Macroeconomic implications of Fdric Holm-Hadulla European Central - - PowerPoint PPT Presentation

▶
macroeconomic implications of
SMART_READER_LITE
LIVE PREVIEW

Macroeconomic implications of Fdric Holm-Hadulla European Central - - PowerPoint PPT Presentation

Mar 16, 2024 295 likes •666 views

Macroeconomic implications of Fdric Holm-Hadulla European Central Bank oil price fluctuations Kirstin Hubrich Federal Reserve Board A regime-switching framework for the euro area Nonlinear Models in Macroeconomics and Finance for an

slide-1
SLIDE 1

Macroeconomic implications of

  • il price fluctuations

A regime-switching framework for the euro area Nonlinear Models in Macroeconomics and Finance for an Unstable World

January.26-27, 2018 The views expressed here are those of the authors and do not necessarily reflect those of the European Central Bank, the Federal Reserve Board or the Federal Reserve System or its staff.

.

Fédéric Holm-Hadulla European Central Bank Kirstin Hubrich Federal Reserve Board

slide-2
SLIDE 2

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

2

Oil price fluctuations typically trigger divergent assessments

Brent crude oil prices

(in different currencies)

Cheaper oil is a rare piece of good news for (…) the euro currency area, since [it] should boost the spending power of Europe’s consumers (…) amid the eurozone’s long slump.

Wall Street Journal, 14 November 2014

(...) a danger [of the oil-price slump] is that an even deeper dip in inflation (…) may have an unwelcome second- round effect by dragging down inflation expectations.

The Economist, 4 December 2014

Source: Bloomberg

Example: oil price slump in 2014H2

slide-3
SLIDE 3

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

3

… but central banks have to form a view on the macro implications in real-time

Brent crude oil prices

(in different currencies)

“In principle, if commodity price changes are of a temporary nature, one can look through the volatility in inflation triggered by their first-round effects. However, the risk of second round effects must be contrasted (…) to prevent that they have a lasting impact on medium-term inflation expectations (…) In such cases, an adjustment of the monetary policy stance would be required to preserve price stability and keep inflation expectations well- anchored.”

Mario Draghi before ECON Committee, June 2011

Source: Bloomberg

Commodity price fluctuations in the ECB’s reaction function

slide-4
SLIDE 4

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

4

Overview of our paper

Aim: model episodic changes in transmission of oil price shocks to the economy in a regime-switching VAR model with time-varying transition matrix Key findings:

  • Oil price fluctuations typically exert limited effects on inflation and economic

activity (‘normal regime’), e.g. downward oil price shock leads to higher growth

  • Occasionally, economy enters into ‘adverse regime’ in which:
  • Oil price shocks trigger sizeable and sustained macroeconomic effects
  • Inflation and economic activity move in the same direction as the oil price

shock

  • …as do inflation expectations, consistent with presence of second-round

effects

  • Role of wage change as channel for a wage-price spiral / second-round effects
  • Model assigns ‘pre-APP episode’ (mid-2014 to early-2015) to adverse regime
slide-5
SLIDE 5

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

5

Related economic literature

Relevance of source of oil price shocks:

  • Disentangle oil supply, aggregate demand & precautionary oil demand shocks

using structural VARs e.g., Kilian (2009); Jo (2014), Caldara, Cavallo & Iacoviello (2016) Differences in transmission of oil price shocks:

  • Assess how impact of oil price shocks has differed across historical episodes

e.g. Blanchard & Galí (2007); Nakov & Pescatori (2010)

  • Explicitly model non-linearities/time-variation in impact of oil shocks (US)

e.g. Hamilton (2003); Baumeister & Peersman (2013); Leduc, Moran & Vigfusson (2016), Bjørnland, Larsen and Maih (2018) Monetary policy response to oil price shocks

