FORWARD GUIDANCE AND THE STATE OF THE ECONOMY Benjamin D. Keen University of Oklahoma Alexander W. Richter Federal Reserve Bank of Dallas Nathaniel A. Throckmorton College of William & Mary The views expressed in this presentation are our own and do not necessarily reflect the views of the Federal Reserve Bank of Dallas or the Federal Reserve System.
WHAT IS FORWARD GUIDANCE? Central bank communication about future policy (e.g., objectives, contingencies, policy actions, etc.) We focus on communication about future policy rates. Campbell et al. (212) differentiate between two types: Delphic forward guidance: A central bank s forecast of its own policy, which is based on its projections for inflation and real GDP growth as well as an established policy rule. Odyssean forward guidance: A central bank s commitment to deviate from its policy rule at some time in the future when the policy rate is expected to rise above zero.
FORWARD GUIDANCE AT THE FED 12/8: FOMC lowered the FFR to.25% and announced it would remain low for an extended period 8/11: low rate warranted at least through mid-213 1/12: updated the date to at least through late 214 and expressed a more pessimistic economic outlook 9/12: considerable time after the economic recovery strengthens likely warranted at least through mid-215 12/12: unemployment rate remains above 6-1/2% 12/13: well past the time the unemployment rate declines 6/14: considerable time after the asset purchase program 1/15: it can be patient in beginning to normalize rates
CONTRIBUTIONS TO THE LITERATURE We examine forward guidance (FG) in a model where FG impacts the economy via news shocks to the policy rule. The news is the difference between the expected policy rates before and after the central bank s announcement. The total weight on the news is held constant to isolate the effect of a longer horizon from a larger policy shock. We show how the following impact the efficacy of FG: ZLB constraint State of the economy Size of the news shocks Speed of the recovery Forward guidance horizon We use our results to interpret the effects of recent FG
KEY FINDINGS 1. The stimulative effect of FG falls as the economy deteriorates or as households expect a slower recovery. 2. Longer FG horizons do not generate increasingly larger impact effects on output when the total amount of news is fixed, unlike with an exogenous interest rate peg. 3. In steady state, an unanticipated shock has a larger impact effect on output than a news shock, but a news shock has a larger cumulative effect in every state of the economy. 4. At the ZLB, the cumulative effect of a longer FG horizon increases over short horizons but decreases thereafter. 5. FG is stimulative in the absence of other shocks, but the observed effect on output is smaller or even negative if another shock simultaneously reduces demand.
Households: KEY MODEL FEATURES Value consumption and leisure with preferences E t= β t [logc t χn 1+η t /(1+η)] where χ >, β 1 and β t = t j=1 β j for t > Cashless economy and bonds are in zero net supply Monopolistically competitive intermediate firms: Choose inputs to minimize costs Choose prices to maximize the present value of profits subject to a quadratic price adjustment cost Perfectly competitive final goods firm: Combines the intermediate inputs to produce a final good Choose intermediate inputs to maximize profits
CENTRAL BANK AND FORWARD GUIDANCE Households receive forward guidance about future policy through a discretionary monetary policy shock. Central bank sets the nominal interest rate according to i t = max{ī,it }, i t = ī(π t / π) φπ (y t /ȳ) φy exp(x t ), x t q j= α jε t j, q j= α j = 1, x: news (either anticipated or unanticipated) ε N(,σ 2 ): monetary policy shock αj [,1]: weight on the shock j periods in the future q: forward guidance horizon The restriction on the weights of the shocks allows us to isolate the effect of a longer horizon from a larger shock.
