Banque nationale de Belgique SA · EURONEXT:BNB

Banque nationale de Belgique SA — Technical Presentation (March 2013)

44 slides from Banque nationale de Belgique SA's technical presentation, March 2013

  1. Slide 1: Macroeconomic Dynamics Near the ZLB: A Tale of Two Equilibria

    Macroeconomic Dynamics Near the ZLB: A Tale of Two Equilibria S. Borağan Aruoba University of Maryland Frank Schorfheide University of Pennsylvania, CEPR, NBER March, 2013

  2. Slide 2: What We Do

    What We Do ZLB on nom. interest rate has become empirically relevant for U.S. Once the ZLB is explicitly included in a New Keynesian DSGE model, it has many equilibria. We consider three: “the” equilibrium near the targeted-inflation steady state; a minimal state-variable equilibrium near the deflat

  3. Slide 3: Some Related Work

    Some Related Work As in Judd, Maliar, and Maliar (2011),Fernandez-Villaverde, Gordon, Guerron-Quintana, and Rubio-Ramirez (2012) and in Gust, Lopez-Salido and Smith (2012), we are using projection methods to solve a stochastic model with 2 endogenous and 3 exogenous states: JMM2011, FVGGQRR2012 and

  4. Slide 4: Roadmap

    Roadmap Multiple equilibria in a two-equation model A New Keynesian DSGE model with ZLB constraint Multiple steady states Local dynamics near deflation and targeted-inflation steady states Nonlinear solution Quantitative Analysis Extract model state variables from U.S. data Model dynamics Conduct po

  5. Slide 5: Two-Equation Model

    Two-Equation Model Adapted from Benhabib, Schmitt-Grohe, and Uribe (2001) and Hursey and Wolman (2010). Fisher equation: Rt = r Et [πt+1 ] Monetary policy rule ( ψ ) πt Rt = max 1, r π∗ exp[σt ] , t ∼ iidN(0, 1), ψ>1 π∗ Combine: ( ψ ) 1 πt Et [πt+1 ] = max , π∗ exp[σt ] r π∗ Model has two steady sta

  6. Slide 6: Local Dynamics Near Steady States

    Local Dynamics Near Steady States We use “hat”s to denote percentage deviations from the respective steady states Dynamics near targeted-inflation steady state (π∗ , R∗ = r π∗ ): Write Et [π̂t+1 ] = max {− ln(r π∗ ), ψπ̂t + σt } For small σ the ZLB is essentially non-binding: Et [π̂t+1 ] = ψπ̂t + σt

  7. Slide 7: Equilibria Considered in this Paper

    Equilibria Considered in this Paper Returning to the original nonlinear difference equation... Mimicking the local dynamics around the targeted-inflation steady state 1 πt = π∗ γ∗ exp − σt . ψ Mimicking similar dynamics around the deflation steady state 1 πt = π∗ γD exp − σt . ψ Sunspot equilibria t

  8. Slide 8: Inflation Decision Rules in Simple Model

    Inflation Decision Rules in Simple Model 1.015 Targeted−Inflation Deflation 1 Deflation 2 1.01 Deflation 3 1.005 1 0.995 0.99 0.985 0.98 −0.015 −0.01 −0.005 0 0.005 0.01 0.015 εR Note: Rt = 1 in the deflation equilibria and Rt = R∗ in the targeted-inflation equilibrium. B. Aruoba and F. Schorfheide

  9. Slide 9: Other Equilibria

    Other Equilibria Benhabib, Schmitt-Grohe, and Uribe (2001) equilibria that move the economy from targeted to deflation, e.g.: 1 t−t0 1 πt = π∗ max γ∗ exp − σt exp −ψ , γD exp − σt ψ ψ “Endogenous Sunspot” Equilibrium: Sunspot switches based on fundamental shocks. (with ¯ > )   1 if st−1 = 1 and t

  10. Slide 10: Inflation Dynamics in Simple Model

    Inflation Dynamics in Simple Model B. Aruoba and F. Schorfheide ZLB Dynamics

  11. Slide 11: Inflation Dynamics in Simple Model

    Inflation Dynamics in Simple Model B. Aruoba and F. Schorfheide ZLB Dynamics

  12. Slide 12: Inflation Dynamics in Simple Model

    Inflation Dynamics in Simple Model B. Aruoba and F. Schorfheide ZLB Dynamics

  13. Slide 13: Inflation Dynamics in Simple Model

    Inflation Dynamics in Simple Model B. Aruoba and F. Schorfheide ZLB Dynamics

  14. Slide 14: The Next Steps

    The Next Steps We now consider a small-scale New Keynesian model... and compute three equilibria: a targeted-inflation equilibrium, a deflation equilibrium, a Markov-switching sunspot equilibrium We solve for “minimal-state-variable” equilibria by postulating flexible functional forms for agents’ de