  • Assess role of monetary policy as propagator of oil price shocks, ZLB

e.g. Bernanke, Gertler and Watson (1997); Bodenstein, Guerrieri and Kilian (2012); Bodenstein, Guerrieri and Gust (2013)

slide-6
SLIDE 6

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

Our paper is the first to

  • Model time-variation in impact of oil price shocks on euro area

macroeconomy

  • Explicitly account for inflation expectations
  • Employ novel regime-switching VAR framework with time-varying transition

matrix Contribution to the literature

6

slide-7
SLIDE 7

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

7

Regime-switching Structural VAR model with time-varying transition matrix

Hubrich, Waggoner and Zha (2015) 𝐵0 𝑡𝑢

𝑑 𝑧𝑢 = 𝐵+ 𝑡𝑢 𝑑 𝑦𝑢 + Ξ−1(𝑡𝑢 𝑤)ε𝑢

(1) 𝑧𝑢: Endogenous variables; 𝑦𝑢

′ = [𝑧𝑢−1 ′

, … , 𝑧𝑢−𝑞

′

, 1] ε𝑢: Vector of standard normal shocks 𝐵0 𝑡𝑢

𝑑 , 𝐵+ 𝑡𝑢 𝑑 : Coefficient matrices

Ξ−1 𝑡𝑢

𝑤 : Diagonal matrix with standard deviations of shocks

  • Previous literature: MS-SVAR constant transition matrix (Sims & Zha, AER, 2006;

Sims, Waggoner & Zha, JoE, 2008; Hubrich and Tetlow, JME, 2015) 𝑡𝑢 = (𝑡𝑢

𝑑, 𝑡𝑢 𝑤): Unobserved state variables evolve according to two independent

first-order Markov processes

  • Hubrich, Waggoner and Zha (2015): time-varying transition matrix
slide-8
SLIDE 8

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

8

Specification of time-varying transition matrix

Regime-Switching SVAR model: Transition matrix 𝑞𝑗,𝑘,𝑢: time-varying probability of switching from regime j to i,

  • 𝑞𝑗,𝑘,𝑢 denotes 𝑞(𝑡𝑢+1 = 𝑗 | 𝑡𝑢 = 𝑘, 𝑍

𝑢,θ, 𝑟)

  • Diagonal elements of 𝑞𝑗,𝑘,𝑢 give the time-varying persistence of 𝑘𝑢𝑢 regime:

𝑞𝑘,𝑘,𝑢 =

1 1+𝑓−𝑣𝑘,𝑢

where 𝑣𝑘,𝑢 = 𝑑

𝑘 + γ𝑘𝑧𝑢,(𝑢−𝑙+1)

and: 𝑧𝑢,(𝑢 −𝑙+1)

′

= [𝑧𝑢

′ ,…, 𝑧 (𝑢−𝑙+1) ′

]

  • Intercept and slopes determine transition process
slide-9
SLIDE 9

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

9

Specification of time-varying probabilities

Regime Switching SVAR model: Transition matrix 𝑞𝑗,𝑘,𝑢: time-varying probability of switching from regime j to i

  • 𝑞𝑗,𝑘,𝑢 denotes 𝑞(𝑡𝑢+1 = 𝑗|𝑡𝑢 = 𝑘, 𝑍

𝑢,θ, 𝑟)

  • Off diagonal elements for application with 2 regimes:

𝑞𝑗,𝑘,𝑢 = 1 − 𝑞𝑘,𝑘,𝑢 where 𝑞𝑗,𝑘,𝑢 + 𝑞𝑘,𝑘,𝑢 = 1

  • Off diagonal elements for more than 2 regimes
  • Off-diagonal elements sum to 1-𝑞𝑘,𝑘,𝑢, (scaled) Dirichlet prior
slide-10
SLIDE 10

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

10

  • Estimation with Bayesian methods
  • Estimation of posterior mode:
  • Blockwise BFGS optimization algorithm
  • Algorithm: parameters divided into blocks; initial guesses for parameters used

in hill-climbing quasi-Newton optimization routine

  • Use draws from the simulations of the posterior distribution as starting points
  • Dynamic Striated Metropolis Hastings sampler (Waggoner, Wong & Zha, 2016)

Model estimation

slide-11
SLIDE 11

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

11

Macroeconomic Effects of Oil Price Fluctuations

Regime-Switching SVAR model Data and Identification

  • 𝑧𝑢 = Δ𝑗𝑞, π, Δ𝑞𝑝𝑗𝑝, 𝐺𝐺, π𝑓, 𝑆
  • 𝑗𝑞: industrial production;
  • π: HICP inflation;
  • 𝑞𝑝𝑗𝑝: Brent crude oil price (in USD);
  • 𝐹𝐺𝑆: USD/EUR exchange rate;
  • π𝑓: 5Y5Y BEIR
  • 𝑆: 3-month EURIBOR
  • Additional specification: change in nominal negotiated wages (Δw) added
  • Baseline sample: euro area aggregates, monthly frequency, Feb 2004 to Jan 2015

