The Phillips Curve at the ECB · The Phillips multiplier is not well defined, and therefore is estimated with wide confidence bands at short horizons (bands up to 6 month are excluded

The Phillips Curve at the ECB

50th Anniversary Conference of the Money, Macro & Finance Research Group

London School of Economics

4 September 2019

Philip R. Lane Member of the Executive Board Based on a forthcoming joint paper with F. Eser, P. Karadi, L. Moretti, C. Osbat

Rubric

www.ecb.europa.eu

© 2

1 The ECB’s policy context

2 The structural Phillips Curve

3 Identifying the slope of the structural Phillips Curve

4 Reduced-form evidence

Conclusion 5

Outline

Rubric

www.ecb.europa.eu

© 3

HICP and HICP excluding energy and food (percentage per annum)

Development of inflation and inflation expectations

Sources: Bloomberg, Thomson Reuters and ECB calculations. Notes: Probabilities implied by five-year zero-coupon inflation options, smoothed over five business days. Risk-neutral probabilities may differ significantly from physical, or true, probabilities. Latest observation: 30 August 2019.

Option-implied distribution of average inflation over the next five years

(percentage)

Sources: Eurostat and ECB calculations. Notes: Based on monthly observations. Latest observation: August 2019 (flash estimate).

0

10

20

30

40

50

60

70

80

90

100

2011 2013 2015 2017 2019

Below 0% Between 0% and 1.5%Between 1.5% and 2.0% Between 2.0% and 2.5%Above 2.5%

-1

0

1

2

3

4

5

2001 2004 2007 2010 2013 2016 2019

HICPHICP excluding energy and food (HICPx)

Rubric

www.ecb.europa.eu

© 4

A package of mutually reinforcing monetary policy measures

The ECB’s package of measures

EONIA, key ECB interest rates and excess liquidity

(lhs: percentage per annum; rhs: € bn)

Sources: Bloomberg, Thomson Reuters and ECB calculations. Latest observation: 30 August 2019.

0

500

1,000

1,500

2,000

-1

0

1

2

3

2013 2014 2015 2016 2017 2018 2019

Excess liquidity (rhs)EONIA (lhs)DFR (lhs)MLF (lhs)

1) Pushing the policy rate into negative territory

2) Forward guidance on the future policy path

3) APP

4) TLTROs

The measures work as a package, with significant complementarities across the different instruments.

Rubric

www.ecb.europa.eu

© 5

Upward pressures on euro area sovereign bond yields in absence of ECB’s non-standard measures 2014-18

(percentage per annum)

Source: Rostagno, Altavilla, Carboni, Lemke, Motto, Saint-Guilhem, Yiangou (2019), forthcoming. Notes: NIRP = negative interest rate policy; FG = forward guidance; APP = asset purchase programme. The chart shows the impact of ECB non-standard measures on the GDP-weighted aggregate of euro area sovereign bond yields. The APP impact is due to Eser, Lemke, Nyholm, Radde, and Vladu (2019). The impact of NIRP and forward guidance is derived from counterfactual analysis of OIS forwards based on the option-implied densities.

Estimated impact of the ECB’s package of measures on the term structure

0.0

0.4

0.8

1.2

1.6

0.0

0.4

0.8

1.2

1.6

2y 5y 10y 2y 5y 10y 2y 5y 10y 2y 5y 10y 2y 5y 10y

2014 2015 2016 2017 2018

NIRPFGAPP

Rubric

www.ecb.europa.eu

© 6

Contribution of ECB non-standard measures to real GDP growth 2014-18


Contribution of ECB non-standard measures to HICP inflation 2014-18


Source: Rostagno, Altavilla, Carboni, Lemke, Motto, Saint-Guilhem, Yiangou (2019), forthcoming. Notes: The chart shows the impact of ECB non-standard measures on macro variables based on a macroeconomic model with financial variables conditioning on the yield curve impact shown on the previous slide.

Source: Rostagno, Altavilla, Carboni, Lemke, Motto, Saint-Guilhem, Yiangou (2019), forthcoming. Notes: The chart shows the impact of ECB non-standard measures on macro variables based on a macroeconomic model with financial variables conditioning on the yield curve impact shown on the previous slide.

