Risk shocks and monetary policy in the new normal Martin Seneca Bank of England Workshop of ESCB Research Cluster on Monetary Economics Banco de España 9 October 17 Views expressed are solely those of the author and do not necessarily reflect those of the Bank of England or its committees
/ 17 Motivation Pre-crisis consensus that ZLB episodes are rare and short (Reifschneider and Williams JMCB, Schmitt-Grohé and Uribe 11) Post-crisis revision in light of incoming data (Chung et al. JMCB 1, Williams 14, Kiley and Roberts BPEA 17) Optimistic view that unconventional policy can stand in after the economy returns to normal (Reifschneider 16) But how should monetary policy be conducted if the new normal is one in which the public worry that policymakers may not always be able to provide sufficient stimulus?
3 / 17 This paper Study of the implications for monetary policy of risk and variation in risk in a new normal close to the zero lower bound (ZLB) Two key differences from pre-crisis prescriptions Policymakers should operate the economy above potential in normal times, but accept that inflation settles below target Changes in the perception of risk lead to trade-offs for monetary policy between inflation and real stability Mechanism is a negative skew in expectations when risk is high relative to the available monetary policy space Inability to respond to large adverse shocks with sufficient stimulus...but not with contractionary action when needed
4 / 17 Literature Follows studies of the implications of the presence of risk with a ZLB under optimal discretionary policy Mechanism: Adam and Billi JMCB 7, Nakov IJCB 8 Applications: Nakata and Schmidt 14, Evans et al. BPEA 15 Complements analysis of risk shocks in a non-linear model with instrument rule by Basu and Bundick 15 by featuring Dynamics away from the ZLB Optimal discretionary policy Quasi-linear model (solution and interpretation) Contemporaneous work on the stochastic steady state in non-linear model by Hills, Nakata and Schmidt 16. More broadly related literature on uncertainty and macroeconomic dynamics
5 / 17 New Keynesian model with ZLB and risk shock Quasi-linear model π t = βe t π t+1 + κx t + u t x t = E t x t+1 1 ς (i t E t π t+1 r t ) i t + i Shock processes with r t = ρ t + ε t ε t = µ ε ε t 1 + ν ε,t ; u t = µ u u t 1 + ν u,t ; ν ε,t N(,σ ε,t) ν u,t N(,σ u,t) Baseline risk shock process with ς 1 σ ε,t = σ u,t = σ t σ t = σ + µ σ (σ t 1 σ) + ν σ,t
6 / 17 Monetary policy Optimal policy under unconstrained discretion minimises subject to Targeting rule Interest rate L = π t + λx t π t = βe t π t+1 + κx t + u t π t = λ κ x t i t = max{ i,i opt t }
7 / 17 Solution algorithm Approximate the shock processes by independent Markov processes using the Rouwenhorst (Frontiers, 1995) method Solve model backwards from distant future period T with E t π t = E t x t = for all t > T Take expectations as given and calculate the unconstrained outcome for a state grid of values for the shock processes Take as solution for each node where ZLB doesn t bind Calculate outcomes from the model equations with it = i imposed for all other nodes Update the ex ante expectations using the Markov transition matrices Progress to previous period Solution if convergence in period t =
Probability 8 / 17 The new normal.4 Federal Funds Rate (US).4 Bank Rate (UK).3.3...1.1-1 1-1 1 1968-199 1993-8 New normal
9 / 17 Baseline parameterisation Parameter Description Value π Inflation target. r Normal real interest rate.1 β Discount factor.995 κ Slope of Phillips curve. ς Relative risk aversion 1 µ ε Persistence of equilibrium rate.75 µ u Persistence of cost-push shock.5 µ σ Persistence of risk shock.75 λ Weight on output gap in loss function. σ Underlying risk.7 n ε,n u Grid size for shock processes 5 T Uncertainty horizon 1
1 / 17 Alternative calibrations Data Unconstrained model Episode E(i) σ(i) E(π) σ(π) E(i) σ(i) E(π) σ(π) 1σ P(i < ) New normal 3....48.7.9 Low risk 3. 1...11.1. US 1968-199 8.7 3.16 5.96 3.73 8.7 3.16 4.16 4.66.39.1 US 1993-8 3.97 1.74.55 3.59 3.97 1.74..31..1 UK 1968-199 1.59.86 8.77 6.83 1.59.86 6.59 6.86.35. UK 1993-8 5.36 1.3 1.93.9 5.36 1.3..11.13.
11 / 17 Risk shock around low-risk steady state.4 Risk (1 ) 3.5 Interest rate (in %).3 3..5.1 1.5.5 Inflation (in %). Output gap (in %).1 1.5 1 Optimal policy -.1 Simple rule
1 / 17 Risk shocks around low-risk steady state.6 Risk (1 ) 3.5 Interest rate (in %).4 3.5. 1.5.5 Inflation (in %). Output gap (in %).1 1.5 1 -.1 r* shock only u shock only
13 / 17 Stochastic steady state Interest rate Inflation Output gap Episode i i E(i) π π E(π) x x E(x) P(ZLB) 1) New normal 3..73.81. 1.8 1.79..5.1.14 ) r shocks only 3..94.94. 1.93 1.9....7 3) u shocks only 3..98.99. 1.98 1.97..1..4 4) Large r shocks 3..78.8. 1.81 1.8..5.1.13 5) Large u shocks 3..64.8. 1.81 1.8..5.1.15 6) Lower r.77.4.41. 1.66 1.64..9.1. 7) Lower π.77.7.43 1.75 1.43 1.41..8.1. 8) Higher r 3.8 3.1 3.14. 1.88 1.87..3.1.1 9) Higher π 3.7 3.9 3.14.5.1.11..3..1 1) High π 5.4 5.3 5.3 4. 3.99 3.99....1
14 / 17 Risk shocks in new normal.35.3 Risk (1 ) 3 Interest rate (in %).5..5.15.1 1.8 1.6 Inflation (in %)..15.1.5 Output gap (in %) 1.4 1. -.5
15 / 17 Risk and economic outcomes 1 ZLB frequency 4 Interest rate (in %).8.6.4. 3 1.1..3 1 3 Inflation (in %) 1.1..3 1.1..3 1 Output gap (in %)..1 -.1 -..1..3 1
16 / 17 Normalisation with risk shock.3.5..15.1 Risk.5 Interest rate (in %) 3 1-1 1.5 Inflation (in %).5 -.5 Output gap (in %) 1-1.5 Low risk Risk shock Risk shift i* -1
17 / 17 Conclusion Risk affects outcomes in the New Keynesian model close to the ZLB In uncertain times, inflation may settle materially below target even when the policy rate is well above the ZLB Even if nothing happens, variation in the perception of risk affects the economy through expectations Risk shocks that are large relative to the available monetary policy space give rise to cost-push effects Stochastic volatility gives rise to occasional trade-offs When underlying risk is high, variation in risk has both negative and positive cost-push effects