Banque nationale de Belgique SA · EURONEXT:BNB

Banque nationale de Belgique SA — Technical Presentation (March 2009)

30 slides from Banque nationale de Belgique SA's technical presentation, Mar 18, 2009

  1. Slide 1: Discretionary Monetary Policy in the Calvo Model

    Discretionary Monetary Policy in the Calvo Model Willem Van Zandweghe1 Alexander L. Wolman2 National Bank of Belgium 18 March, 2009 1 Federal Reserve Bank of Kansas City 2 Federal Reserve Bank of Richmond, currently on sabbatical at the Bundesbank 1

  2. Slide 2: What are we doing?

    What are we doing? Studying optimal monetary policy without commitment in a model where monop. competitive firms face an exogenous, constant probability of price adjustment Why are we doing it? • Calvo model = most popular framework for applied monetary policy analysis, yet studied almost exclusivel

  3. Slide 3: Why (details)? What we find

    Why (details)? • Papers that take LQ approach (New Keynesian models) never find multiple equilibria. • In two previous papers (w/King QJE 2004; w/Khan, King FRBWP 2001), we found multiple private-sector equilibria in a Markov-perfect discretionary equilibrium. – Taylor style models, with fixed durat

  4. Slide 4: Plan of talk

    Plan of talk 1. Related Literature 2. General Model, nesting Calvo and Taylor (2-pd prices) 3. State Variables: Real vs. Nominal 4. Background: Discretionary Equilibrium in Taylor Model 5. Discretionary Equilibrium in Calvo Model (a) definition (b) algorithm and computational method (c) properties (

  5. Slide 5: Related Literature

    Related Literature • Discretionary Policy in Calvo-Style Models- LQ Approach – Clarida, Gali, Gertler; Woodford – Adam and Billi (zero bound) – many others • Nonlinear Solutions for Discretion with Sticky Prices – (Khan), King and Wolman (2001) 2004, Siu (2008) – Dotsey and Hornstein (2008) – Yun (2

  6. Slide 6: Model: households

    Model: households • Representative Xagent, infinitely lived, preferences, ∞ β j (ln(ct+j ) + χ (1 − nt+j )) , β ∈ (0, 1) j=0 – efficiency condition: χct = wt, – consumption Euler equation: 1/ct = β (1 + Rt) (1/ (ct+1π t+1)) , • Consumption is Dixit-Stiglitz aggregate:. ∙Z 1 ¸ ε−1 ε ε−1 ct = ct(z) ε

  7. Slide 7: Model: firms

    Model: firms • Technology: yt (z) = nt (z) , • Current period profits: µ ¶−ε µ ¶−ε Xt Xt d (Xt; Pt, ct, wt) = Xt ct − Ptwt ct. Pt Pt • Value ⎧ with probability η. : – Calvo firms adjust price ⎫ ⎨X ∞ ⎬ VtC = max Qt,t+j (1 − η)j d (Xt; Pt+j , ct+j , wt+j ) ; Xt ⎩ ⎭ j=0 – Taylor firms adjust price ever

  8. Slide 8: Model: optimal prices and price indices

    Model: optimal prices and price indices • Calvo optimal price: ³ ´ε µ ¶ P∞ j j Pt+j P0,t p0,t ε j=0 (1 − η) β Pt wt+j P0,t = = P∞ ³ ´ε−1 , p0,t ≡ Pt πt ε−1 j j Pt+j Pt−1 j=0 (1 − η) β Pt • Taylor optimal price: ³ ´ε µ ¶ w + β pt+1 pJT w pJT 0,t ε t pt 0,t t+1 JT P0,t Pt = · ³ ´ε−1 , p0,t ≡ , pt ≡ pt

  9. Slide 9: Model: monetary authority

    Model: monetary authority • Choose nominal money supply, Mt. Without loss of generality, view m.a. as choosing M normalized by Pt−1 (Calvo) or P0,t−1 (Taylor) • Optimizing m.a. without commitment ⇐⇒ different policymaker each period. 9

