Banque Nationale De Belgique Sa · EURONEXT:BNB
Banque Nationale De Belgique Sa — Technical Presentation (October 2008)
31 slides from Banque Nationale De Belgique Sa's technical presentation, October 2008
Slide 1: Imperfect Information, Macroeconomic Dynamics and the Term Structure of Interest Rates
Imperfect Information, Macroeconomic Dynamics and the Term Structure of Interest Rates: An Encompassing Macro-Finance Model Hans Dewachter KULeuven and RSM, EUR October 2008 NBB Colloquium (KULeuven and RSM, EUR) An Encompassing Macro-Finance Model October 2008 1 / 29
Slide 2: Actual Law of Motion: Summary
Additional slides Actual Law of Motion The Macro-Finance framework Actual Law of Motion: Summary Actual Law of Motion of observed macroeconomic dynamics follows the Evans and Honkapohja (2001) Euler approach to learning and is consistent with : A (semi) structural New-Keynesian model containing: Lon
Slide 3: Actual Law of Motion: Macroeconomic dynamics
Additional slides Actual Law of Motion The Macro-Finance framework Actual Law of Motion: Macroeconomic dynamics Actual Law of Motion of observed macroeconomic dynamics is a VAR(I): Xt = CA + A Xt 1 + A A A S "t ; Xt0 = [ t ; yt ; it ; t P ; Pt ; t ; t] New-Keynesian model with stoch. trends t and t:
Slide 4: Actual Law of Motion: Macroeconomic dynamics
Additional slides Actual Law of Motion The Macro-Finance framework Actual Law of Motion: Macroeconomic dynamics Actual Law of Motion of observed macroeconomic dynamics is a VAR(I): Xt = CA + A Xt 1 + A A A S "t ; Xt0 = [ t ; yt ; it ; t P ; Pt ; t ; t] New-Keynesian model with stoch. trends t and t:
Slide 5: Actual Law of Motion: Macroeconomic dynamics
Additional slides Actual Law of Motion The Macro-Finance framework Actual Law of Motion: Macroeconomic dynamics Actual Law of Motion of observed macroeconomic dynamics is a VAR(I): Xt = CA + A Xt 1 + A A A S "t ; Xt0 = [ t ; yt ; it ; t P ; Pt ; t ; t] New-Keynesian model with stoch. trends t and t:
Slide 6: Actual Law of Motion: Learning dynamics
Additional slides Actual Law of Motion The Macro-Finance framework Actual Law of Motion: Learning dynamics Learning dynamics (imperfect information models) are modeled along the lines of Kozicki and Tinsley (2005): P P t = t 1+w " ;t + (1 w ) b ;t + g ( t EtP 1 t ) P P t = t 1+w " ;t + (1 w ) b ;t +
Slide 7: Actual Law of Motion: Learning dynamics
Additional slides Actual Law of Motion The Macro-Finance framework Actual Law of Motion: Learning dynamics The encompassing model, although ad hoc, embeds standard expectations formation processes as limiting cases: P P t = t 1+w " ;t + (1 w ) b ;t + g ( t EtP 1 t ) P P t = t 1+w " ;t + (1 w ) b ;t
Slide 8: Summary: Bayesian estimation framework
Additional slides Econometric issues Econometric issues Summary: Bayesian estimation framework Identi cation of the posterior density p( i j Z T )of the parameter vector i : p( i j Z T ) = L(Z T j i )p( i )=p(Z T ) with: L(Z T j i ) : the likelihood of the data, Z T , given i : p( i ) : the prior de
Slide 9: Summary: sets of estimated parameters
Additional slides Econometric issues Econometric issues Summary: sets of estimated parameters We estimate jointly six sets of parameters included in i : Deep parameters related to the structural equations: Phillips curve: in ation indexation, ; in ation sensitivity to output, : IS curve: risk aversi
Slide 10: Summary: Prior distributions