Contribution of ECB non-standard measures to growth and inflation 2014-18

0.0

0.5

1.0

1.5

2.0

2.5

3.0

0.0

0.5

1.0

1.5

2.0

2.5

3.0

2014 2015 2016 2017 2018

TLTRO NIRPFG APPReal GDP growth Counterfactual

-0.5

0.0

0.5

1.0

1.5

2.0

-0.5

0.0

0.5

1.0

1.5

2.0

2014 2015 2016 2017 2018

TLTRO NIRPFG APPHICP inflation Counterfactual

Rubric

www.ecb.europa.eu

© 7

Real GDP and short-term indicators (lhs: index, rhs: quarter-on-quarter percentage changes)

Sources: Markit, Eurostat, European Commission and ECB calculations. Latest observations: 2019Q2 for real GDP, August 2019 for PMI and ESI.

Unemployment rate, wages and core HICP (lhs: percentage per annum, rhs: percentage inverted)

Sources: Eurostat and ECB calculations. Latest observation: 2019Q1 for compensation per employee and 2019Q2 for the rest.

-0.25

0.00

0.25

0.50

0.75

1.00

47.5

50.0

52.5

55.0

57.5

60.0

2014 2015 2016 2017 2018 2019

Real GDP (rhs) Composite PMI (lhs)

EC ESI (lhs)

6

7

8

9

10

11

12

130.5

1.0

1.5

2.0

2.5

3.0

3.5

4.0

1999 2003 2007 2011 2015 2019

HICP excluding energy and food (lhs)Compensation per employee (lhs)Unemployment rate (inverted) (rhs)

Growth and unemployment

Rubric

www.ecb.europa.eu

© 8

Pass‐through of short‐term (1y) to long‐term (5y) SPF inflation expectations

(coefficients and confidence bands)

Sources: ECB, ECB Survey of Professional Forecasters (SPF), ECB calculations. Latest observations: 2019Q3.

Pass‐through of 1y1y to 5y5y market-based inflation expectations

(coefficients and confidence bands)

Sources: ECB, ECB calculations. Latest observation: August 2019.

