Banque Nationale De Belgique Sa · EURONEXT:BNB
Banque Nationale De Belgique Sa — National Bank of Belgium — 2018 NBB International Conference on Understanding inflation dynamics
30 slides from Banque Nationale De Belgique Sa's conference presentation, Oct 25, 2018
Slide 1: Is euro area lowflation here to stay?
Is euro area lowflation here to stay? Insights from a time-varying parameter model with survey data * 2018 NBB International Conference on “Understanding inflation dynamics” Brussels, October 25-26, 2018 Arnoud Stevens and Joris Wauters * The views expressed in this presentation are those of the aut
Slide 2: Do we still understand inflation dynamics?
Do we still understand inflation dynamics? “The root of the current insecurity around monetary policy is that in advanced economies — from Japan to the US — inflation is not behaving in the way economic models predicted” (FT - Oct 11th, 2017) Mario Draghi: (Sept 2017 policy meeting) “the ongoing eco
Slide 3: Motivation: euro area inflation is undershooting its target
Motivation: euro area inflation is undershooting its target ➢ ECB target: inflation below, but close to, 2% in the medium term ➢ Persistent low inflation since 2013, despite the economic recovery HICP inflation (annual percentage change) 5 4 3 2 1 0 -1 1999 2001 2003 2005 2007 2009 2011 2013 2015 20
Slide 4: What explains this so-called lowflation period?
What explains this so-called lowflation period? Permanent effects… …or cyclical effects ? HICP inflation HICP inflation (annual percentage change) (annual percentage change) 5 5 4 4 3 3 2 2 1 1 0 0 -1 -1 2017 2011 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2018 2007 2008 2009 2010 2012 2013 2
Slide 5: This paper
This paper ➢ Trend vs cycle: what do survey inflation expectations tell? Survey forecasts typically… …(a) incorporate new information regarding structural changes …(b) perform well in forecast comparisons (e.g., Ang et al, 2007, and Faust and Wright, 2013) ➢ To this end: a) Estimate a time-varying p
Slide 6: Relation to the literature
Relation to the literature 1. Literature on understanding euro area lowflation ➢ Only few studies analyse the role of survey forecasts; the ones that do, focus on long-term inflation expectations only (e.g., Garcia and Poon, 2018; Banbura and van Vlodrop, 2018) The short end of the expectations curv
Slide 7: Main Findings
Main Findings ➢ The lowflation is mainly due to cyclical drivers ● The model with survey expectations finds a more muted decline in trend inflation in recent years and a larger degree of economic slack ● The impact of economic slack and import prices on inflation increased in recent years ➢ Includin
Slide 8: Outline
Outline I. Model II. Data and Estimation III. Results 8
Slide 9: Outline
Outline I. Model II. Data and Estimation III. Results 9
Slide 10: Baseline model: time-varying parameter Phillips curve
Baseline model: time-varying parameter Phillips curve ➢ Key ingredients (based on Chan, Koop and Potter, 2016): 𝜋 𝜋𝑡 − 𝜏𝑡𝜋 = 𝜌𝑡𝜋 𝜋𝑡−1 − 𝜏𝑡−1 + 𝜆𝑡 𝑢𝑡 − 𝜏𝑡𝑢 + 𝛾𝑡 𝜋𝑡𝑚 − 𝜏𝑡𝑚 + 𝜖𝑡𝜋 Inflation Unemployment Import price gap gap (slack) inflation gap 𝑢𝑡 − 𝜏𝑡𝑢 = 𝜌1𝑢…
Slide 11: Extended model with survey inflation expectations
Extended model with survey inflation expectations ➢ Measurement equation for the survey forecast for inflation at horizon t+h given the information at the start of period t 𝑒 𝑡−1 𝑒 𝑛 ℎ 𝜋𝑡+ℎ ȁ 𝑛 𝑡 = 1 − 𝜉𝑡−1 𝑓ℎ𝑛 𝜃𝑡−1 , 𝑌 + 𝜉𝑡−1 𝜋 ȁ 𝑡−1+ℎ𝑛 𝑡−1 + 𝜖 𝑡 Period t’s Model forecast…
Slide 12: Outline
Outline I. Model II. Data and Estimation III. Results 12
Slide 13: Data details
Data details ➢ Sample: 1990Q1 - 2017Q4 ➢ Macroeconomic data: • 𝜋𝑡 : HICP inflation (annualised percentage quarterly changes) • 𝑢𝑡 : civilian unemployment rate • 𝜋𝑡𝑚 : annualized quarterly change in the import-price relative to the GDP deflator (Matheson and Stavrev, 2013) ➢ Inflation…
Slide 14: Estimation
Estimation ➢ All latent states (i.e., time varying trends and coefficients ) and static coefficients are jointly estimated … ➢ …applying Bayesian Gibbs sampling techniques ➢ Issue: the model forecast function 𝑓ℎ𝑛 𝜃𝑡−1 , 𝑌 𝑡−1 is nonlinear in the parameters 𝜃𝑡−1 E.g., consider the simplified…
Slide 15: Outline
Outline I. Model II. Data and Estimation III. Results 15
Slide 16: Permanent or cyclical forces? Using survey data leads to different conclusions