  10. Slide 10: Model: labor market clearing

    Model: labor market clearing • labor supplied by household = labor demanded by all firms: Z 1 nt = nt(z)dz. 0 • Calvo model: ∞ X nt = η (1 − η)j nj,t, j=0 nj,t = labor input in t by firm that set price in t − j. Technology ⇒ X∞ nt = η (1 − η)j cj,t, now use demand curves: j=0 ∞ µ ¶−ε µ µ ¶ ¶ nt X P0

  11. Slide 11: Model: timing

    Model: timing 1. Predet. prices (P0,t−j , j > 0) known at beginning of pd. 2. Monetary authority chooses M (equivalently m). 3. Adjusting firms set prices, all other pd.-t variables determined. 11

  12. Slide 12: Model: state variables

    Model: state variables • Want study Markov Perfect Equilibrium (i.e. no reputational equilibria, no trigger strategies.) • Need know what are payoff-relevant state variables • Principle in nominal models: one less p-r.s.v. than # predet variables • Calvo, 2 predet. (Pt−1, nt−1/ct−1) one state: µ µ ¶

  13. Slide 13: Discretionary eqlb in Taylor model (King and Wolman 2004)

    Discretionary eqlb in Taylor model (King and Wolman 2004) • No state vars., so mt constant in MPE • Given m, eqlb is fixed µ point ¶ of pricing best response fn: JT εχ ¡ JT ¢ p0,t = · (1 − θt,t+1) · mt + θt,t+1 · mt+1p0,t , ε−1 where µ ¶ε−1 p(pJT ) JT β p p0,t+1 p0,t ( JT 0,t ) θt,t+1 ≡ µ ¶ε−1 p(p0,

  14. Slide 14: Two or zero point-in-time equilibria under arbitrary monetary policy

    Two or zero point-in-time equilibria under arbitrary monetary policy • Strong complementarity generated by combination of – any candidate policy makes future nominal M increase with current p0 (firm’s optimal price increasing in future nom. marginal cost, hence M), and – positive effect of prices se

  15. Slide 15: Two or zero point-in-time equilibria under arbitrary monetary policy

    Two or zero point-in-time equilibria under arbitrary monetary policy • Strong complementarity generated by combination of – any candidate policy makes future nominal M increase with current p0 (firm’s optimal price is increasing in future nominal M), and – positive effect of prices set by other firm

  16. Slide 16: Background to discretionary eqlb in Calvo model

    Background to discretionary eqlb in Calvo model • State variable. ∆, so m = Γ (∆) in MPE • Eqlb for arbitrary m = Γ (∆) , two functions N (∆) and D (∆): N (∆) = πε [w + β (1 − η) N (∆0)] , D (∆) = πε−1 [1 + β (1 − η) D (∆0)] , where µ ¶ ε N (∆) p0 = · , ε−1 D (∆) ¡ 1−ε ¢1/(1−ε) π = ηp0 + (1 − η) 0 ε

  17. Slide 17: Discretionary eqlb defined in Calvo model

    Discretionary eqlb defined in Calvo model A policy function Γ∗ (∆) and a value function v ∗ (∆) that satisfy v ∗ (∆) = max {ln c + χ (1 − n) + βv (∆0)} m Γ (∆) = arg max {ln c + χ (1 − n) + βv (∆0)} ∗ m subject to the price index, market clearing, money demand, labor supply, definition of the state

  18. Slide 18: Properties of discretionary equilibrium

    Properties of discretionary equilibrium • “Highest level”: m, welfare etc. as functions of state – For given value of state: objective function, policy tradeoffs ∗ For given value of state and given value of m : private sector equilibrium through lens of best response function • Now particular examp

  19. Slide 19: Figure 1: Equilibrium as a function of the state

    Figure 1: Equilibrium as a function of the state A. State transition and inflation (dashed line, right scale) 1.01 inflation rate 6 steady state 1.0075 inflation 5 % 1.005 4 1.0025 state transition equation 1 3 1.0000 1.0025 1.0050 1.0075 1.0100 State variable (∆) B. Policy instrument and welfare (d