  15. Slide 15: “Standard” Small-Scale New Keynesian DSGE Model

    “Standard” Small-Scale New Keynesian DSGE Model No discount factor shock. Households maximize "∞ # X (Ct+s /At+s )1−τ − 1 Mt+s s Et β − Ht+s + χV . s=0 1−τ Pt+s At+s B. Aruoba and F. Schorfheide ZLB Dynamics

  16. Slide 16: “Standard” Small-Scale New Keynesian DSGE Model

    “Standard” Small-Scale New Keynesian DSGE Model No discount factor shock. Households maximize "∞ # X (Ct+s /At+s )1−τ − 1 Mt+s s Et β − Ht+s + χV . s=0 1−τ Pt+s At+s Intermediate good j is produced by a monopolist with technology: Yt (j) = At Ht (j), where lnAt = lnγ + lnAt−1 + lnzt B. Aruoba and F.

  17. Slide 17: “Standard” Small-Scale New Keynesian DSGE Model

    “Standard” Small-Scale New Keynesian DSGE Model No discount factor shock. Households maximize "∞ # X (Ct+s /At+s )1−τ − 1 Mt+s s Et β − Ht+s + χV . s=0 1−τ Pt+s At+s Intermediate good j is produced by a monopolist with technology: Yt (j) = At Ht (j), where lnAt = lnγ + lnAt−1 + lnzt Intermediate goo

  18. Slide 18: “Standard” Small-Scale New Keynesian DSGE Model

    “Standard” Small-Scale New Keynesian DSGE Model No discount factor shock. Households maximize "∞ # X (Ct+s /At+s )1−τ − 1 Mt+s s Et β − Ht+s + χV . s=0 1−τ Pt+s At+s Intermediate good j is produced by a monopolist with technology: Yt (j) = At Ht (j), where lnAt = lnγ + lnAt−1 + lnzt Intermediate goo

  19. Slide 19: “Standard” Small-Scale New Keynesian DSGE Model

    “Standard” Small-Scale New Keynesian DSGE Model No discount factor shock. Households maximize "∞ # X (Ct+s /At+s )1−τ − 1 Mt+s s Et β − Ht+s + χV . s=0 1−τ Pt+s At+s Intermediate good j is produced by a monopolist with technology: Yt (j) = At Ht (j), where lnAt = lnγ + lnAt−1 + lnzt Intermediate goo

  20. Slide 20: Equilibrium Conditions

    Equilibrium Conditions Equilibrium is {ct , πt , yt , Rt } (in terms of detrended variables, i.e., ct = Ct /At and yt = Yt /At ) " −τ # ct+1 1 Rt 1 = βEt ct γzt+1 πt+1 1 τ 1 π̄ 1 = (1 − ct ) + φ(πt − π̄) 1 − πt + ν 2ν 2ν " −τ # ct+1 yt+1 −φβEt (πt+1 − π̄)πt+1 ct yt 1 φ 2 ct = − (πt − π̄) yt gt 2  "

  21. Slide 21: Two Steady States

    Two Steady States Let λ = ν(1 − β), assume ψ1 > 1. Targeted-Inflation steady state inflation equals π∗ and R∗ = r π∗ " 2 #1/τ φ λ2 φ 1−λ 2 c∗ = 1 − v − (1 − 2λ) π∗ − π̄ + π̄ 2 1 − 2λ 2 1 − 2λ Deflation steady state nominal interest rate is RD = 1 and: πD = 1/r " 2 #1/τ φ λ2 φ 1−λ cD = 1 − v − (1 − 2

  22. Slide 22: Approximate Dynamics In Targeted-Inflation Equilibrium

    Approximate Dynamics In Targeted-Inflation Equilibrium Let τ = 1, γ = 1, π̄ = π∗ , ψ1 = ψ, ψ2 = 0, and ρR = ρg = ρz = 0. Linearizing around targeted-inflation steady state yields the system R̂t = max − ln(r π∗ ), ψπ̂t + σR R,t ĉt = Et [ĉt+1 ] − (R̂t − Et [π̂t+1 ]) π̂t = βEt [π̂t+1 ] + κ∗ ĉt , The

  23. Slide 23: Sketch of Solution Method

    Sketch of Solution Method Approximate decision rules for π(St ) and E(St ). πt = π(St ; Θ) = ζt fπ1 (St ; Θ) + (1 − ζt )fπ2 (St ; Θ) Et = E(St ; Θ) = ζt fE1 (St ; Θ) + (1 − ζt )fE1 (St ; Θ) −τ where ζt = I {R(St ) > 1} and Et ≡ IE t ct+1 /γzt+1 πt+1 fji are linear combinations of a complete set Cheb

