Banque nationale de Belgique SA · EURONEXT:BNB
Banque nationale de Belgique SA — Investor Presentation (January 2014)
82 slides from Banque nationale de Belgique SA's investor presentation, January 2014
Slide 1: Designing a Simple Loss Function for the Fed: Does the Dual Mandate Make Sense?
Designing a Simple Loss Function for the Fed: Does the Dual Mandate Make Sense? Davide Debortoli, Jinill Kim, Jesper Lindé and Ricardo Nunes UCSD, KU & Federal Reserve Board January, 2014 DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 1 / 27
Slide 2: Motivation
Motivation Conduct of monetary policy delegated to central banks Variable and high rates of price in‡ation in the 1970s and 80s caused many economies to delegate the conduct of monetary policy to instrument independent central banks DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Man
Slide 3: Motivation
Motivation Conduct of monetary policy delegated to central banks Variable and high rates of price in‡ation in the 1970s and 80s caused many economies to delegate the conduct of monetary policy to instrument independent central banks Advances in academic research Rogo¤ (1985) and Persson and Tabellin
Slide 4: Motivation
Motivation Conduct of monetary policy delegated to central banks Variable and high rates of price in‡ation in the 1970s and 80s caused many economies to delegate the conduct of monetary policy to instrument independent central banks Advances in academic research Rogo¤ (1985) and Persson and Tabellin
Slide 5: Motivation
Motivation Fed puts large weight on economic activity One exception to common central banking practice is the Fed’s “Dual Mandate” which stipulates it to “promote maximum employment in a context of price stability” DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 201
Slide 6: Motivation
Motivation Fed puts large weight on economic activity One exception to common central banking practice is the Fed’s “Dual Mandate” which stipulates it to “promote maximum employment in a context of price stability” In January 2012, the Fed adopted an explicit in‡ation target and clari…ed its intenti
Slide 7: Motivation
Motivation Fed puts large weight on economic activity One exception to common central banking practice is the Fed’s “Dual Mandate” which stipulates it to “promote maximum employment in a context of price stability” In January 2012, the Fed adopted an explicit in‡ation target and clari…ed its intenti
Slide 8: Motivation
Motivation Fed puts large weight on economic activity One exception to common central banking practice is the Fed’s “Dual Mandate” which stipulates it to “promote maximum employment in a context of price stability” In January 2012, the Fed adopted an explicit in‡ation target and clari…ed its intenti
Slide 9: What we do
What we do Study the design of simple mandates (loss functions) within the context of an estimated medium-scale model of the U.S. economy – the workhorse Smets and Wouters (2007) model DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 4 / 27
Slide 10: What we do
What we do Study the design of simple mandates (loss functions) within the context of an estimated medium-scale model of the U.S. economy – the workhorse Smets and Wouters (2007) model Speci…cally, we examine how to design simple objectives to maximize welfare of households DKLN (UCSD, KU & Federal
Slide 11: What we do
What we do Study the design of simple mandates (loss functions) within the context of an estimated medium-scale model of the U.S. economy – the workhorse Smets and Wouters (2007) model Speci…cally, we examine how to design simple objectives to maximize welfare of households For instance, does the Fe
Slide 12: What we do
What we do Study the design of simple mandates (loss functions) within the context of an estimated medium-scale model of the U.S. economy – the workhorse Smets and Wouters (2007) model Speci…cally, we examine how to design simple objectives to maximize welfare of households For instance, does the Fe
Slide 13: What we do
What we do Study the design of simple mandates (loss functions) within the context of an estimated medium-scale model of the U.S. economy – the workhorse Smets and Wouters (2007) model Speci…cally, we examine how to design simple objectives to maximize welfare of households For instance, does the Fe
Slide 14: What we do
What we do Study the design of simple mandates (loss functions) within the context of an estimated medium-scale model of the U.S. economy – the workhorse Smets and Wouters (2007) model Speci…cally, we examine how to design simple objectives to maximize welfare of households For instance, does the Fe
Slide 15: What we do
What we do Study the design of simple mandates (loss functions) within the context of an estimated medium-scale model of the U.S. economy – the workhorse Smets and Wouters (2007) model Speci…cally, we examine how to design simple objectives to maximize welfare of households For instance, does the Fe
Slide 16: Key findings