  20. Slide 20: Properties of discretionary equilibrium

    Properties of discretionary equilibrium • “Highest level”: m, welfare etc. as functions of state – For given value of state: objective function, policy tradeoffs ∗ For given value of state and given value of m : private sector equilibrium through lens of best response function 16

  21. Slide 21: Figure 3: Policymaker's objective function

    Figure 3: Policymaker’s objective function A. Welfare 198.834 198.832 198.83 198.828 198.826 State=1.000 State=1.005 198.824 0.19 0.192 0.194 0.196 0.198 0.2 0.202 0.204 0.206 0.208 0.21 m B. Current period component of welfare 1.994 1.992 1.99 1.988 1.986 1.984 State=1.000 State=1.005 1.982 0.19 0.

  22. Slide 22: Figure 4: Distortions as functions of m

    Figure 4: Distortions as functions of m A. Markup 1.2 1.18 1.16 1.14 1.12 1.1 State=1.000 State=1.005 1.08 0.19 0.192 0.194 0.196 0.198 0.2 0.202 0.204 0.206 0.208 0.21 m B. Relative price distortion 1.007 State=1.000 1.006 State=1.005 1.005 1.004 1.003 1.002 1.001 1 0.19 0.192 0.194 0.196 0.198 0.2

  23. Slide 23: Figure 5: Relative price distortion as function of p0

    Figure 5: Relative price distortion as function of p0 1.009 State=1.000 State=1.005 1.008 1.007 1.006 1.005 1.004 1.003 1.002 1.001 1 0.92 0.94 0.96 0.98 1 1.02 1.04 1.06 p0 27

  24. Slide 24: Properties of discretionary equilibrium

    Properties of discretionary equilibrium • “Highest level”: m, welfare etc. as functions of state – For given value of state: objective function, policy tradeoffs ∗ For given value of state and given value of m : private sector equilibrium through lens of best response function 16

  25. Slide 25: Figure 6: Pricing best response function

    Figure 6: Pricing best response function: State = 1.0025, m = 0.2022 1.5 1.25 1 0.75 0.5 0.25 0.25 0.5 0.75 1 1.25 1.5 p0 28

  26. Slide 26: Figure 2: Steady state inflation rate

    Figure 2: Steady state inflation rate 11 10 9 % 8 7 6 5 0.3 0.32 0.34 0.36 0.38 0.4 0.42 0.44 0.46 0.48 0.5 Fraction of firms adjusting price 24

  27. Slide 27: Comparing Calvo and Taylor

    Comparing Calvo and Taylor • Taylor: two fixed points, convex best response. If all adjusting firms raise prices very high, (a) nominal state responds one-for-one (b) individual firm optimally overcompensates, because tomorrow’s m.a. raises M one for one. Don’t want get stuck with low price and high

  28. Slide 28: Price Stickiness and inflation

    Price Stickiness and inflation • In Calvo model, can ask how steady state inflation varies with price stickiness under discretionary policy. – stickier prices ⇒ higher marginal benefit to surprise inflation (reduce markup), needs to be offset with higher marginal cost (relative price distortion) – t

  29. Slide 29: Price Stickiness and inflation

    Price Stickiness and inflation • In Calvo model, can ask how steady state inflation varies with price stickiness under discretionary policy. – stickier prices ⇒ higher marginal benefit to surprise inflation (reduce markup), needs to be offset with higher marginal cost (relative price distortion) – t

  30. Slide 30: Conclusions

    Conclusions • Discretionary monetary policy in sp models involves interesting ec. mecha- nisms, incentives for monetary authority that are missing in local approxima- tions around zero inflation: – accommodate nominal state in Taylor model⇒ multiple equilibria – counteract real state in Calvo model

Banque Nationale De Belgique — Technical Presentation (March 2009) on 6ix