ADVANTAGES OF NEWS SHOCKS A way to model innovations in households expectations Enables the policy rate to endogenously respond to changes in economic conditions, whereas an interest rate peg fixes the policy rate regardless of economic conditions Households expectations incorporate the possibility that the central bank alters its previous forward guidance policy Households form expectations about the possibility the central bank will provide news that it plans to exit the ZLB Allows us to isolate the effects of different FG horizons They are important for matching data [Gomes et al. (213); Milani and Treadwell (212); Campbell et al. (212)]
COMPETITIVE EQUILIBRIUM Consists of sequences of quantities {c t,n t,y t } t=, prices {w t,i t,π t } t=, and shocks {β t} t= that satisfy: w t = χn η tc t 1 = i t E t [β t+1 (c t /c t+1 )/π t+1 ] y t = n t ( πt ) ϕ π 1 πt π = 1 θ +θw t +ϕe t [ c ( t πt+1 ) β t+1 c t+1 π 1 πt+1 π c t = [1 ϕ(π t / π 1) 2 /2]y t y gdp t i t = max{ī,ī(πt /π ) φπ (y t /ȳ) φy exp(x t )} β t = β(β t 1 / β) ρ β exp(υ t ) ] y t+1 y t
DATA AND CALIBRATION Calibration (Quarterly) Steady-State Disc. Factor β.9957 Lower Bound ī 1.22 Frisch Labor Supply Elasticity 1/η 3 Response to Inflation φ π 2 Elasticity of Substitution θ 6 Response to Output φ y.8 Price Adjustment Cost ϕ 16 Disc. Factor Persistence ρ β.87 Steady-State Labor n.33 Disc. Factor St. Dev. σ ε.225 Steady-State Inflation π 1.57 Policy Shock St. Dev. σ ν.3 Standard Deviations Real GDP Growth Inflation (Deflator) Interest Rate (T-Bill) Data 2.58%.99% 2.79% Model 2.45% 1.7% 2.29% (1.92%,3.67%) (.74%,1.63%) (1.83%,2.9%)
BLUE CHIP CONSENSUS FORECASTS 2.5 29 21 211 212 213 2 3-Month T-Bill (%) 1.5 1.5 1 2 3 4 5 6 7 Forecast Horizon (Quarters)
DISTRIBUTION OF ZLB EVENTS 35 Deep Recession (ĩ =.5%) 3 Frequency (%) 25 2 15 1 mean = 3.1 5 1 2 3 4 5 6 7 8 9 1 ZLB Duration (Quarters)
SOLUTION METHOD Compute nonlinear solutions using policy function iteration Linear interpolation and Gauss Hermite quadrature We are the first to study FG using a global solution method This method enhances our analysis in several ways: 1. Enables ZLB events to endogenously reoccur, which impacts households expectations of future policy rates and the central bank s ability to provide economic stimulus 2. Allows us to examine FG in any state of the economy 3. We can study FG in a setting where changes in economic conditions alter the probability and duration of a ZLB event 4. We are able to analyze FG across all possible realizations of shocks, which nonlinearly impact the economy
EXPERIMENTS: 1-QUARTER HORIZON Key Assumptions: We initialize the discount factor at ˆβt =.6, which is the minimum value of ˆβ necessary for the ZLB to bind. In the absence of any news shocks, agents expect ˆβ to gradually revert to its mean so that E t [i t+k ] >, k >. Three types of forward guidance are examined: 1. No FG (α = 1) x t = ε t 2. 1-Quarter FG (α = and α 1 = 1) x t = ε t 1 3. 1-Quarter Distributed FG (α =.13 and α 1 =.87) x t =.13ε t +.87ε t 1. These weights eliminate feedback effects on the policy rate.
.3.2.1.1.2.3 SOLUTION: 1-QUARTER HORIZON Real GDP (ŷ gdp t ).4 1.75.5.25.25.5 Monetary Policy Shock (ˆε t ).2.1.1.2 No FG 1-Quarter FG 1-Quarter Distributed FG Stimulative Effect Inflation Rate ( π t ) Feedback Effect.3 1.75.5.25.25.5 Monetary Policy Shock (ˆε t ).3.2.1.1 Exp. Int. Rate ( Et [i t+1 ]).2.3 1.75.5.25.25.5 Monetary Policy Shock (ˆε t ).4.3.2.1 Cause of the Stimulative Effect Nom. Int. Rate (ĩ t ) Feedback Effect ZLB Constraint 1.75.5.25.25.5 Monetary Policy Shock (ˆε t )