Key …ndings A large weight on resource utilization is warranted DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 5 / 27
Slide 17: Key findings
Key …ndings A large weight on resource utilization is warranted For a standard in‡ation - output gap loss function, we …nd “lambda” about 1 DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 5 / 27
Slide 18: Key findings
Key …ndings A large weight on resource utilization is warranted For a standard in‡ation - output gap loss function, we …nd “lambda” about 1 More ad hoc utilization measures – like detrended output or output growth – also get a large weight DKLN (UCSD, KU & Federal Reserve Board) On the Design of Sim
Slide 19: Key findings
Key …ndings A large weight on resource utilization is warranted For a standard in‡ation - output gap loss function, we …nd “lambda” about 1 More ad hoc utilization measures – like detrended output or output growth – also get a large weight Note: Absolute welfare gains/losses small (Lucas, 1987, and
Slide 20: Key findings
Key …ndings A large weight on resource utilization is warranted For a standard in‡ation - output gap loss function, we …nd “lambda” about 1 More ad hoc utilization measures – like detrended output or output growth – also get a large weight Note: Absolute welfare gains/losses small (Lucas, 1987, and
Slide 21: Key findings
Key …ndings A large weight on resource utilization is warranted For a standard in‡ation - output gap loss function, we …nd “lambda” about 1 More ad hoc utilization measures – like detrended output or output growth – also get a large weight Note: Absolute welfare gains/losses small (Lucas, 1987, and
Slide 22: Key findings
Key …ndings A large weight on resource utilization is warranted For a standard in‡ation - output gap loss function, we …nd “lambda” about 1 More ad hoc utilization measures – like detrended output or output growth – also get a large weight Note: Absolute welfare gains/losses small (Lucas, 1987, and
Slide 23: Presentation outline
Presentation outline Model and parameterization Our exercise Benchmark results Robustness of results Simple rule results Concluding remarks DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 6 / 27
Slide 24: Model
Model Key features of model structure We use the estimated SW07 model. This model features monopolistic competition in both goods and labour markets Nominal price and wage stickiness: Calvo price contracts, indexation of non-optimizers PtNO = Πtp 1 Π1 ιp PtNO1 ι Calvo wage contracts, indexation of n
Slide 25: Model: Shock structure
Model Shock structure Four structural shocks that we assume a¤ect potential output: Total factor productivity (εat ), Investment-speci…c (εit ), Risk-shock on …nancial assets (εbt ), Government-NX (εgt ), Two “ine¢ cient” shocks: εpt - “price markup” shock εwt - “wage markup” shock Pay particular at
Slide 26: Parameterization
Parameterization Parameters adopted from Smets and Wouters We use the posterior mode parameters from SW07 (Tables 1.A-B in their paper, Table 1 in our paper) Make assumptions on adjustment functions and how we introduce the shocks so that linearized representation of our model coincides exactly with
Slide 27: Our Exercise
Our Exercise Quadratic approximation of utility and Ramsey policy Benigno and Woodford (2006) demonstrated that households utility function could be written as: ∞ ∞ ∑ E0 βt U (Xt ) ' constant ∑ E0 βt Xt0 W society Xt , (1) t =0 t =0 DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Man
Slide 28: Our Exercise
Our Exercise Quadratic approximation of utility and Ramsey policy Benigno and Woodford (2006) demonstrated that households utility function could be written as: ∞ ∞ ∑ E0 βt U (Xt ) ' constant ∑ E0 βt Xt0 W society Xt , (1) t =0 t =0 Xt0 W society Xt on the RHS is LQ approximation of the economy DKLN
Slide 29: Our Exercise
Our Exercise Quadratic approximation of utility and Ramsey policy Benigno and Woodford (2006) demonstrated that households utility function could be written as: ∞ ∞ ∑ E0 βt U (Xt ) ' constant ∑ E0 βt Xt0 W society Xt , (1) t =0 t =0 Xt0 W society Xt on the RHS is LQ approximation of the economy De…n
Slide 30: Our Exercise
Our Exercise Quadratic approximation of utility and Ramsey policy Benigno and Woodford (2006) demonstrated that households utility function could be written as: ∞ ∞ ∑ E0 βt U (Xt ) ' constant ∑ E0 βt Xt0 W society Xt , (1) t =0 t =0 Xt0 W society Xt on the RHS is LQ approximation of the economy De…n
Slide 31: Our Exercise
Our Exercise Quadratic approximation of utility and Ramsey policy Benigno and Woodford (2006) demonstrated that households utility function could be written as: ∞ ∞ ∑ E0 βt U (Xt ) ' constant ∑ E0 βt Xt0 W society Xt , (1) t =0 t =0 Xt0 W society Xt on the RHS is LQ approximation of the economy De…n
Slide 32: Our Exercise
Our Exercise Simple Mandate approximation to policy behavior We assume (arguably realistically) that the CB minimizes: E0 ∑ βt Xt0 W CB Xt , DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 11 / 27