-0.25

0.00

0.25

0.50

0.75

1.00

2010 2012 2014 2016 2018

84th percentileTime-varying parameter estimate - median16th percentile

-0.25

0.00

0.25

0.50

0.75

1.00

2010 2012 2014 2016 2018

84th percentileTime-varying parameter estimates - median16th percentile

The pass-through of short-term to long-term inflation expectations

Rubric

www.ecb.europa.eu

© 9





Conclusion 5

Outline

Rubric

www.ecb.europa.eu

© 10

New Keynesian Phillips Curve

The Structural Phillips Curve

𝜋𝜋�𝑡𝑡 − 𝜋𝜋�𝑡𝑡−1 = 𝜅𝜅 𝑦𝑦�𝑡𝑡 + 𝛽𝛽 𝐸𝐸𝑡𝑡 𝜋𝜋𝑡𝑡+1 − 𝛾𝛾𝜋𝜋𝑡𝑡 + 𝜑𝜑𝑡𝑡

slack inflation expectations

mark-up inflation

Rubric

www.ecb.europa.eu

© 11

New Keynesian Phillips Curve

The Structural Phillips Curve

𝜋𝜋�𝑡𝑡 − 𝜋𝜋�𝑡𝑡−1 = 𝜅𝜅 𝑦𝑦�𝑡𝑡 + 𝛽𝛽 𝐸𝐸𝑡𝑡 𝜋𝜋𝑡𝑡+1 − 𝛾𝛾𝜋𝜋𝑡𝑡 + 𝜑𝜑𝑡𝑡

slack inflation expectations

mark-up

𝑦𝑦�𝑡𝑡 = −1𝜎𝜎𝑖𝑖𝑡𝑡 − 𝐸𝐸𝑡𝑡 𝜋𝜋𝑡𝑡+1 − 𝑟𝑟𝑡𝑡𝑛𝑛 + 𝐸𝐸𝑡𝑡 𝑦𝑦𝑡𝑡+1

𝑖𝑖𝑡𝑡∗ = 𝜌𝜌 + 𝜋𝜋∗ + 𝜙𝜙𝜋𝜋 𝜋𝜋𝑡𝑡 − 𝜋𝜋∗ + 𝜙𝜙𝑦𝑦 𝑦𝑦�𝑡𝑡 + 𝜂𝜂𝑡𝑡

inflation

IS-relation

Policy rule

Rubric

www.ecb.europa.eu

© 12

The Phillips Curve slope in different ECB models

Coefficients Notation NAWM II ECB-Base Reduced-form

Slope �̂�𝜅 0.008 0.12 0.08

Forward term �̂�𝛽 0.998 0.63 0.16

Indexation term 𝛾𝛾� 0.230 0.39 0.51

Maximum multiplier �̂�𝜅

1 − �̂�𝛽 4.000 0.33 0.02

Sample 1985Q1- 2014Q4

2000Q1- 2017Q4

1995Q1-2019Q2

Type of model Structural Semi-structural Reduced-form

Sources: ECB calculations. Notes: Estimated Phillips-Curve parameters of a structural open-economy DSGE model (New Area-Wide Model II), a semi-structural model (ECB-BASE) and an average of a suite of reduced-form estimations. The reduced-form estimation uses the HICPx index as a dependent variable, the internal output-gap estimate of the ECB as a slack measure and various inflation expectations. The results for NAWM II and ECB-BASE report a maximum output-gap multiplier, which obtains as the change in the output gap becomes permanent.

Comparison of selected parameter estimates across ECB models

• reduced-form approaches tend to yield lower estimates of the PC slope

• the slope coefficients 𝜅𝜅 cannot be assessed independently of the estimated coefficient of the

forward-looking term 𝛽𝛽: higher 𝛽𝛽, future output gap matters more for today’s inflation → lower 𝜅𝜅 (see NAWM II vs. ECB-BASE)

Rubric

www.ecb.europa.eu

© 13

Phillips Curve slope estimates in light of the response to a monetary policy shock Short-term nominal interest rate

(x-axis: quarters, y-axis: pp)

GDP deflator inflation rate (x-axis: quarters, y-axis: pp)

Real output (x-axis: quarters, y-axis: percentage deviation of GDP from

steady state)

• the slope coefficients 𝜅𝜅 cannot be assessed independently of the estimated coefficient of the forward looking term 𝛽𝛽: higher 𝛽𝛽, future output gap matters more for today’s inflation → lower 𝜅𝜅 (see NAWM II vs. ECB-BASE)

• this effect is visible in the impulse response to a monetary policy shock being larger in the NAWM II than in ECB-BASE despite the fact that NAWMII has a lower PC slope-coefficient

Sources: ECB calculations. Notes: The graphs show the impulse responses to a one percentage point monetary policy shock in the structural New Area-Wide Model II (NAWM II) and the semi-structural ECB-BASE model.

-0.25

0.00

0.25

0.50

0.75

1.00

1.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-1.25

-1.00

-0.75

-0.50

-0.25

0.00

0.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-0.50

-0.40

-0.30

-0.20

-0.10

0.00

0.10

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-0.25

0.00

0.25

0.50

0.75

1.00

1.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-1.25

-1.00

-0.75

-0.50

-0.25

0.00

0.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-0.50

-0.40

-0.30

-0.20

-0.10

0.00

0.10

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-0.25

0.00

0.25

0.50

0.75

1.00

1.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-1.25

-1.00

-0.75

-0.50

-0.25

0.00

0.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-0.50

-0.40

-0.30

-0.20

-0.10

0.00

0.10

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-0.25

0.00

0.25

0.50

0.75

1.00

1.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

-1.25

-1.00

-0.75

-0.50

-0.25

0.00

0.25

4 8 12 16 20 24 28 32 36 40

ECB BASE NAWMII

Rubric

www.ecb.europa.eu

© 14





Conclusion 5

Outline

Rubric

www.ecb.europa.eu

© 15

One-year German sovereign bond yield Core consumer price index

Sources: ECB calculations. Notes: The figures plot the impulse responses of the one-year German sovereign bond yield, of the harmonized index of consumer prices excluding energy and food, and of the unemployment rate. The Phillips multiplier represents the coefficient 𝜅𝜅ℎ in the regression ∑ 𝜋𝜋𝑡𝑡+𝑖𝑖 = 𝜅𝜅ℎℎ