IMPULSE RESPONSES: 1-QUARTER HORIZON.3.2.1 Real GDP (ŷ gdp t ) 1 2 3 4 5 Exp. Real GDP ( E t [ŷ gdp t+1 ])).3.2.1 No FG 1-Quarter FG 1-Quarter Distributed FG 1 2 3 4 5.2.1 Inflation Rate ( π t ) 1 2 3 4 5 Exp. Infl. Rate ( Et [π t+1 ]).12.8.4 1 2 3 4 5.1.2 Nom. Int. Rate (ĩ t ).3 1 2 3 4 5 Exp. Int. Rate ( Et [i t+1 ]).1.2.3 1 2 3 4 5
IMPORTANCE OF THE ZLB CONSTRAINT.8.7.6.5.4.3.2.1 Real GDP (ŷ gdp t ) Additional Stimulative Effect (Unconstrained Model) Stimulative Effect (Constrained Model) With a ZLB Constraint 1.75.5.25 Monetary Policy Shock (ˆε t ).1.2.3.4.5.6.7.8 Without a ZLB Constraint Exp. Int. Rate ( Et [i t+1 ]) Cause of the Stimulative Effect (Constrained Model) Cause of the Additional Stimulative Effect (Unconstrained Model).9 1.75.5.25 Monetary Policy Shock (ˆε t )
FUTURE INTEREST RATE DISTRIBUTIONS % of Simulations 8 7 6 5 4 3 2 1 ZLB (ĩ t = ) E t [i t+1 ] =.23.2.4.6.8 1 Future Nominal Interest Rate (ĩ t+1 ) % of Simulations 8 7 6 5 4 3 2 1 Deep ZLB (ĩ t =.5) Et [i t+1 ] =.1.2.4.6.8 1 Future Nominal Interest Rate (ĩ t+1 )
STATE OF THE ECONOMY Steady State (1) Low State (.25) ZLB () Deep ZLB (.5) Real GDP (ŷ gdp t ) Nom. Int. Rate (ĩt) No Forward Guidance.5.4.3.2.1 1 2 3 4 5.1.1.2.3.4.5 No Forward Guidance 1 2 3 4 5 Real GDP (ŷ gdp t ) Nom. Int. Rate (ĩt) 1-Quarter Distributed FG.5.4.3.2.1 1 2 3 4 5.1.1.2.3.4.5 1-Quarter Distributed FG 1 2 3 4 5
Real GDP (ŷ gdp t ) Nom. Int. Rate (ĩt) 1.75.5.25.75 SIZE OF THE SHOCK No Forward Guidance 1.75.5.25 Monetary Policy Shock (ˆε t ).25.5 Steady State (1) Low State (.25) ZLB () Deep ZLB (.5) No Forward Guidance 1 1.75.5.25 Monetary Policy Shock (ˆε t ) Real GDP (ŷ gdp t ) Exp. Int. Rate ( Et[it+1]) 1.75.5.25 1-Quarter Distributed FG 1.75.5.25 Monetary Policy Shock (ˆε t ).25.5.75 1-Quarter Distributed FG 1 1.75.5.25 Monetary Policy Shock (ˆε t )
SPEED OF THE RECOVERY Slow Recovery (p 21 =.19) Fast Recovery (p 21 =.21).2.15 Real GDP (ŷ gdp t ) Additional Stimulative Effect (Fast Recovery).5 Exp. Int. Rate ( Et [i t+1 ]) Cause of the Stimulative Effect (Slow Recovery).1.1.5 Stimulative Effect (Slow Recovery) 1.75.5.25 Monetary Policy Shock (ˆε t ).15 Cause of the Additional Stimulative Effect (Fast Recovery).2 1.75.5.25 Monetary Policy Shock (ˆε t )
EXPERIMENTS: LONGER HORIZONS To reduce the dimensionality of our problem, the continuous distribution of the news shock is discretized using the method described in Tauchen (1986). We specify three values ( 6,,6) for each news shock and calculate the probabilities of each event. Agents learn in period 1 about a 6 basis point policy shock that is distributed over the forward guidance horizon. Four types of FG are examined: 1-quarter, 4-quarter, 8-quarter, and 1-quarter distributed FG. In each simulation, the weights (α j, j =,1,...,q) are set to eliminate any feedback effects on the policy rate.
RESPONSES AT STEADY STATE No FG 1-Quarter 4-Quarter 8-Quarter 1-Quarter.5 Real GDP (ŷ gdp t ) Nom. Int. Rate (ĩ t ).4.1.3.2.2.3.1.4 1 2 3 4 5 6 7 8 9 1 11 12.5 1 2 3 4 5 6 7 8 9 1 11 12
RESPONSES AT THE ZLB No FG 1-Quarter 4-Quarter 8-Quarter 1-Quarter.5 Real GDP (ŷ gdp t ) Nom. Int. Rate (ĩ t ).4.1.3.2.2.3.1.4 1 2 3 4 5 6 7 8 9 1 11 12.5 1 2 3 4 5 6 7 8 9 1 11 12
RESPONSES AT A NEGATIVE NOTIONAL No FG 1-Quarter 4-Quarter 8-Quarter 1-Quarter.5 Real GDP (ŷ gdp t ) Nom. Int. Rate (ĩ t ).4.1.3.2.2.3.1.4 1 2 3 4 5 6 7 8 9 1 11 12.5 1 2 3 4 5 6 7 8 9 1 11 12
CUMULATIVE EFFECT ON OUTPUT Present value of the cumulative percent change in real GDP: Cumulative Effect ŷ(q) = 1 N q+1 N j=1 t=1 1(yj,t ε /yno j,t ε 1) t k=2 r j,k, Forward Guidance Horizon Initial State of the Economy 1 4 8 1 Steady State (ĩ = 1).5.83 1.19 1.2 1.17 Recession (ĩ = ).23.51 1. 1.9 1.9 Deep Recession (ĩ =.5).11.33.87 1.3 1.4 Limit on how far the horizon can extend and add stimulus.