Slide 33: Our Exercise
Our Exercise Simple Mandate approximation to policy behavior We assume (arguably realistically) that the CB minimizes: E0 ∑ βt Xt0 W CB Xt , W CB is a sparse matrix with many zeros DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 11 / 27
Slide 34: Our Exercise
Our Exercise Simple Mandate approximation to policy behavior We assume (arguably realistically) that the CB minimizes: E0 ∑ βt Xt0 W CB Xt , W CB is a sparse matrix with many zeros Given W CB , the expected loss for the society is 0 Loss obj = E Xtobj W CB W society Xtobj W CB . (2) DKLN (UCSD, KU &
Slide 35: Our Exercise
Our Exercise Simple Mandate approximation to policy behavior We assume (arguably realistically) that the CB minimizes: E0 ∑ βt Xt0 W CB Xt , W CB is a sparse matrix with many zeros Given W CB , the expected loss for the society is 0 Loss obj = E Xtobj W CB W society Xtobj W CB . (2) Measure welfare
Slide 36: Our Exercise
Our Exercise Simple Mandate approximation to policy behavior We assume (arguably realistically) that the CB minimizes: E0 ∑ βt Xt0 W CB Xt , W CB is a sparse matrix with many zeros Given W CB , the expected loss for the society is 0 Loss obj = E Xtobj W CB W society Xtobj W CB . (2) Measure welfare
Slide 37: Our Exercise
Our Exercise Two alternative assumptions about mandate We will explore two alternative assumptions about the speci…city of the law that governs the behavior of the central bank DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 12 / 27
Slide 38: Our Exercise
Our Exercise Two alternative assumptions about mandate We will explore two alternative assumptions about the speci…city of the law that governs the behavior of the central bank 1 We assume that the law only speci…es both the variables and weights. For example, the law could specify Lt = (π at π a )2
Slide 39: Our Exercise
Our Exercise Two alternative assumptions about mandate We will explore two alternative assumptions about the speci…city of the law that governs the behavior of the central bank 1 We assume that the law only speci…es both the variables and weights. For example, the law could specify Lt = (π at π a )2
Slide 40: Benchmark results
Benchmark results Results for standard in‡ation-output mandate We start with a standard in‡ation - output based function Lt = (π at π a )2 + λa xt2 DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 13 / 27
Slide 41: Benchmark results
Benchmark results Results for standard in‡ation-output mandate We start with a standard in‡ation - output based function Lt = (π at π a )2 + λa xt2 π at is annualized in‡ation (4 ln [Pt /Pt 1 ]) DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 13 / 27
Slide 42: Benchmark results
Benchmark results Results for standard in‡ation-output mandate We start with a standard in‡ation - output based function Lt = (π at π a )2 + λa xt2 π at is annualized in‡ation (4 ln [Pt /Pt 1 ]) Consider three alternative measures of xt : yt ytpot , yt ȳt and 4 ( yt yt 1 ) DKLN (UCSD, KU & Federal
Slide 43: Benchmark results
Benchmark results Results for standard in‡ation-output mandate We start with a standard in‡ation - output based function Lt = (π at π a )2 + λa xt2 π at is annualized in‡ation (4 ln [Pt /Pt 1 ]) Consider three alternative measures of xt : yt ytpot , yt ȳt and 4 ( yt yt 1 ) CEV as function for λa fo
Slide 44: Benchmark results
Benchmark results CEV for simple mandates with alternative utilization measures DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 14 / 27
Slide 45: Benchmark results
Benchmark results Volatility trade-o¤s for alternative utilization measures DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 15 / 27
Slide 46: Benchmark results
Benchmark results Drivers of our results Key …ndings: DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 16 / 27
Slide 47: Benchmark results
Benchmark results Drivers of our results Key …ndings: 1 Optimal weight on resource utilization is about 1.05. This is substantially higher than Woodford’s (2003) value of 0.048 and Yellen’s (2012) value of 0.25 DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 16
Slide 48: Benchmark results
Benchmark results Drivers of our results Key …ndings: 1 Optimal weight on resource utilization is about 1.05. This is substantially higher than Woodford’s (2003) value of 0.048 and Yellen’s (2012) value of 0.25 2 Important volatility trade-o¤ between in‡ation and the output gap (at odds with Justini
Slide 49: Benchmark results
Benchmark results Drivers of our results Key …ndings: 1 Optimal weight on resource utilization is about 1.05. This is substantially higher than Woodford’s (2003) value of 0.048 and Yellen’s (2012) value of 0.25 2 Important volatility trade-o¤ between in‡ation and the output gap (at odds with Justini
Slide 50: Benchmark results