𝑖𝑖=0 ∑ 𝑢𝑢𝑡𝑡+𝑖𝑖 + ΓℎΦ 𝐿𝐿 𝑋𝑋𝑡𝑡 + 𝑒𝑒𝑡𝑡,ℎℎ𝑖𝑖=0 where the cumulative unemployment is instrumented by a proxy of a monetary policy shock.

The Phillips multiplier is not well defined, and therefore is estimated with wide confidence bands at short horizons (bands up to 6 month are excluded from the figure). At horizons between 7 and 18 months, the Phillips multiplier is estimated to be negative with a coefficient between -0.05 and -0.1. The methodology follows Barnichon-Mesters (2019).

Identifying the Phillips Curve slope using external instruments

Unemployment Phillips multiplier

-2.0-1.00.01.02.03.04.0

0 5 10 15 20

84th

One-year German sovereign bond yield16th

-0.50.00.51.01.52.02.5

0 5 10 15 20

84th

HICPx16th

-4.0-3.0-2.0-1.00.01.02.03.0

0 5 10 15 20

84th

Unemployment16th

-0.5-0.4-0.3-0.2-0.10.00.10.20.3

0 5 10 15 20

84th

Phillips multiplier16th

Rubric

www.ecb.europa.eu

© 16

Parameter estimates from panel regressions

Sources: ECB calculations. Notes: (1): OLS and (2) to (5) fixed-effects panel estimation using annual HICPx inflation and output gap, 1998-2018. For (1) heteroskedasticity-robust standard errors; for (2) to (5) robust standard errors are clustered at country level. P-values in parentheses. The countries included are: Austria, Belgium, Cyprus, Germany, Spain, Finland, France, Greece, Ireland, Italy, Luxembourg, Latvia, Malta, Netherlands, Portugal and Slovakia.

Identifying the Phillips Curve slope using cross-country variation

(1) (2) (3) (4) (5)

Regression Euro area Pooled Country FE

Time FE

Country, Time

Output gap 0.012 0.011 0.013 0.01 0.012

(0.032) (0.01) (0.007) (0.036) (0.02)

Constant 0.068 0.049 0.058 0.173 0.191

(0.012) (0.000) (0.001) (0.001) (0.001)

𝐇𝐇𝐇𝐇𝐇𝐇𝐇𝐇𝒙𝒙𝐭𝐭−𝟏𝟏 0.806 0.886 0.865 0.883 0.857

(0.001) (0.001) (0.001) (0.001) (0.001)

Observations 81 1296 1296 1296 1296

𝑹𝑹𝟐𝟐 0.74 0.85 0.85 0.87 0.87

Country FE No No Yes No Yes

Time FE No No No Yes Yes

Rubric

www.ecb.europa.eu

© 17





Conclusion 5

Outline

Rubric

www.ecb.europa.eu

© 18

Comparison between official and alternative slack measures (percentage)

Sources: Eurostat, European Commission, IMF, ECB calculations based on Jarocinski and Lenza (2018). Notes: the IMF and the EC output gap are depicted based on quarterly linear interpolation on annual data. Latest observation: 2019Q2.

Estimating the output gap that best predicts inflation

-6

-4

-2

0

2

4

6

1995 2000 2005 2010 2015

European Commission output gap IMF output gap

Output gap - Jarocinski-Lenza

Rubric

www.ecb.europa.eu

© 19

Conditional forecasts of HICPx inflation for alternative indicators of economic conditions


Sources: Eurostat, IMF, OECD, European Commission, ECB calculations. Notes: the range was obtained using alternative indicators of slack: unemployment rate, unemployment gap (the difference between unemployment and NAIRU), broad unemployment rate (U6), output gaps estimated by the OECD, IMF and European Commission. Latest observation: 2019Q2.

