Banque Nationale De Belgique Sa · EURONEXT:BNB
Banque Nationale De Belgique Sa — Technical Presentation (April 2005)
37 slides from Banque Nationale De Belgique Sa's technical presentation, April 2005
Slide 1: New-Keynesian Macroeconomics and the Term Structure
New-Keynesian Macroeconomics and the Term Structure Geert Bekaert, Seonghoon Cho and Antonio Moreno April 2005
Slide 2: Macro 102: Standard New Keynesian Dynamic System
Macro 102: Standard New Keynesian Dynamic System 1) AS equation (Phillips curve): Inflation dynamics - Inflation depends on real variables; unemployment, output - Concept of natural rate of unemployment or output 2) IS equation: Output dynamics - Output is linked to past and future output and the re
Slide 3: Dual Motivation
Dual Motivation Standard New-Keynesian (NK) Macro Model: has some clear deficiencies: Truly Minimalist (e.g. Clarida, Galí and Gertler (1999)): Only 3 equations: AS, IS, MP rule Only 3 variables: inflation, output, short-term interest rate Very Limited Information Sets Does not generate enough persi
Slide 4: Models of the Term Structure of Interest Rates
Models of the Term Structure (T-S) of Interest Rates In principle, no link from Factors to Economics (e.g. Dai and Singleton (2000)) Lack of Economic Interpretation of the Term Structure Typically do not match the yield data
Slide 5: Goals in this paper
Goals in this paper Enlarge the NK Macro Model with relevant unobservable variables which are filtered out using Term Structure Information Provide a Better Fit of the data Get more reliable Estimates of structural parameters Enrich Agents´ Information Sets: Add realism Nest the NK Macro Model into
Slide 6: Outline Talk
Outline Talk I. Related Literature/Contributions II. Model - Motivation New-Keynesian Model - Term Structure Model III. Data and Estimation; Model Selection IV. Empirical Results - Macro - Term Structure V. Extensions 6
Slide 7: Related Literature/Motivation
Related Literature/Motivation NK Macro Model Estimation: Rotemberg, Woodford (1999), Ireland (2000), Cho, Moreno (2004) Phillips Curves and IS Curves: Galí, Gertler (1999), Fuhrer, Rudebusch (2004) - weak transmission mechanism - hard to generate persistence 7
Slide 8: Related Literature/Motivation
Related Literature/Motivation Empirical links between spreads and macro variables: - Estrella, Mishkin (1998); Harvey (1988); Ang, Piazzesi and Wei (2004) Term structure variables predict real growth - Diebold, Rudebusch, Aruoba (2003): Macro variables predict term structure variables Persistence Pu
Slide 9: Related Literature/Motivation
Related Literature/Motivation Macro finance: - Ang, Piazzesi (2003): Gaussian price of risk model rt a b Xt Xt Xt 1 t Xt [3 latent variables, 2 macro variables] and block diagonal! Macro factors particularly important for the spread factor - Competing papers: Rudebusch, Wu (2003), Hördal, Tristani,
Slide 10: Key Features of our NK Model
Key Features of our NK Model DSGE model with Monopolistic competition with Inflation Indexation ( AS ) External Habit Persistence ( IS ) Modified Taylor Rule ( MP ) Plus Two unobservables
Slide 11: Unobservables
Unobservables Time-Varying Inflation Target: t * * Central Bank targets t Natural Rate of Output: y n t Output Gap is yt y n t Central Bank targets y n t 11
Slide 12: IS Curve
