Banque nationale de Belgique SA · EURONEXT:BNB

Banque nationale de Belgique SA — Technical Presentation (December 2014)

55 slides from Banque nationale de Belgique SA's technical presentation, Dec 1, 2014

  1. Slide 1: Collateral Constraints and Macroeconomic Asymmetries

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Collateral Constraints and Macroeconomic Asymmetries Luca Guerrieri Matteo Iacoviello Federal Reserve Board Federal Reserve Board National Bank of Belgium December 1, 2014 The views expressed in this paper a

  2. Slide 2: Consumption, Year-on-Year % Change

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras 1975 1980 1985 1990 1995 2000 2005 2010 Year Consumption, Year-on-Year % Change 4 1983 1998 1999 2 1992 1985 19841997 1988 19862000 1978 1977 1982 19941996 1993 2003 2004 2010 2006 0 1995 1989 2001 1987 2005

  3. Slide 3: This Paper

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras This Paper How much do Housing Boom and Bust contribute to movements in consumption? We address this question with a general equilibrium model estimated with Bayesian methods. In the model, housing collatera

  4. Slide 4: This Paper

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras This Paper How much do Housing Boom and Bust contribute to movements in consumption? We address this question with a general equilibrium model estimated with Bayesian methods. In the model, housing collatera

  5. Slide 5: This Paper

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras This Paper How much do Housing Boom and Bust contribute to movements in consumption? We address this question with a general equilibrium model estimated with Bayesian methods. In the model, housing collatera

  6. Slide 6: Related Papers

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Related Papers DSGE model Financial factors and …nancial shocks and the Great Recession Del Negro et al. (2011), Jermann and Quadrini (2012), Christiano, Motto and Rostagno (2013) These papers are either cal

  7. Slide 7: The Basic Idea

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras The Basic Idea Household maximizes U = E0 ∑∞ t t=0 β (log ct + j log ht ) subject to ct + qt ht = y + bt Rbt 1 + qt ht 1 (1 δ) bt mqt ht log qt = ρ log qt 1 + υt , υt N 0, σ2

  8. Slide 8: The Basic Idea

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras The Basic Idea Household maximizes U = E0 ∑∞ t t=0 β (log ct + j log ht ) subject to ct + qt ht = y + bt Rbt 1 + qt ht 1 (1 δ) bt mqt ht log qt = ρ log qt 1 + υt , υt N 0, σ2 Assume impatience (βR < 1), …x y

  9. Slide 9: The Basic Idea

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras The Basic Idea Household maximizes U = E0 ∑∞ t t=0 β (log ct + j log ht ) subject to ct + qt ht = y + bt Rbt 1 + qt ht 1 (1 δ) bt mqt ht log qt = ρ log qt 1 + υt , υt N 0, σ2 Assume impatience (βR < 1), …x y

  10. Slide 10: Consumption

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Model’s solution. Consumption function, ct = C (qt , bt 1 , ht 1 ) 1 0.95 0.9 Borrowing Consumption C onstraint 0.85 Slack 0.8 0.75 Borrowing C onstraint 0.7 Binds 0.65 0.9 0.95 1 1.05 1.1 H ousing Price

  11. Slide 11: In summary

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras In summary: 1. High house prices are associated with slack borrowing constraints, and with a lower sensitivity of consumption to changes in house prices. 2. When household borrowing is constrained – more lik

  12. Slide 12: The Full Model: Overview

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras The Full Model: Overview Standard monetary DSGE model augmented to include housing collateral constraint along the lines of Kiyotaki and Moore (1997), Iacoviello (2005), and Liu, Wang, and Zha (2013). Allow

  13. Slide 13: Full Model: Households and Preferences

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Full Model: Households and Preferences Two household types, each measure 1: patient and impatient. Within each group, a representative household maximizes: 1 E0 ∑∞ t 1+ η t=0 β zt Γ log ( ct εct 1 ) + jt log