A simple bivariate approach

0.0

0.5

1.0

1.5

2.0

2.5

3.0

2001 2003 2005 2007 2009 2011 2013 2015 2017 2019

HICPx InflationIMF Unemployment rate

Broad unemployment rate

Rubric

www.ecb.europa.eu

© 20

Estimated Phillips Curve slope across all specifications

(regression coefficients of HICPx inflation on standardised slack measures)

Reduced-form estimates of the Phillips Curve slope based on different slack measures

Sources: European Commission, Eurostat, IMF, OECD and ECB calculations. Notes: We consider the following measures of slack: (1) output gap model-based estimate; (2) output gap IMF; (3) output gap European Commission; (4) output gap OECD; (5) unemployment rate; (6) unemployment gap model-based estimate; (7) unemployment gap IMF; (8) unemployment gap European Commission; (9) unemployment gap OECD; (10) short-term unemployment rate; (11) broad unemployment rate; (12) Jarocinski-Lenza output gap. The unemployment rates/gaps have been inverted. All measures of slack/tightness are standardised for the coefficients to be comparable across specifications. The vertical bars show the range of coefficients across all specifications including a particular measure of economic slack/tightness or activity.

0.00

0.05

0.10

0.15

0.20

0.25

0.30

1 2 3 4 5 6 7 8 9 10 11 12

Median

Rubric

www.ecb.europa.eu

© 21

Thick modelling: Phillips Curve-based decomposition of core inflation (percentage per annum and pp contributions;

all values in terms of deviations from their averages since 1999)

Sources: ECB calculations. Notes: The bars show average contributions across 600 models with different permutations of external price, economic slack and expectations measures. Contributions are derived as in Yellen, J. L., “Inflation Dynamics and Monetary Policy”, speech at the Philip Gamble Memorial Lecture, University of Massachusetts, Amherst, 24 September 2015. Latest observation: 2019Q2.

Decomposition of core inflation based on thick modelling

-0.8

-0.6

-0.4

-0.2

0.0

0.2

0.4

0.6

0.8

2008 2010 2012 2014 2016 2018

HICPx inflationEconomic slackInflation Expectations

External pricesUnexplained

Rubric

www.ecb.europa.eu

© 22

Phillips Curve-based decomposition of wage growth into its main drivers

(deviations from mean in year-on-year growth terms and pp contributions)

Source: Nickel, Bobeica, Koester, Lis, Porqueddu (2019) “Understanding low wage growth in the euro area and European countries”. Notes: Sample: 1995Q1-2018Q4. The blue line shows deviations of compensation per employee growth from its model-implied mean. Contributions (including residuals) are also shown as deviations from their model-implied mean. Contributions are derived as in Yellen, J.L. (2015). Latest observation: 2018Q4.

Decomposition of wage growth

-1.25

-1.00

-0.75

-0.50

-0.25

0.00

0.25

0.50

0.75

2011 2012 2013 2014 2015 2016 2017 2018

Compensation per employeeLabour market slack (unemployment rate)Past inflation

ProductivityUnexplained

Rubric

www.ecb.europa.eu

©

-4.00

-3.00

-2.00

-1.00

0.00

1.00

2.00

2000 2002 2004 2006 2008 2010 2012 2014 2016 2018

Unit profits - deviation from model meanDomestic demand shockWage mark-up shock

Oil supply shockForeign demand shockOther

23

BVAR-based structural decomposition of unit profits

(deviations from mean in year-on-year growth terms and pp contributions)

Source: ECB calculations. Latest observation: 2018Q4.

Historical decomposition of unit profit growth

Rubric

www.ecb.europa.eu

© 24





Conclusion 5

Outline

Rubric

www.ecb.europa.eu

© 25

Conclusion

• The structural Phillips Curve is a useful framework

• Slack in the economy, inflation expectations and markups are key determinants of inflation

• Using cross-country variation or external instruments are promising routes to identify the Phillips Curve slope

• Reduced-form empirical relation between slack and inflation provides a helpful contribution to the suite of forecasting models that we use at the ECB

The Phillips Curve at the ECB · The Phillips multiplier is not well defined, and therefore is estimated with wide confidence bands at short horizons (bands up to 6 month are excluded

Documents