Benchmark results Drivers of our results Key …ndings: 1 Optimal weight on resource utilization is about 1.05. This is substantially higher than Woodford’s (2003) value of 0.048 and Yellen’s (2012) value of 0.25 2 Important volatility trade-o¤ between in‡ation and the output gap (at odds with Justini
Slide 51: Benchmark results
Benchmark results Drivers of our results Key …ndings: 1 Optimal weight on resource utilization is about 1.05. This is substantially higher than Woodford’s (2003) value of 0.048 and Yellen’s (2012) value of 0.25 2 Important volatility trade-o¤ between in‡ation and the output gap (at odds with Justini
Slide 52: Benchmark results
Benchmark results Drivers of our results Key …ndings: 1 Optimal weight on resource utilization is about 1.05. This is substantially higher than Woodford’s (2003) value of 0.048 and Yellen’s (2012) value of 0.25 2 Important volatility trade-o¤ between in‡ation and the output gap (at odds with Justini
Slide 53: Benchmark results
Benchmark results Drivers of our results To understand the role of the parameters, we compute λa using Woodford’s (2003) formula 16 κθ px in a variant of the model without capital and wage stickiness, and …nd that λa = 0.87 DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates Jan
Slide 54: Benchmark results
Benchmark results Drivers of our results To understand the role of the parameters, we compute λa using Woodford’s (2003) formula 16 κθ px in a variant of the model without capital and wage stickiness, and …nd that λa = 0.87 This is higher than Woodford’s λa as he assumed …rm-speci…c labor (which c.p
Slide 55: Benchmark results
Benchmark results Drivers of our results To understand the role of the parameters, we compute λa using Woodford’s (2003) formula 16 κθ px in a variant of the model without capital and wage stickiness, and …nd that λa = 0.87 This is higher than Woodford’s λa as he assumed …rm-speci…c labor (which c.p
Slide 56: Benchmark results
Benchmark results Drivers of our results To understand the role of the parameters, we compute λa using Woodford’s (2003) formula 16 κθ px in a variant of the model without capital and wage stickiness, and …nd that λa = 0.87 This is higher than Woodford’s λa as he assumed …rm-speci…c labor (which c.p
Slide 57: Benchmark results
Benchmark results Drivers of our results To understand the role of the parameters, we compute λa using Woodford’s (2003) formula 16 κθ px in a variant of the model without capital and wage stickiness, and …nd that λa = 0.87 This is higher than Woodford’s λa as he assumed …rm-speci…c labor (which c.p
Slide 58: Benchmark results
Benchmark results Drivers of our results To understand the role of the parameters, we compute λa using Woodford’s (2003) formula 16 κθ px in a variant of the model without capital and wage stickiness, and …nd that λa = 0.87 This is higher than Woodford’s λa as he assumed …rm-speci…c labor (which c.p
Slide 59: Benchmark results
Benchmark results Sensitivity of results w.r.t. parameters and shocks DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 18 / 27
Slide 60: Benchmark results
Benchmark results Drivers of our results We now turn to the second question, namely why we get an important trade-o¤ between stabilizing the output gap and in‡ation DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 19 / 27
Slide 61: Benchmark results
Benchmark results Drivers of our results We now turn to the second question, namely why we get an important trade-o¤ between stabilizing the output gap and in‡ation Justiniano, Primiceri and Tambalotti (2012) …nds that the output gap can be stabilized without generating higher in‡ation volatility DK
Slide 62: Benchmark results
Benchmark results Drivers of our results We now turn to the second question, namely why we get an important trade-o¤ between stabilizing the output gap and in‡ation Justiniano, Primiceri and Tambalotti (2012) …nds that the output gap can be stabilized without generating higher in‡ation volatility Bu
Slide 63: Benchmark results
Benchmark results Drivers of our results We now turn to the second question, namely why we get an important trade-o¤ between stabilizing the output gap and in‡ation Justiniano, Primiceri and Tambalotti (2012) …nds that the output gap can be stabilized without generating higher in‡ation volatility Bu
Slide 64: Benchmark results
Benchmark results Drivers of our results We now turn to the second question, namely why we get an important trade-o¤ between stabilizing the output gap and in‡ation Justiniano, Primiceri and Tambalotti (2012) …nds that the output gap can be stabilized without generating higher in‡ation volatility Bu
Slide 65: Benchmark results
Benchmark results Variance frontiers for alternative calibrations DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 20 / 27
Slide 66: Benchmark results
Benchmark results Drivers of our results While we do not neccessarily disagree with JPT, our analysis makes clear that their “no trade-o¤” result is a special case in the sense that it applies only if BOTH price and wage markup shocks are irrelevant DKLN (UCSD, KU & Federal Reserve Board) On the Des