(availability of 5Y5Y BEIR is restraining factor for start of sample period);

  • Different sample extensions
  • Identification: Cholesky decomposition, variables ordered as shown above
  • Persistence of regime: depends on oil price inflation
slide-12
SLIDE 12

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

12

Impulse responses to downward oil price shock

Impulse response functions Downward (negative) oil price shock

  • Model reveals relevant differences

in economic dynamics across regimes

  • Normal regime:
  • Oil price shocks only trigger

small macroeconomic effects

  • Increase in growth
  • Adverse regime:
  • Inflation declines and inflation

expectations decline

  • Output growth declines
  • Effects are long-lasting
  • MP loosens but not sufficiently to

pre-empt second-round effects

slide-13
SLIDE 13

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

13

Regime probabilities

Probability of being in a normal regime (grey-shaded area) and conditional probability of staying in that regime (black line)

Note: on the x-axis ’05 refers to the beginning of the year 2005 etc.

  • Euro area economy entered adverse regime at various occasions
  • Typically switch after sequence of pronounced, unidirectional oil price changes
  • Conditional probability of staying in normal regime declined steeply in 2014H2
  • Overall, supports unfavourable interpretation of that episode of oil price declines
slide-14
SLIDE 14

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

14

Counterfactual

Counterfactual Experiment Main findings

  • Consider regime switch in August

2014: What if no regime change and stay in normal regime?

  • Actual compared to Counterfactual

path

  • Higher path for oil price changes
  • Inflation higher
  • Inflation expectations 0.4pp

higher, substantial since move within a narrow range

  • Growth substantially higher
slide-15
SLIDE 15

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

15

Impulse responses to downward oil price shock, Model with nominal negotiated wages

Impulse response functions Downward oil price shock

  • Model reveals relevant differences in

economic dynamics across regimes

  • Normal regime:
  • Increases in growth
  • Inflation declines
  • Declines in Nom. wage growth, but
  • nly modestly
  • Adverse regime:
  • Inflation declines (after a year

increase due to oil price dynamics)

  • Inflation expectations
  • Nominal wage growth declines (with

lag)

  • Substantial growth decline
  • MP loosens but not sufficiently to

pre-empt second-round effects

slide-16
SLIDE 16

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

16

Summary and conclusion

  • Depending on source and transmission of underlying shock, observed oil price

fluctuations may have very different macroeconomic consequences

  • Aim of our paper is to model episodic changes in transmission of oil price shocks to the

economy in a regime-switching SVAR with time-varying transition matrix

  • Key findings:
  • Oil price fluctuations typically exert limited effects on inflation and economic activity

(‘normal regime’), e.g. downward oil price shock leads to higher growth

  • Occasionally, economy enters into ‘adverse regime’ in which:
  • il price shocks trigger sizeable and sustained macroeconomic effects
  • inflation and economic activity move in the same direction as the oil price shock
  • …as do wage changes and inflation expectations, consistent with presence of second-round

effects

  • Model assigns ‘pre-APP episode’ (mid-2014 to early-2015) to adverse regime
  • Key contribution:
  • Model helps assess effect of oil price fluctuations in real-time and inform

deliberations on the adequate policy response.

slide-17
SLIDE 17

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

17

Background

slide-18
SLIDE 18

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

18

Robustness check (oil prices in EUR)

Impulse response functions Main findings

  • Ultimately, it is the oil price in EUR that

matters for EA consumers and firms

  • Baseline specification includes oil price

in USD and USD/EUR exchange rate

  • Robustness test (incl. oil price in EUR)

confirms key results of baseline spec.

  • Nearly identical responses of growth,

inflation, and inflation expectations

slide-19
SLIDE 19

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

19

Regime probabilities

Probability of being in a normal regime (grey-shaded area) and conditional probability of staying in that regime (black line)

Note: on the x-axis ’05 refers to the beginning of the year 2005 etc.