  24. Slide 24: Sketch of Solution Method

    Sketch of Solution Method Given a solution, we run an auxiliary particle filter with fixed parameters to obtain filtered states. Iterative process (ergodic-set method with a twist): Start with a guess for Θ. Use simulated (and filtered) states to obtain a “representative” grid. Use a time-separated

  25. Slide 25: Solution Grid

    Solution Grid New Grid 0.2 1.1 0.15 1.05 y-1 g 0.1 1 0.95 0.05 1 1.005 1.01 1.015 1.02 1 1.005 1.01 1.015 1.02 R-1 R-1 0.04 0.02 Ergodic Distribution 0.015 2008Q4 onwards 0.02 Up to 2008Q3 0.01 0 R 0.005 z -0.02 0 -0.005 -0.04 -0.01 1 1.005 1.01 1.015 1.02 1 1.005 1.01 1.015 1.02 R-1 R-1 B. Aruoba

  26. Slide 26: Sample Decision Rules - Targeted-Inflation Equilibrium

    Sample Decision Rules - Targeted-Inflation Equilibrium Interest Rate Inflation 10 4 8 3 6 2 4 1 2 0 0 0.05 0.1 0.15 0.2 0.05 0.1 0.15 0.2 g g Output Consumption 0.94 1.15 0.93 0.92 1.1 0.91 1.05 0.9 1 0.89 Piece−wise Smooth Smooth 0.95 0.88 0.05 0.1 0.15 0.2 0.05 0.1 0.15 0.2 g g Gray bands represen

  27. Slide 27: Quantitative Analysis

    Quantitative Analysis 1 Parameter values (from estimating a non-linear version using second-order perturbation, Aruoba, Bocola and Schorfheide, 2011): 1 ν = 0.1, g∗ = , γ = 1.059, τ = 1.09, ψ1 = 1.5, ψ2 = 0.8, 0.85 ρR = 0.55, r = 1.0076, π∗ = 1.007, π̄ = 1, φ = 221.70, ρG = 0.89, ρZ = 0.26, σR = 0.0

  28. Slide 28: Ergodic Distributions

    Ergodic Distributions Targeted-Inflation Eq. Deflation Eq. 4 x 10 15000 8 6 10000 4 5000 2 0 0 10 8 4 6 6 0 4 4 2 −1 2 2 −2 0 −3 R (annualized %) 0 −4 π (annualized %) R (annualized %) π (annualized %) Sunspot Equilibrium 10 8000 5 6000 R (annualized %) 8 4 R (annualized %) 4000 6 3 2000 4 0 2 10 2

  29. Slide 29: Simulated Paths

    Simulated Paths Nominal Interest Rate (annualized %) Inflation (annualized %) 8 Deflation Eq. 4 Targeted−Inflation Eq. 6 2 4 0 2 −2 0 50 100 150 50 100 150 Time Time Shocks Output 2 1.1 1 1.08 0 1.06 −1 1.04 −2 1.02 g z εR −3 50 100 150 50 100 150 Time Time B. Aruoba and F. Schorfheide ZLB Dynamics

  30. Slide 30: Simulated Paths

    Simulated Paths Nominal Interest Rate (annualized %) Inflation (annualized %) 8 4 6 2 0 4 −2 2 −4 0 50 100 150 50 100 150 Time Time Government Spending Shock Output 3 Government Spending 1.12 Technology 2 Monetary Policy 1.1 1 1.08 0 1.06 −1 Sunspot Eq. −2 1.04 Targeted−Inflation Eq. 50 100 150 50 1

  31. Slide 31: Properties of Equilibria

    Properties of Equilibria Inflation responds negatively to a positive demand shock in the deflation equilibrium and in the deflation state of the sunspot equilibrium. Similar to the findings in Eggertson (2009). AD becomes upward sloping. Log-linearized AD equation. ŷt = Et ŷt+1 − τ (it − Et πt+1 )

  32. Slide 32: Extracting Historical Shocks

    Extracting Historical Shocks We now use a(n auxiliary) particle filter to extract the latent states for our model. Data: output growth, inflation (annualized), and interest rates (annualized) from 2000:Q1 to 2010:Q3 Show results for year-on-year CPI inflation. We consider the targeted-inflation equi

  33. Slide 33: Data and Historical Shocks

    Data and Historical Shocks Federal Funds Rate GDP Growth Inflation 7 2 6 6 5 1 4 5 0 3 4 2 3 -1 1 2 0 -2 1 -1 0 -3 -2 2000 2002 2004 2006 2008 2010 2000 2002 2004 2006 2008 2010 2000 2002 2004 2006 2008 2010 Demand Shock (g-hat) Technology Shock (z-hat) Monetary Policy Shock (er-hat) 4 4 .6 2 3 .4 0