IS Curve Utility function: 1 1 Ft C t 1 Nt U (C t , N t ) , Ft C t 1G t 1 1 - Exogenous Habit - Separable Labour Supply yt Et yt 1 (1 ) yt 1 (it Et t 1 ) IS ,t 1 ;
Slide 13: AS Curve
AS Curve Monopolistically competitive firms produce goods but prices are partially sticky: t Et t 1 (1 ) t 1 wsˆt where = f (the degree of indexation to previous inflation) w = f (proportion "optimizing" firms) sˆt = log real marginal cost Natural rate of output (Yt n ): output that would prevail at
Slide 14: Inflation Target
Inflation Target Anchored in the expectations of long run inflation, adjusted for lagged inflation target. LR j t (1 d) j 0 d Et t j LR LR t dE t t 1 (1 d) t * * LR t t 1 (1 ) t * ,t * * * t 1Et t 1 2 t 1 3 t * ,t Also more standard specification: * * t 2 t 1 3 t 1 * ,t 14
Slide 15: Enlarged NK Model
Enlarged NK Model n t Et t 1 (1 ) t1 ( yt yt ) AS ,t yt Et yt 1 (1 ) yt 1 (it Et t 1 ) IS ,t n n y t yt 1 y n ,t * * * t E 1 t t 1 2 t 1 3 t * ,t * it it 1 (1 )i t MP,t * * n i t it ( Et t 1 t ) ( yt yt )
Slide 16: Desired Nominal Interest Rate
Desired Nominal Interest Rate i *t it ( Et t 1 * t ) ( yt ytn ) (1) i t r Et t 1 ( 2) i t rt Et t 1 (3) i t i 1 n n n rt [ Et y t (1 ) yt 1 yt ] 16
Slide 17: Rational Expectations Equilibrium
Rational Expectations Equilibrium Structural Model Bxt AEt xt 1 Jxt 1 C t n * xt t , y t , it , yt , t Reduced Form Model xt xt 1 t subject to non-linear restrictions Serve as State Variable Dynamics
Slide 18: Term Structure (T-S) Model
Term Structure (T-S) Model IS Curve Implied by: E t M t 1 (1 it ) 1 Implied Pricing Kernel: Yt 1 Yt G t 1 1 Mt 1 t 1 Yt Yt 1 G t Bond pricing Et M t 1 Pn 1,t 1 Pn ,t Pn , t n-period bond price
Slide 19: Term Structure (T-S) Model
Term Structure (T-S) Model Implied Pricing Kernel (in logs): mt 1 ln( ) yt 1 ( ) yt yt 1 t 1 gt 1 gt Bond pricing falls into the Affine class: p n ,t an b n xt p n ,t n-period bond price a n constant risk premium (EH embedded) bn factor loading
Slide 20: Term Structure (T-S) Model
Term Structure (T-S) Model an bn Bond yield: i n ,t xt n n an bn Yield Spread: sp n ,t ( e3 )' xt n n
Slide 21: Strategy
Strategy Filter unobservable inflation target and the output gap through the term structure. Macro Model Solution expressed as: n * xt xt 1 t xt t , y t , it , y t , t T-S Model Implied by NK System: zt Bz xt zt t , y t , it , sp n , t , sp n , t 1 2 Full System of Observables can be estimated: zt z
Slide 22: Data and Estimation
Data and Estimation Model estimated with U.S. data (1961-2003): CPI inflation, Detrended Output, 3 month T-Bill, 3 and 5 year spreads GMM Estimation Approach with Bootstrapped Standard Errors - First stage with “theoretical” weighting matrix - Iterate till convergence - 7 of 12 specifications conver
Slide 23: Autocorrelogram: Criterion to Choose among Models
Autocorrelogram: Criterion to Choose among Models
Slide 24: Estimated NK Model
Estimated NK Model n t 0.61* Et t 1 0.39 * t 1 0.06 * ( yt yt ) AS ,t yt 0.42 * Et yt 1 0.58* yt 1 0.13* (it Et t 1 ) IS ,t it 0.72 * it 1 0.28* i *t MP , t n i *t Et t 1 0.53* ( Et t 1 * t ) 0.001( yt yt ) n n yt 0.96 * yt 1 y n ,t * t 0.492 * Et * t 1 0.498 * * t 1 0.01 t * ,t
Slide 25: Parameter