Additional slides Econometric issues Econometric issues Summary: Prior distributions Structural parameters (Distr, (mean, std. dev)): Phillips curve: In ation indexation, ; : Beta (0.7, 0.05); Output sensitivity of in ation, : Normal (0.12, 0.03). IS curve: Risk aversion : Gamma (1.5, 0.34); Habit p
Slide 11: Summary: Likelihood
Additional slides Econometric issues Econometric issues Summary: Likelihood Likelihood based on the prediction error decomposition: Transition equation is based on the ALM dynamics, consistent with the learning dynamics and the structural model. Measurement equation includes macroeconomic variables,
Slide 12: Summary: Bayesian estimation
Additional slides Econometric issues Econometric issues Summary: Bayesian estimation Parameters are estimated by simulation-based Bayesian techniques (Metropolis-Hastings): A rst round optimization based on simulated annealing techniques. Metropolis-Hastings based on normal random walk dynamics. Acc
Slide 13: Summary: Likelihood
Additional slides Econometric issues Econometric issues Summary: Likelihood Transition equation (ALM) models observed macroeconomic dynamics: Xt = C A + A Xt 1 + A A A S "t ; Xt0 = [ t ; yt ; it ; t P ; Pt ; t ; t] Structural model (conditioned on true stochastic endpoints, t and t ): t = EtP t+1 +
Slide 14: Summary: Likelihood
Additional slides Econometric issues Econometric issues Summary: Likelihood The measurement equation incorporates three types of information: macroeconomic variables, Xto , the yield curve, Yt ; and survey data on in ation expectations, St : 2 3 2 3 2 3 2 3 2 3 Xto 0 Io 0 0 4 Yt 5 = 4 Ay 5 + 4 By 5
Slide 15: Versions of the model
Additional slides Econometric issues Econometric issues Versions of the model Table: P ROPERTIES OF ALTERNATIVE VERSIONS OF THE MODEL Model Macro (# stoch.trends) Prices of Risk Expectations Mispricing IS NK0 NK model (0) Consistent 0 Full-info RE Yes Stand. NK mod IS MF1 NK model (1) Consistent 0 F
Slide 16: Data
Additional slides Econometric issues Econometric issues Data Model is estimated on US data: 1960Q2 till 2006Q4. Quarter-by-quarter in ation (GDP de ator) rate (p.a. terms) are used. Source: Federal Reserve Economic Data archive (FRED). CBO output gap measure is used (no- real time data). Source: Con
Slide 17: Yield curve fit: posterior moments
Additional slides Results Results for the encompassing model Yield curve t: posterior moments Table: P OSTERIOR DENSITY ESTIMATES I: E NCOMPASSING MODEL MFL A Param. Mean Std. Dev Mode Crit.val. 5% Crit. val. 95% Average mispricing yields (1/2) -0.0034 0.0016 -0.0032 -0.0054 -0.0001 (1) -0.0001 0.00
Slide 18: Risk premiums: expected excess holding returns
Additional slides Results Macro factors and the yield curve Risk premiums: expected excess holding returns Conclusion 6: The encompassing model implies signi cant and time-varying risk premiums, covarying with the output and interest rate gaps. The encompassing Macro-Finance model explains unconditi
Slide 19: Risk premiums: expected excess holding returns
Additional slides Results Macro factors and the yield curve Risk premiums: expected excess holding returns Risk premium maturity 1 year Risk premium maturity 3 years 0.025 0.05 0.02 0.04 0.015 0.03 0.01 0.02 0.005 0.01 0 0 -0.005 -0.01 -0.01 -0.02 -0.015 -0.03 1960 1965 1970 1975 1980 1985 1990 1995
Slide 20: Risk premiums: expected excess holding returns
Additional slides Results Macro factors and the yield curve Risk premiums: expected excess holding returns Table: P OSTERIOR DENSITY ESTIMATES III: E NCOMPASSING MODEL MFL Param Mean Std. Dev Mode Crit.val. 5% Crit. val. 95% 2 Price of risk: 0( 10 ) 0; -0.0700 0.1379 -0.1257 -0.3218 0.1365 0;y -0.06