EFFECT OF LOWER DEMAND 4-Quarter Distributed FG Lower Demand FG + Lower Demand.2.2.4.6 Real GDP (ŷ gdp t ).3.6.9 8-Quarter Yield ((Π 7 j= E [i j ]) 1/8 ).8.12 1 1 2 3 4 5 6 7 8.15 1 2 3 4 5 6 7 8
DISTRIBUTION OF REAL GDP 25 2 Impact Effect on Real GDP (ŷ gdp 1 ) No Forward Guidance 4-Quarter Distributed FG % of Simulations 15 1 5 5 4 3 2 1 1 2
EFFECTS OF AN INTEREST RATE PEG Nominal interest rate: i t = { max{ī,i t } for e t = ī for e t = 1 FG is characterized by a vector of nominal interest rate policies, [e t,e t+1,...,e t+q ], communicated in period t over horizon q. Example: 1-Quarter FG: p 1 p F(1) = 1 1 P(1) = p 1 p p 1 p 1 1 p 1 p
EFFECTS OF AN INTEREST RATE PEG No FG 1-Q 2-Q 4-Q 6-Q Real GDP (ŷ gdp t ) Nom. Int. Rate (ĩ t ) 4 3 2 1 1 2 3 4 5 6 7 8.8.6.4.2 1 2 3 4 5 6 7 8
DRAWBACKS OF AN INTEREST RATE PEG Cannot respond to changes in economic conditions Households never expect the central bank to modify previously announced forward guidance policies An interest rate peg does not allow the effects of additional news to be separated from a longer horizon Less flexible than news shocks (i.e., a peg represents a specific sequence of news that pushes the expected nominal interest rate to zero over a given horizon)
CASE STUDY: FORWARD GUIDANCE IN 211 The FOMC announced in its August 9, 211 policy statement that it expected the federal funds rate to remain between -25 basis points until mid-213. Estimating the effect of that announcement is complicated by the fact that GDP was revised downward just 11 days before the FOMC s statement was released. To separate the impact of the events, we use consensus forecasts from the Blue Chip Financial Forecasts (BCFF) and the Blue Chip Economic Indicators (BCEI) survey.
3-MONTH T-BILL CONSENSUS FORECASTS 211Q4 212Q1 212Q2 212Q3 212Q4 BCFF (7/2-21).14.26.43.75 1.8 GDP revision (7/29) BCEI T-bill (8/4-5).13.19.29.5.77 FOMC Announcement (8/9) BCFF (8/24-25).7.9.12.14.2 Total Change.7.17.31.61.88 Change after GDP.1.7.13.25.31 Change after FOMC.6.1.18.36.57 GDP revision limited the ability of FG to reduce rates Larger decline after the FOMC than the GDP revision
REAL GDP GROWTH CONSENSUS FORECASTS 211Q4 212Q1 212Q2 212Q3 212Q4 BCFF (7/2-21) 3.9 2.75 2.97 3.7 3.17 GDP revision (7/29) BCEI T-bill (8/4-5) 2.53 2.38 2.59 2.81 2.88 FOMC Announcement (8/9) BCFF (8/24-25) 2.17 2.13 2.44 2.69 2.9 Total Change.92.62.53.38.27 Change after GDP.56.37.38.26.29 Change after FOMC.36.25.15.12.2 The statement expressed pessimism about the economy: The Committee now expects a somewhat slower pace of recovery over coming quarters... Smaller decline after the FOMC than the GDP revision
MAIN TAKEAWAYS We study FG in a New Keynesian model with news shocks The FG horizon, the state of the economy, the speed of the recovery, and the size of policy shocks all nonlinearly impact the effects of FG due to the ZLB constraint At the ZLB, the cumulative effect on real GDP from lengthening the horizon decreases beyond two years There are limits on how far FG can extend into the future and continue to add stimulus, unlike with a policy rate peg Recent FG often associated with declines in real GDP: News often accompanied by weak economic assessments Prior expectations of a weak economy gave policymakers a small margin to lower expected future policy rates