Permanent or cyclical forces? Using survey data leads to different conclusions 𝒖 Trend inflation (𝝉𝝅 𝒕) Natural rate of unemployment (𝝉𝒕 ) 6 13 5 12 4 11 3 2 10 1 9 0 8 -1 7 -2 -3 6 1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015 baseline SPF model Inflation baseline SPF model…
Slide 17: Important changes in the transmission of cyclical factors
Important changes in the transmission of cyclical factors: Intrinsic persistence (𝝆𝝅𝒕 ) Phillips curve slope (−𝝀𝒕 ) Import price coefficient (𝜸𝒕 ) 1 1 0.5 0.8 0.8 0.4 0.6 0.6 0.3 0.4 0.4 0.2 0.2 0.2 0.1 0 0 0 1999 2004 2009 2014 1999 2004 2009 2014 1999 2004 2009 2014 baseline SPF model ➢…
Slide 18: Forecasters updated their projections more frequently during the financial crisis period
Forecasters updated their projections more frequently during the financial crisis period Forecast smoothing 𝜉𝑡 1 0.8 0.6 0.4 0.2 0 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 18
Slide 19: SPF improves the forecast accuracy of our model in predicting future inflation
SPF improves the forecast accuracy of our model in predicting future inflation… RMSE relative to baseline model CLPD relative to baseline model 1Q ahead 1Y ahead 2Y ahead 5Y ahead 1Q ahead 1Y ahead 2Y ahead 5Y ahead QoQ inflation YoY inflation YoY inflation YoY inflation QoQ inflation YoY inflation
Slide 20: Low inflation mainly due to cyclical factors
… which provides confidence that the low inflation is mainly due to cyclical factors Gradual declining slack with increasing Phillips curve slope Downward import price pressures Historical decomposition of inflation with the SPF smoothing model Relative contribution to (in %, yoy, mean contribution)
Slide 21: Relative forecast performance (CLPD) over time
Relative forecast performance (CLPD) over time (relative to baseline) 1Q ahead QoQ inflation 1Y ahead YoY inflation 10 20 8 15 6 4 10 2 5 0 0 -2 -4 -5 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 2Y ahead YoY inflation 5Y ahead YoY inflation 14
Slide 22: Robustness: Relevance short-term forecast data & TVP?
Robustness: Relevance short-term forecast data & TVP ? Two robustness checks: a) Long-term survey forecasts only b) Constant parameters ➢ These two alternative model variants attribute less weight to domestic slack in explaining the lowflation ▪ Excluding short-term forecast data: stronger weakening
Slide 23: Conclusions
Conclusions ➢ The model with survey expectations finds a more muted decline in trend inflation in recent years and a larger degree of economic slack ➢ The inclusion of survey data improves the forecast performance of the model in predicting inflation From the perspective of a practitioner, intereste
Slide 24: Thank you for your attention
Thank you for your attention - Questions? 24
Slide 25: Backup slides
Backup slides 25
Slide 26: SPF no smoothing model: Trend inflation and the natural rate of unemployment
SPF no smoothing model: Trend inflation and the natural rate of unemployment 𝒖 Trend inflation (𝝉𝝅 𝒕) Natural rate of unemployment (𝝉𝒕 ) 6 13 5 12 4 11 3 2 10 1 9 0 8 -1 -2 7 -3 6 1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015 baseline baseline SPF no smoothing SPF no smoothing…
Slide 27: SPF no smoothing model: Time varying Phillips curve parameters
SPF no smoothing model: Time varying Phillips curve parameters Intrinsic persistence (𝝆𝝅𝒕 ) Phillips curve slope (−𝝀𝒕 ) Import price coefficient (𝜸𝒕 ) 1 1 0.5 0.8 0.8 0.4 0.6 0.6 0.3 0.4 0.4 0.2 0.2 0.2 0.1 0 0 0 1999 2004 2009 2014 1999 2004 2009 2014 1990 1995 2000 2005 2010 2015 baseline…
Slide 28: Time varying stochastic volatility
Time varying stochastic volatility 1.2 1 0.8 0.6 0.4 0.2 0 1990 1995 2000 2005 2010 2015 Baseline model SPF model SPF no smoothing 28
Slide 29: Robustness checks I: Trend inflation and the natural rate of unemployment
Robustness checks I: Trend inflation and the natural rate of unemployment Trend inflation (𝝉𝝅 𝒕) Natural rate of unemployment (𝝉𝒖𝒕 ) 4 13 3.5 12 3 11 2.5 10 2 9 1.5 8 1 7 0.5 0 6 1990 1995 2000 2005 2010 2015 1990 1995 2000 2005 2010 2015 Baseline SPF model Baseline SPF model LT-exp model CST…
Slide 30: Robustness checks II: forecast performance
Robustness checks II: forecast performance RMSE relative to baseline model CLPD relative to baseline model 1Q ahead 1Y ahead 2Y ahead 5Y ahead 1Q ahead 1Y ahead 2Y ahead 5Y ahead QoQ inflation YoY inflation YoY inflation YoY inflation QoQ inflation YoY inflation YoY inflation YoY inflation 1.2 20 1.