0.5 0
2
4
6 Time
8
10
−1 −2 −3 −4
2
4
6 Time
8
% Deviation from the Det. Steady State
C (φp=1.0) 0 −1
Stochastic Deterministic
−2 −3 −4 −5
2
4
6 Time
8
10
10
I
My example is, of course, contingent on the specific parameter values chosen.
I
It would be useful if the authors provide us with some evidence behind their assumption that the contractionary bias channel is eliminated under their parameterization.
I
Mechanism
I
Empirical relevance of the price-level targeting rule
I
They model monetary policy as following a mixture of inflation and price-level targeting. I I
I
This is a reasonable modeling approach to increase the persistence of the ZLB episode. I
I
I
rt = max rtd , 0 rtd = r + φπ (πt − π) + φx xt + φpl (pt − p ∗ )
Difficulty of generating a persistent ZLB episode can be seen in Fernandez-Villarverde et al. (2012), among many others. This difficulty is systematically analyzed by Richter and Throckmorton (2013) and Nakata and Schmidt (2014).
But, do we think PLT is a realistic description of monetary policy in the U.S.?
PLT
I
It is an unusual modelling assumption. I I
I
The Fed’s policy is typically modelled as an inflation-targeting rule. The Fed also explicitly declared itself as an inflation targeter in January 2013.
As in optimal commitment policy, the PLT implies overshooting of inflation. I I
Inconsistent with data (below-trend inflation last several years). Inconsistent with any forecasts of inflation (for example, SEP and Blue Chip).
Former Governor J. Stein: I understand how the model works when you write down a model in which you say that it might make sense to proactively in some sense try to create inflation. My own view is that’s just not right. I just disagree fundamentally with that. I think it would be a mistake to seek higher inflation not only because it would do damage to our inflation leg of the mandate. I don’t accept the proposition that actively seeking higher inflation would be helpful on the activity and employment leg of the mandate. *In response to a question by Former Vice Chairman Donald Kohn at the Brookings Institution (October 11, 2012). Source: “Uncorrected Transcipt: Unconventional Times, Unconventional Measures: A Conversation with Federal Reserve Board Governor Jeremy Stein” www.brookings.edu/events/2012/10/11-fed-reserve-stein
I
Challenge: Policy rules that can generate persistent ZLB episodes tend to generate overshooting of inflation. I
I
Nominal-income targeting rule, the Reifschnieder-Williams Rule, and an inertial Taylor rule with the lagged shadow rates suffer from the same problem.
Wanted: policy rules consistent with a prolonged ZLB episode and undershooting of inflation.
Summary
I
A must-read paper for anyone interested in the policy implications of the ZLB. I I
I
Interesting answers to policy relevant questions. Technically competent.
Clarifications of the mechanism and some discussions on the empirical plausibility of using the PLT rule would further improve the paper.
References J. Fern´ andez-Villarverde, G. Gordon, P. Geurr´ on-QUintana, and J. Rubio-Ram´ırez. (2012). “Nonlinear Adventures at the Zero Lower Bound” NBER Working Paper No. 18058. B. Johannsen (2014). “When are the Effects of Fiscal Uncertainty Large?” FEDS 2014-40, Federal Reserve Board J. Stein (2012) “Uncorrected Transcipt: Unconventional Times, Unconventional Measures: A Conversation with Federal Reserve Board Governor Jeremy Stein” The Brookings Institution. T. Nakata (2013). “Uncertainty at the Zero Lower Bound,” FEDS 2013-40, Federal Reserve Board T. Nakata and S. Schmidt (2014). “Conservatism and Liquidity Traps,” Mimeo A. Nakov (2008). “Optimal and Simple Monetary Policy Rules with Zero Floor on the Nominal Interest Rate,” International Journal of Central Banking. A. Richter and N. Throckmorton (2014). “The Zero Lower Bound: Frequency, Duration, and Numerical Convergence,” B.E. Journal of Macroeconomics.