  34. Slide 34: Policy Experiments

    Policy Experiments Standing at the beginning of 2009:Q1 and taking only the filtered states in 2009:Q1 as given, we consider 1 a fiscal policy intervention calibrated to ARRA; 2 a combination of the fiscal policy intervention with an expansionary monetary policy that lasts for one year. Mechanics: c

  35. Slide 35: Calibration of Fiscal Policy Intervention

    Calibration of Fiscal Policy Intervention Fiscal policy intervention is calibrated to portion of the American Recovery and Reinvestment Act (ARRA) of February 2009: Tax cuts and benefits; entitlement programs; funding for federal contracts, grants, and loans; Convert expenditures into ĝtARRA and co

  36. Slide 36: Targeted-Inflation vs. Sunspot Equilibrium: Fiscal Policy

    Targeted-Inflation vs. Sunspot Equilibrium: Fiscal Policy 100(ln(YI) − ln(Y)) INFI − INF (% annualized) 1 0.5 0.8 0 0.6 0.4 −0.5 0.2 0 −1 2 4 6 8 10 12 2 4 6 8 10 12 I I R − R (% annualized) R (% annualized) 1 10 8 0.5 6 4 0 2 −0.5 0 2 4 6 8 10 12 2 4 6 8 10 12 Figure depicts pointwise medians (soli

  37. Slide 37: Government Spending Multipliers

    Government Spending Multipliers Sunspot Equilibrium Targeted-Inflation Eq. µµC µC 1.5 1 1 0.8 0.8 1 0.6 0.6 0.4 0.4 0.5 0.2 0.2 10 12 0 0 0 2 2 4 4 66 88 1010 12 12 2 4 6 8 10 12 Pt η C η (YτI − Yτ ) ηC c 0.12 Multiplier: 0.2 µt = Pτt =1 I ; 0.2 0.1 τ =1 (Gτ − Gτ ) 0.15 0.08 0.15 0.06 0.1 0.1 0.04 0

  38. Slide 38: Expansionary Fiscal and Monetary Policy

    Expansionary Fiscal and Monetary Policy At the beginning of 2009:Q1 the Fed contemplates to amplify the effect of the expansionary fiscal policy by an expansionary monetary policy that keeps interest rates at or near zero. Recall monetary policy rule:  "  ψ1 ψ2 #1−ρR  πt Yt ρR σR R,t  Rt = max 1

  39. Slide 39: Sunspot Equilibrium: Fiscal and Monetary Policy

    Sunspot Equilibrium: Fiscal and Monetary Policy Figure depicts pointwise medians (solid); 20%-80% percentiles (shaded area) for only fiscal intervention and both interventions. B. Aruoba and F. Schorfheide ZLB Dynamics

  40. Slide 40: Targeted-Inflation Equilibrium: Fiscal and Monetary Policy

    Targeted-Inflation Equilibrium: Fiscal and Monetary Policy Figure depicts pointwise medians (solid); 20%-80% percentiles (shaded area) for only fiscal intervention and both interventions. B. Aruoba and F. Schorfheide ZLB Dynamics

  41. Slide 41: Government Spending Multipliers

    Government Spending Multipliers Sunspot Equilibrium Targeted-Inflation Eq. C µµ µC 43 3 2.5 2.5 3 2 2 1.5 2 1.5 1 1 1 0.5 0.5 00 0 22 44 66 88 10 10 12 12 2 4 6 8 10 12 ηηC ηC 0.4 0.4 0.4 0.3 0.3 0.3 0.2 0.2 B. Aruoba and F. Schorfheide 0.2 ZLB Dynamics

  42. Slide 42: An Alternative Policy Exercise

    An Alternative Policy Exercise Instead of standing at the beginning of 2009Q1, do an ex-post analysis. Use the filtered shocks for 2000Q1-2010Q2. Two experiments: Fiscal intervention only, calibrated to ARRA. Small issue: Data may already include ARRA. Both fiscal and monetary intervention. Interven

  43. Slide 43: Cumulative Government Spending Multipliers

    Cumulative Government Spending Multipliers 2009Q1 2007Q1 Period Both Only Fiscal Both Only Fiscal Targeted-Inflation Equilibrium 1 1.74 1.42 1.44 0.59 2 1.69 1.54 1.85 0.66 3 1.65 1.54 2.21 0.73 4 1.76 1.64 2.46 0.78 5 1.79 1.67 2.56 0.83 6 1.79 1.68 2.62 0.88 Sunspot Equilibrium 1 0.58 0.58 1.45 0.

  44. Slide 44: Summary of Empirical Results

    Summary of Empirical Results 2008-09 is consistent with a change from the targeted-inflation regime to the deflation regime in the sunspot equilibrium. Data prefers the sunspot model over the targeted-inflation model which needs very large shocks. Ex-ante policy exercise in 2009Q1: Cumulative multip

Banque Nationale De Belgique — Technical Presentation (March 2013) on 6ix