  14. Slide 14: Full Model: Households Constraints

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Full Model: Households Constraints Patient households maximize their utility subject to: R t 1 bt 1 wt nt ct + qt ∆ht + it = + rk,t k t 1 bt + divt , πt xw,t resources are given by wage, capital income, hous

  15. Slide 15: Full Model: Monetary Policy and Supply Side

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Full Model: Monetary Policy and Supply Side Monetary policy follows Taylor rule that responds to annual in‡ation and GDP in deviation from trend, subject to the zero lower bound (ZLB): h i r ( 1 r ) r e ( 1

  16. Slide 16: Two Important Parameters

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Two Important Parameters Parameter Restriction Measures σ b/w 0 and 1 Collateral Constraints β0 less than β Asymmetries σ measures wage share of impatient households and importance of collateral constraints.

  17. Slide 17: Data and Shocks

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Data and Shocks The estimation is based on observations from 1985Q1 to 2011Q4: 1. Real Total Household Consumption, 2. Price (GDP de‡ator) In‡ation, 3. Wage In‡ation (compensation per hour, nonfarm), 4. Real

  18. Slide 18: Data series

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Interest Rate House Prices Consumption 15 20 5 % from HP trend % from HP trend 10 0 0 Percent 5 -20 -5 0 -40 -10 1990 2000 2010 1990 2000 2010 1990 2000 2010 Investment Wage Inflation Price Inflation 20 5 6

  19. Slide 19: Solution Method

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Solution Method We solve the model using the Occbin algorithm (see Guerrieri and Iacoviello, JME): the algorithm extends a …rst-order perturbation approach and applies it in a piecewise fashion to handle DSG

  20. Slide 20: The Local Linearity of the Policy Functions

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras The Local Linearity of the Policy Functions 0.02 c` 0.01 0 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 u j,t 0.06 1 Number of periods the constraint is expected to be slack 0.04 Q( ) 23 4 5 6 7 8 0.02 9 10 12 13

  21. Slide 21: Computing the Likelihood

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Computing the Likelihood Each period, the model solution takes the form: Xt = P ( Xt 1 , e t ) Xt 1 + D ( Xt 1 , e t ) + Q ( Xt 1 , e t ) e t

  22. Slide 22: Computing the Likelihood

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Computing the Likelihood Each period, the model solution takes the form: Xt = P ( Xt 1 , e t ) Xt 1 + D ( Xt 1 , e t ) + Q ( Xt 1 , e t ) e t In terms of observables, through observation equation Yt = HXt ,:

  23. Slide 23: Computing the Likelihood

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Computing the Likelihood Each period, the model solution takes the form: Xt = P ( Xt 1 , e t ) Xt 1 + D ( Xt 1 , e t ) + Q ( Xt 1 , e t ) e t In terms of observables, through observation equation Yt = HXt ,:

  24. Slide 24: Computing the Likelihood

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Computing the Likelihood Each period, the model solution takes the form: Xt = P ( Xt 1 , e t ) Xt 1 + D ( Xt 1 , e t ) + Q ( Xt 1 , e t ) e t In terms of observables, through observation equation Yt = HXt ,:

  25. Slide 25: Computing the Likelihood

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Computing the Likelihood Each period, the model solution takes the form: Xt = P ( Xt 1 , e t ) Xt 1 + D ( Xt 1 , e t ) + Q ( Xt 1 , e t ) e t In terms of observables, through observation equation Yt = HXt ,:

  26. Slide 26: Calibration

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Calibration m Maximum LTV 0.9 η labor disutility 1 β discount factor, patient agents 0.995 π steady-state gross in‡ation rate 1.005 α capital share in production 0.3 δ capital depreciation rate 0.025 j housi

  27. Slide 27: Model Results: Selected Estimated Parameters

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Model Results: Selected Estimated Parameters Prior type [mean, std] Posterior Mode β0 discount factor, impatients normal [0.99, .0015] 0.9895 σ wage share, impatients beta [0.5, 0.20] 0.4151 ε habit in consu