Slide 21: Variance decomposition: level, slope and curvature factors
Additional slides Results Macro factors and the yield curve Variance decomposition: Interpretation of the yield curve: level, slope and curvature factors Conclusion 7: A variance decomposition of the level, slope and curvature factor identi es the following interpretations: Level factor. Unlike stan
Slide 22: Variance decomposition: Encompassing model
Additional slides Results Macro factors and the yield curve Variance decomposition: Interpretation of the yield curve: level, slope and curvature factors Table: VARIANCE DECOMPOSTION : E NCOMPASSING MODEL MFL Type of shock Fed fund rate Level Slope Curvature In exp 1y In exp 10y Frequency: 1 quarter
Slide 23: Instantaneous impulse-response analysis
Additional slides Results Macro factors and the yield curve Variance decomposition: Instantaneous impulse-response analysis (deviation from baseline) Figure: I NSTANTANEOUS IMPULSE RESPONSE FUNCTIONS OF THE YIELD CURVE MFL MODEL x 10 -3 Yield curve response supply shock x 10 -3 Yield curve response
Slide 24: Inflation scares
Additional slides In ation Scares In ation scares In ation scares NBB Colloquium (KULeuven and RSM, EUR) An Encompassing Macro-Finance Model October 2008 22 / 29
Slide 25: Historical decomposition of the Great Inflation and Disinflation episodes
Additional slides In ation Scares In ation scares Historical decomposition of the 'Great In ation and Disin ation' episodes Conclusion 8: The estimation results establish the empirical relevance of the "In ation Scares" argument for both in ation expectations and the yield curve. In ation Scares: "
Slide 26: Historical decomposition of the Great Inflation episode
Additional slides In ation Scares In ation scares Historical decomposition of the 'Great In ation' episode Figure: H ISTORICAL DECOMPOSITION BASED ON THE MFL- MODEL OF THE G REAT I NFLATION PERIOD Short term yield (1 quarter) Long-term yield (5 year) Long-term inflation expectations (10 year) 0.14 0
Slide 27: Historical decomposition of the Great Disinflation episode
Additional slides In ation Scares In ation scares Historical decomposition of the 'Great Disin ation' episode Figure: H ISTORICAL DECOMPOSITION BASED ON THE MFL- MODEL OF THE G REAT D ISINFLATION PERIOD Short term yield (1 quarter) Long-term yield (5 year) Long-term inflation expectations (10 year)
Slide 28: Time variation eq. real rate: Trehan Wu
Additional slides T-variation eq. real rate Time variation eq. real rate Trehan Wu (JME, 2007) NBB Colloquium (KULeuven and RSM, EUR) An Encompassing Macro-Finance Model October 2008 26 / 29
Slide 29: Time variation eq. real rate: Bjorland et al
Additional slides T-variation eq. real rate Time variation eq. real rate Bjorland et al (2006) NBB Colloquium (KULeuven and RSM, EUR) An Encompassing Macro-Finance Model October 2008 27 / 29
Slide 30: Time variation eq. real rate: Bekaert et al
Additional slides T-variation eq. real rate Time variation eq. real rate Bekaert et al (2005) NBB Colloquium (KULeuven and RSM, EUR) An Encompassing Macro-Finance Model October 2008 28 / 29
Slide 31: Time variation eq. real rate: Edge et al
Additional slides T-variation eq. real rate Time variation eq. real rate Edge et al (FED Working Paper) NBB Colloquium (KULeuven and RSM, EUR) An Encompassing Macro-Finance Model October 2008 29 / 29
Banque Nationale De Belgique — Technical Presentation (October 2008) on 6ix