Parameter CRLELN Estimate GMM Std Error Max 0.611 0.064 3.156 4.294 0).723 1.525 0.001 01.958 0.877 01.866 1.249 0.671 2.177 1.380 0.730 Cross-Model Bootstrap Average Std Error Min 0572 (0038) 0.5 0.075 Ode 0.041 3.205 2.727 4.780 3.469 01.799 0.705 2.381 1.525 0.013 0.001 01.945 0.910 01,924 0.810
Slide 26: Implied Parameter Estimates
1.685 Implied = CR,EIN Std Error Cross-Model Parameter Estimate GMM Bootstrap Average Std Error Min Max 0.424 (0.026) 0420 (0013) O0d0l 040 0.134 0.126 (0.023) 0.085 0.160 0.500 0,395 (0.229) 0.492 0.533 : 0.492 0.377 (0.219) 0.466 0.497 ps 0.008 0.003 (0.004) 0.000 0.009 op) 0.934 (0.006) 0.929 0.9
Slide 27: Macro Findings
Macro Findings Statistically significant Monetary Transmission Mechanism: Phillips curve parameter and elasticity of output to real rate significant: T-S Information is doing the job! Both Forward and Backward-Looking Components significant. Unobservables quite persistent What do the unobservables l
Slide 28: Figure 2: Output Gap and Natural Rate of Output
Figure 2: Output Gap and Natural Rate of Output OUTPUT GAP Percent
Slide 29: Figure 3: Inflation Target and Inflation
Percent Figure 3: Inflation Target and Inflation T AA T T 20
Slide 30: Figure 4: Impulse Response Functions of Macro Variables
Figure 4: Impulse Response Functions of Macro Variables ‘AS is be ¥ = Quarters
Slide 31: Figure 5: Variance Decompositions for the Macro Variables
F igure 5: Variance Decompositions for the Macro Variables . > on 2 BTR IRIO LLCS OB OSESESESSESESESESESS i * * 7 eT eS SS SS SS Se 0 5 10 15 20 25 30 35 40 69005 9 0-0 [elroy . o- “ S ©9990000009000000000000000004 ® ¥ oat 9 090683 an ee a A a ee ie ae Oy °900000000000000000000004:% * oa 10 15 20 25
Slide 32: Term Structure Findings
Term Structure Findings 5 Macro factors match the Yield Fairly well Interpretable Yield Movement Factors are not orthogonal Can be interpreted in terms of Structural Shocks b 'n b 'n b 'n i n ,t xt xt 1 t n n n Monetary Policy and Inflation Target shocks are the major driving forces of the T-S. Pred
Slide 33: Figure 6: Fit for Yields
Figure 6: Fit for Yields 16 16 — DATA — - MODEL 14 14 2 2 Q1-61 Q1-80 Q4-03 a1-61 Q1-280 Q4-03 — DATA wah = - MODEL 14 Yq 2 Q1-61 Q1-80 Q4-03 a1-61 Q1-80 Q4-03
Slide 34: Figure 8: Variance Decompositions for the Term Structure Factors
Level Slope Curvature Figure 8: Variance Decompositions for the Term Structure Factors T T T T T T T * *. r aereerer Pero eee eee) a r 0000600600-06000000006000600000000000 fhe lacliliaaaidindtndaindndiadintatialitnattaliads | T T T T T T T — AS Le0S90006600600600000000000000006060¢ > MP - yy = ‘* 1
Slide 35: Figure 9: Contemporaneous Responses of Yields to Macro Shocks
Figure 9: Contemporaneous Responses of Yields to Macro Shocks o 10 20 30 40
Slide 36: Main Results
Main Results Phillips Curve and Interest Rate Demand Channel both Large and Statistically Significant Strong and Significant Monetary Policy Transmission Mechanism to Output and Inflation Monetary Factors Drive the Yield Curve Level is driven by inflation target shocks Slope and Curvature driven by
Slide 37: Extensions
Extensions Include the permanent component of output in the Macro System: Assume that Output and Natural Rate are Cointegrated Generate Time-Varying Risk Premia Square Root Processes Devise Theoretical Model with endogenous Risk Premia Think about regime shifts/structural breaks 37
Banque Nationale De Belgique — Technical Presentation (April 2005) on 6ix