  • Euro area economy entered adverse regime at various occasions
  • Typically switch after sequence of pronounced, unidirectional oil price changes
  • Conditional probability of staying in normal regime declined steeply in 2014H2
  • Overall, supports unfavourable interpretation of that episode of oil price declines
slide-20
SLIDE 20

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

20

Past policy action motivated by risk of oil-price induced 2nd-round effects

Brent crude oil prices and inflation expectations (LHS: USD; RHS: % per annum)

“(…) we decided at today’s meeting to increase the key ECB interest rates by 25 basis points. This decision was taken to prevent broadly based second-round effects.”

Introductory Statement, 3 July 2008

“While the sharp fall in oil prices over recent months remains the dominant factor driving current headline inflation, the potential for second-round effects (…) has increased. This assessment is underpinned by a further fall in market-based measures of inflation expectations.”

Introductory Statement, 22 January 2015

Source: Bloomberg

Commodity price fluctuations in the ECB’s reaction function

slide-21
SLIDE 21

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

Markov Switching Model Literature

Markov switching with constant transition matrix Hamilton (1989); Chauvet (1998); Kim and Nelson (1999); Fruehwirth-Schnatter (2004); Sims and Zha (2006), Sims, Waggoner, Zha (2008); Luetkepohl, Lanne & Maciejowska (2010); Herwartz & Luetkepohl (2014); Brunnermeier, Palia & Sims (2014) Regime-switching regression models with time-varying transition matrix Filardo (1994); Diebold, Lee and Weinbach (1994); Kim (2004); Kim, Piger and Startz (2008); Bazzi, Blasques, Koopman, Lucas (2014); Chang, Choi and Park (2014)

Related methodological literature

21

slide-22
SLIDE 22

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

22

Impulse response functions Main findings

  • Model reveals relevant differences

in economic dynamics across regimes

  • Constant parameter VAR:
  • may underestimate effect of oil

price shock in adverse regime

  • may give wrong sign for output

and inflation response in normal regime

Impulse responses to negative oil price shock

slide-23
SLIDE 23

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

23

Extension of Sample until December 2015, with long-term interest rate

Impulse response functions Main findings

  • We extend the sample to December

2015

  • Long-term real interest rate included to

capture potential effects of non- standard measures

  • No inflation expectations to keep

specification parsimonious

  • Very similar responses of growth and

inflation in respective regimes

slide-24
SLIDE 24

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

24

Robustness check (oil prices in EUR)

Probability of being in a normal regime (grey-shaded area) and conditional probability of staying in that regime (black line)

Note: on the x-axis ’05 refers to the beginning of the year 2005 etc.

  • Assignment of time periods to different regimes broadly unaffected
  • Some additional adverse-regime episodes
  • Period around the turn of 2015 again assigned to adverse regime
  • …. and drop in cond. probability of staying in normal regime in 2014H2

confirmed

slide-25
SLIDE 25

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

25

Impulse responses to negative oil price shock, Extended sample until 2015(12)

Impulse response functions Main findings

  • Model reveals relevant differences

in economic dynamics across regimes

  • Normal regime:
  • il price shocks only trigger small

macroeconomic effects

  • Adverse regime:
  • Growth and inflation decline
  • effects on growth long-lasting
  • MP loosens but not sufficiently to

pre-empt second-round effects

slide-26
SLIDE 26

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

26

Counterfactual

Counterfactual Experiment Main findings

  • Consider regime switch in August

2014: What if no regime change?

  • Assume inflation expectations do

not drift down

  • Impose actual average interest

rate path

  • Actual compared to Counterfactual

path

  • Higher path for oil price and

inflation expectations

  • Growth and inflation higher
slide-27
SLIDE 27

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

27

Counterfactual

Counterfactual Experiment Main findings

  • Consider regime switch in August

2014: What if no regime change?

  • Assume inflation expectations do

not drift down

  • Impose actual average interest

rate path

  • Actual compared to Counterfactual

path

  • Higher path for oil price and

inflation expectations

  • Growth and inflation higher
slide-28
SLIDE 28

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

28

Evolution of oil-price changes (variable included in VAR)

Y-o-Y changes in price of oil (in %)

Source: Bloomberg

slide-29
SLIDE 29

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

29

Evolution of inflation expectations (variable included in VAR)

Breakeven inflation rate, 5y5y (in %)