Slide 67: Robustness of results
Robustness of results Importantly, we …nd that our results hold up when we put restrictions on std(rta ): Results for loss function with interest rate term Loss Function λa ytgap λr CEV (%) std(rta ) Woodford 0.048 0.1381 8.92 Optimized 1.042 0.0128 9.00 Optimized : rta r a 1.161 0.0770 0.0222 2.24
Slide 68: Robustness of results
Robustness of results Also study the merits of an alternative mandate with nominal wage in‡ation and a labor market gap (lt ltpot ): 2 Lt = (∆wta ∆w a )2 + λa lt ltpot DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 23 / 27
Slide 69: Robustness of results
Robustness of results Also study the merits of an alternative mandate with nominal wage in‡ation and a labor market gap (lt ltpot ): 2 Lt = (∆wta ∆w a )2 + λa lt ltpot Find that labor market variables warrant further attention; not surprising given that the model features labor market frictions (nom
Slide 70: Robustness of results
Robustness of results On the importance of labor market variables DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 24 / 27
Slide 71: Simple rule results
Simple rule results Rule and basic results We also study the performance of simple rules on the form h i rta = (1 ρr ) r a + $π (π at π a ) + $y ytgap + $∆y ∆ytgap +ρr rta 1 (4) Look at Taylor (1993) and (1999) and coe¢ cients that minimize CEV Parameterization $π $y $∆y ρr CEV (%) std(rta ) Taylor
Slide 72: Simple rule results
Simple rule results Interpretation of …ndings We …nd that optimized rule is characterized by DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 26 / 27
Slide 73: Simple rule results
Simple rule results Interpretation of …ndings We …nd that optimized rule is characterized by High degree of interest rate smoothing (ρr ) DKLN (UCSD, KU & Federal Reserve Board) On the Design of Simple Mandates January, 2014 26 / 27
Slide 74: Simple rule results
Simple rule results Interpretation of …ndings We …nd that optimized rule is characterized by High degree of interest rate smoothing (ρr ) Large response coe¢ cients for in‡ation ($π ) and growth rate of the output gap ($∆y ); coe¤ on the level of the output gap ($y ) substantially smaller DKLN (UCSD
Slide 75: Simple rule results
Simple rule results Interpretation of …ndings We …nd that optimized rule is characterized by High degree of interest rate smoothing (ρr ) Large response coe¢ cients for in‡ation ($π ) and growth rate of the output gap ($∆y ); coe¤ on the level of the output gap ($y ) substantially smaller CEV for op
Slide 76: Simple rule results
Simple rule results Interpretation of …ndings We …nd that optimized rule is characterized by High degree of interest rate smoothing (ρr ) Large response coe¢ cients for in‡ation ($π ) and growth rate of the output gap ($∆y ); coe¤ on the level of the output gap ($y ) substantially smaller CEV for op
Slide 77: Concluding remarks
Concluding remarks Our analysis suggest that resource utilization should carry a large weight in formulation of monetary policy, consistent with the spirit of the dual mandate and recent papers by Reifschneider et al. (2013) and English et al. (2013) DKLN (UCSD, KU & Federal Reserve Board) On the De
Slide 78: Concluding remarks
Concluding remarks Our analysis suggest that resource utilization should carry a large weight in formulation of monetary policy, consistent with the spirit of the dual mandate and recent papers by Reifschneider et al. (2013) and English et al. (2013) Weight on resource utilization also substantially
Slide 79: Concluding remarks
Concluding remarks Our analysis suggest that resource utilization should carry a large weight in formulation of monetary policy, consistent with the spirit of the dual mandate and recent papers by Reifschneider et al. (2013) and English et al. (2013) Weight on resource utilization also substantially
Slide 80: Concluding remarks
Concluding remarks Our analysis suggest that resource utilization should carry a large weight in formulation of monetary policy, consistent with the spirit of the dual mandate and recent papers by Reifschneider et al. (2013) and English et al. (2013) Weight on resource utilization also substantially
Slide 81: Concluding remarks
Concluding remarks Our analysis suggest that resource utilization should carry a large weight in formulation of monetary policy, consistent with the spirit of the dual mandate and recent papers by Reifschneider et al. (2013) and English et al. (2013) Weight on resource utilization also substantially
Slide 82: Concluding remarks
Concluding remarks Our analysis suggest that resource utilization should carry a large weight in formulation of monetary policy, consistent with the spirit of the dual mandate and recent papers by Reifschneider et al. (2013) and English et al. (2013) Weight on resource utilization also substantially
Banque Nationale De Belgique — Investor Presentation (January 2014) on 6ix