An Estimated Two-Country DSGE Model for the Euro Area and the US Economy

1
An Estimated Two-Country DSGE Model for the Euro
Area and the US Economy

• Gregory de Walque - Frank Smets - Raf
  Wouters
• NBB ECB NBB
• EABCN/CEPR, Swiss National Bank
• Estimation and Validation of Structural Models
  for BC Analysis
• 29-30 August

2
1. INTRODUCTION

• This contribution extends the SW closed economy
  models for the EA US to an integrated
  two-country model. The result is an estimated
  medium-sized model in the NOEM-tradition.
• Most academic research on NOEM has been
  theoretical, with a few exceptions on small scale
  models
• Ghironi (1999), Bergin (2004), Lubik
  Schorfheide (2003-2005)...
• Central Banks have developed large-scale open and
  multi-country models based on the NOEM approach
  GEM, SIGMA, models at Riksbank, BoF, BoC, NAWM.
• Advances in Bayesian estimation techniques make
  it feasible to estimate medium-sized NOEM models
  now
• see Adolfson et al. (2004-2005), Adjémian and
  Darrac-Paries (2004). Justiniano and Preston
  (2004), Rabanal and Tuesta (2005)...

3
1. INTRODUCTION (continued)

• Insights from an estimated two-country EA-US
  model
• empirical test of a consistent model for the
  Trade Balance and the Exchange Rate behaviour and
  their relation with the domestic variables and
  the monetary policy reaction
• based on the marginal likelihood for different
  model specifications
• based on the stylised facts for the main open
  economy variables.
• the focus of this empirical exercise is
• on the estimation of the elasticity of
  substitution between domestic and foreign goods
• on the restrictions implied by the UIRP condition
  for the exchange rate dynamics and the overall
  model dynamics.

4
1. INTRODUCTION (continued)

• Insights from an estimated two-country EA-US
  model
• the estimated model allows to assess the role of
  various types of shocks in explaining the
  exchange rate volatility as well as trade balance
  variations
• contribution of domestic shocks
• contribution of spill-over effects between the
  two main economies
• contribution of ROW shocks oil shocks, UIRP
  shocks, demand and price shocks.

5
1. INTRODUCTION (continued)

• Overview of the presentation
• model specification
• estimation results issues
• model validation
• structural IR productivity, monetary policy and
  oil shocks
• reproducing stylised facts
• variance decomposition
• historical decomposition

6
2. MODEL DESCRIPTION

• Euro Area and US are modelled symmetrically
• Each country block contains three types of agents
• households consume, work, set wages, invest,
  allocate wealth between one period domestic and
  foreign bonds of which relative return determines
  the exchange rate through UIP
• firms
• central bank sets short-term interest rate
• Rest of the World is introduced via exogenous
  shocks

7
2. MODEL DESCRIPTION household and firm sector

• Contains a relatively large number of real and
  nominal frictions (CEE 2005 ACEL 2005 - SW
  2003)
• Monopolistic competition in goods and labour
  markets with sticky nominal prices and wages
• Calvo pricing with partial indexation of prices
  and wages on lagged inflation
• Endogenous elasticity of demand (Eichenbaum
  Fischer 2004)
• Costs of adjustment in capital accumulation as a
  function of change in investment
• External habit formation
• Variable capital utilisation and fixed costs.

8
2. MODEL DESCRIPTION detailed firm sector
9
(No Transcript)
10
(No Transcript)
11
(No Transcript)
12
(No Transcript)
13
2. MODEL DESCRIPTION shocks

• supply TFP and investment specific technology
  shocks
• demand intertemporal risk premium and government
  spending shocks
• monetary policy interest rate shocks
• mark-up shocks wages and producer prices
  (ARMA(1,1)), consumer prices (i.i.d. measurement
  error)
• ROW shocks demand and price
• oil price shocks
• UIRP shocks
• zero covariance between all shocks is assumed.

14
3. DESCRIPTION OF THE DATA

• 1974Q1-2004Q4 data for euro area and US (SW JAE
  2005)
• 10 country specific series
• growth rate in real GDP, consumption, investment,
  wages and employment (hours)
• inflation in GDP, consumption and import price
  deflator
• nominal short term interest rate
• real trade balance
• 2 global series
• / exchange rate depreciation and oil price
  inflation.