1,5 1,7 1,9 2,1 2,3 2,5 2,7 2,9 20050331 20050630 20050930 20051231 20060331 20060630 20060930 20061231 20070331 20070630 20070930 20071231 20080331 20080630 20080930 20081231 20090331 20090630 20090930 20091231 20100331 20100630 20100930 20101231 20110331 20110630 20110930 20111231 20120331 20120630 20120930 20121231 20130331 20130630 20130930 20131231 20140331 20140630 20140930 20141231 20150331 IE_5Y5YBEIR

slide-30
SLIDE 30

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

30

Nominal wage changes (variable included in VAR)

Nominal wage changes (yoy change in %)

0,5 1 1,5 2 2,5 3 3,5 4 4,5 20040201 20040501 20040801 20041101 20050201 20050501 20050801 20051101 20060201 20060501 20060801 20061101 20070201 20070501 20070801 20071101 20080201 20080501 20080801 20081101 20090201 20090501 20090801 20091101 20100201 20100501 20100801 20101101 20110201 20110501 20110801 20111101 20120201 20120501 20120801 20121101 20130201 20130501 20130801 20131101 20140201 20140501 20140801 20141101 20150201 20150501 20150801 20151101

dWneg

slide-31
SLIDE 31

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

31

Inflation (variable included in VAR)

HICP (yoy change in %)

  • 1

1 2 3 4 5 20000131 20000531 20000930 20010131 20010531 20010930 20020131 20020531 20020930 20030131 20030531 20030930 20040131 20040531 20040930 20050131 20050531 20050930 20060131 20060531 20060930 20070131 20070531 20070930 20080131 20080531 20080930 20090131 20090531 20090930 20100131 20100531 20100930 20110131 20110531 20110930 20120131 20120531 20120930 20130131 20130531 20130930 20140131 20140531 20140930 20150131 20150531 20150930

dHICP

slide-32
SLIDE 32

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

32

Industrial production growth (variable included in VAR)

Industrial production growth (yoy change in %)

  • 30
  • 25
  • 20
  • 15
  • 10
  • 5

5 10 15 20000131 20000531 20000930 20010131 20010531 20010930 20020131 20020531 20020930 20030131 20030531 20030930 20040131 20040531 20040930 20050131 20050531 20050930 20060131 20060531 20060930 20070131 20070531 20070930 20080131 20080531 20080930 20090131 20090531 20090930 20100131 20100531 20100930 20110131 20110531 20110930 20120131 20120531 20120930 20130131 20130531 20130930 20140131 20140531 20140930 20150131 20150531 20150930

dIP

slide-33
SLIDE 33

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

33

5Y5YBEIR vs 5Y5YILS

5Y5YBEIR vs 5Y5YILS

1,5 1,7 1,9 2,1 2,3 2,5 2,7 2,9 2005… 2005… 2005… 2005… 2006… 2006… 2006… 2006… 2007… 2007… 2007… 2007… 2008… 2008… 2008… 2008… 2009… 2009… 2009… 2009… 2010… 2010… 2010… 2010… 2011… 2011… 2011… 2011… 2012… 2012… 2012… 2012… 2013… 2013… 2013… 2013… 2014… 2014… 2014… 2014… 2015… IE_5Y5YBEIR HICP_5Y5Y

slide-34
SLIDE 34

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

Dynamic Striated Metropolis Hastings sampler Basic idea:

  • Tractable initial distribution one can sample from
  • Transform initial distribution gradually to desired posterior distribution through sequence of

stages

  • Grounded in Metropolis-Hastings, but combines with the strength of equi-energy and

sequential Monte Carlo samplers

  • Differs from other methods in how information from previous stage is transmitted to current

stage

  • Allows to compute MDDs as by-product

Sampler

34

slide-35
SLIDE 35

Rubric

Holm-Hadulla & Hubrich Macroeconomic implications of oil price fluctuations

Simulation of Posterior Distribution Posterior distribution and model evaluation (statistical):

  • Marginal Data Densities often via Modified Harmonic Mean (Gelfand & Dey, 1994)
  • MHM might be unreliable when posterior distributions far from Gaussian and extremely

irregular with multiple peaks

  • Recently growing literature on new methods to compute posterior distributions
  • Different methods within class of Sequential Monte Carlo methods developed, e.g.

Durham & Geweke (2014), Herbst & Schorfheide (2014), Bognanni & Herbst (2014), Waggoner, Wong & Zha (2016)

  • Here: Dynamic Striated Metropolis Hastings sampler,

Waggoner, Wong & Zha (2016)

35

Simulation