  28. Slide 28: Specification Checks and Sensitivity Analysis

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Speci…cation Checks and Sensitivity Analysis Checks on the Estimation Procedure 1. Re-estimate model assuming di¤erent initialization scheme 2. Filter shocks assuming true parameters are known 3. Estimate sh

  29. Slide 29: Model Results: Effect of housing demand shocks

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Model Results: E¤ect of housing demand shocks House Prices Consumption % from steady state % from steady state 30 4 20 2 10 0 0 -10 -2 -20 -30 -4 0 10 20 30 40 0 10 20 30 40 Hours W orked Multiplier on Borro

  30. Slide 30: Consumption and the Housing Boom and Bust

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Consumption and the Housing Boom and Bust How much did collateral constraints contribute to the decline in consumption? By construction, estimated model explains everything in sample. However, it is importan

  31. Slide 31: The estimated simulated multiplier

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras The estimated simulated multiplier House Price (% from trend) M odel impli ed m ultipl ier (level) 0.08 10 0 0.06 -10 0.04 -20 0.02 -30 0 1990 2000 2010 1990 2000 2010

  32. Slide 32: Consumption and House Prices: Data and Model

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Consumption and House Prices: Data and Model (Data, and all shocks) House Prices (Housing Demand) Consumption (Housing Demand) data 4 10 data 2 0 0 % from trend -10 -2 -20 -4 -30 -6 1990 1995 2000 2005 2010

  33. Slide 33: Consumption and House Prices: Data and Model

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Consumption and House Prices: Data and Model (Housing Demand Shocks Only) House Prices (Housing Demand) Consumption (Housing Demand) data 4 10 data 2 model 0 0 % from trend -10 -2 -20 -4 -30 -6 1990 1995 200

  34. Slide 34: Consumption and House Prices: Data and Model

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Consumption and House Prices: Data and Model (Housing Demand Shocks Only) House Prices (Housing Demand) Consumption (Housing Demand) data 4 10 data 2 model model 0 0 % from trend -10 -2 -20 -4 -30 -6 1990 19

  35. Slide 35: Consumption and House Prices: Data and Model

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Consumption and House Prices: Data and Model (Housing Demand Shocks Only - model w/o frictions) House Prices (Housing Demand) Consumption (Housing Demand) data 4 10 data 2 model model 0 0 % from trend -10 -2

  36. Slide 36: Summary of Comparison

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Summary of Comparison The baseline model comes close to matching the evolution of both housing and consumption with just the housing shocks, By contrast, housing shocks have no bearing on consumption for the

  37. Slide 37: Evidence from Regional Data

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Evidence from Regional Data We use state-level and MSA-level data from 1990 to 2010

  38. Slide 38: Evidence from Regional Data

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Evidence from Regional Data We use state-level and MSA-level data from 1990 to 2010 We regress an indicator of economic activity against housing prices (when low or high)

  39. Slide 39: Evidence from Regional Data

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Evidence from Regional Data We use state-level and MSA-level data from 1990 to 2010 We regress an indicator of economic activity against housing prices (when low or high) Regressions (y is log EMP/CARS/ELE,

  40. Slide 40: Evidence from Regional Data

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Evidence from Regional Data We use state-level and MSA-level data from 1990 to 2010 We regress an indicator of economic activity against housing prices (when low or high) Regressions (y is log EMP/CARS/ELE,

  41. Slide 41: Evidence from Regional Data

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Evidence from Regional Data We use state-level and MSA-level data from 1990 to 2010 We regress an indicator of economic activity against housing prices (when low or high) Regressions (y is log EMP/CARS/ELE,

  42. Slide 42: US States: Auto Sales and House Prices

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras US States: Auto Sales and House Prices % Change in Auto Sales (∆autot ) ∆hpt 1 0.24*** (0.03) ∆hp_hight 1 -0.05 0.16*** 0.11*** 0.07** (0.04) (0.04) (0.03) (0.03) ∆hp_lowt 1 0.62*** 0.33*** 0.27** 0.20** (0.

  43. Slide 43: Instrumenting House Prices, MSA

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Instrumenting House Prices, MSA We instrument housing price using housing supply elasticity at the MSA level (data from Albert Saiz), as in Mian and Su… (2011). Cross-sectional Regressions Sample Sample 2002

  44. Slide 44: Summary of the Empirical Evidence

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Summary of the Empirical Evidence Average elasticities measured from the various regressions State MSA ∆empl ∆auto ∆elec ∆empl ∆hp_high 0.02 0.07 0.12 0.04 ∆hp_low 0.07 0.20 0.19 0.09 1. Average sensitivity

  45. Slide 45: Asymmetry in Motion

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Asymmetry in Motion House Price (% from trend) Consumption (% from trend) Data and Baseline 0 10 -2 0 Small Shock -10 -4 -20 Large Shock -6 -30 2010 2015 2020 2010 2015 2020 Model implied multiplier λ (level

  46. Slide 46: An application to mortgage relief

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras An application to mortgage relief Consider a simple proposal such as mortgage relief for debtors The marginal e¤ects of mortgage relief depend drastically on whether house prices are high (and few people are

  47. Slide 47: Two transfers from saver to borrower under different house price scenarios

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Two transfers from saver to borrower under di¤erent house price scenarios Housing Prices Transfer 20 1 % of S S consumption % dev. from S S 10 0.8 0 0.6 0.4 -10 0.2 -20 0 20 40 60 80 100 20 40 60 80 100 Cons

  48. Slide 48: Deleveraging in SCF

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Deleveraging in SCF Mean V alue of Mortgages for Families with P os itive Mortgages , by Liquidi ty 220 Illiqui d Hous ehol ds Liquid Hous eholds 200 A verage Thousands of 2009 $ 180 160 140 120 100 1985 199

  49. Slide 49: Deleveraging in Aggregate Data and the Model

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Deleveraging in Aggregate Data and the Model Leverage (Total Household Debt over Owner-Occupied Real Estate) 15 Model Data 10 Deviation from mean, ppt 5 0 -5 1990 1995 2000 2005 2010

  50. Slide 50: Does our approach uniquely identify shocks?

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Does our approach uniquely identify shocks? In practice, yes. Note: exclude interest rate and monetary shocks from observables/shocks when at the ZLB. 1 Consumption, % from s.s. 0 -1 -2 -3 -4 -0.8 -0.6 -0.4

  51. Slide 51: The ZLB

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras The ZLB House Price (% from trend) Consumption (% from trend) 10 2 0 0 -2 -10 -4 -20 -6 -30 1990 1995 2000 2005 2010 1990 1995 2000 2005 2010 Interest Rate (APR) Model implied multiplier λ (level) 8 0.08 6 0

  52. Slide 52: Conclusions

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Conclusions A model with housing collateral generates asymmetric e¤ects of house prices on aggregate economic activity These asymmetries can be found in national, US state and MSA level data, using Bayesian

  53. Slide 53: Conclusions

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Conclusions A model with housing collateral generates asymmetric e¤ects of house prices on aggregate economic activity These asymmetries can be found in national, US state and MSA level data, using Bayesian

  54. Slide 54: Conclusions

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Conclusions A model with housing collateral generates asymmetric e¤ects of house prices on aggregate economic activity These asymmetries can be found in national, US state and MSA level data, using Bayesian

  55. Slide 55: Conclusions

    Introduction The Model Solution, Estimation and Results Regional Evidence Conclusions Extras Conclusions A model with housing collateral generates asymmetric e¤ects of house prices on aggregate economic activity These asymmetries can be found in national, US state and MSA level data, using Bayesian

Banque Nationale De Belgique — Technical Presentation (December 2014) on 6ix