15
4. ESTIMATION RESULTS 3 model versions

• model with a high and with a low substitution
  elasticity
• ? around a critical value of this parameter
  (1.5) the exchange rate is extremely volatile
  and the model has an very low marginal likelihood
  (see discussion for IR of productivity shock
  below).
• model without UIRP and exogenous AR(1) process is
  assumed for the exchange rate
• ? empirical test of the restrictions imposed
  by UIRP on the overall model dynamics and the
  systematic response of the exchange rate.

-
-
-
16
4. ESTIMATION RESULTS parameter estimates

• parameter related to the domestic economy block
  are very similar to the estimates obtained in the
  closed economy models for all three model
  versions
• the exception is the calvo price parameter due to
  the introduction of the endogenous demand
  elasticity (following Eichenbaum and Fischer
  2004)
• stochastic structure will become clear when
  looking at the variance decomposition
• parameters related to the open economy block are
  most interesting both structural and behavioural
  parameters are estimated.

17
4. ESTIMATION RESULTS domestic economy block
18
4. ESTIMATION RESULTS parameter estimates

• parameter related to the domestic economy block
  are very similar to the estimates obtained in the
  closed economy models for all three model
  versions
• the exception is the calvo price parameter due to
  the introduction of the endogenous demand
  elasticity (following Eichenbaum and Fischer
  2004)
• stochastic structure will become clear when
  looking at the variance decomposition
• parameters related to the open economy block are
  most interesting both structural and behavioural
  parameters are estimated.

19
4. ESTIMATION RESULTS open economy block (1)
20
4. ESTIMATION RESULTS open economy block (2)
21
5. MODEL VALIDATION

• structural impulse response
• TFP productivity shock example of a supply shock
  to illustrate the importance of the elasticity
  substitution
• monetary policy shock
• OIL shock
• stylised facts variances and correlations
  between main open economy macroeconomic series
• variance decomposition

22
5.1 MODEL VALIDATION IRF TFP shock

• In the typical closed economy model
• prod. frontier,
• - MC and int. dom. price,
• real wage, cons. and invest (- hours worked)
• the exact size of these effect depends on the
  mon. pol. react.
• In the open economy set-up, RER acts as a risk
  sharing device distributing wealth across
  economies
• the elast. of sub. plays an important role

23
5.1 MODEL VALIDATION IRF TFP shock

• Large elast. of substitution
• large switching expenditure effects
• lead to a depreciation of the RER to redirect the
  increased domestic good supply towards the
  foreign market
• terms of trade losses leads to CA deficit
• CA stabilised in the LR through the positive net
  trade effect
• Everything else equal, if the elast. of subst.
  decreases the switching expenditure effect
  decreases too and a larger RER depreciation is
  needed to stabilise the CA

24
5.1 MODEL VALIDATION IRF TFP shock

• Marshall-Lerner
• there is some critical value for the elast. of
  substitution such that the impact of the RER on
  the CA reverses
• below this threshold, the terms of trade wealth
  effects dominate the switching expenditure effect
• In our dynamic set-up, such a critical value
  exists too and its exact value is influenced
• by the degree of the bilateral RER pass through
• by the speed of the bilateral RER pass through
• by the share of the domestic good component in
  the import distribution sector

25
5.1 MODEL VALIDATION IRF TFP shock

• IRF of the RER and consumption to a TFP shock in
  the EA for different values of the elasticity of
  substitution

2
3
3
4
1
4
1
1
4
2
2
3
3.01
1.60
1.25
0.80
1
2
3
4
26
5.1 MODEL VALIDATION IRF TFP shock

• IRF of the RER and consumption to a TFP shock in
  the EA for different values of the elasticity of
  substitution

3
3
2
4
4
1
1
1
4
2
2
3
3.01
1.60
1.25
0.80
1
2
3
4
27
5.1 MODEL VALIDATION IRF TFP shock



28
5.1 IMPULSE-RESPONSE MON. POLICY SHOCK

• Interest rate increase leads to negative effects
  on
• consumption and investment,
• real wage and prod. price inflation
• ... but it also lead to an exchange rate
  appreciation
• positive wealth effect on consumption
• negative effect on foreign demand of dom. goods
  through substitution effect (adj. costs)
• pass-through effect on inflation PCP and sticky
  prices
• spillover effects remain small and depend on
  elasticity of substitution

29
5.1 IMPULSE-RESPONSE MON. POLICY SHOCK
30
5.1 IMPULSE-RESPONSE OIL PRICE SHOCK

• impact on GDP is around 0.1 the first two years
  for a 1s shock (with s 17)
• effects on prices complex pass-through
• immediate through final good production
• gradual through intermediate domestic good
  production
• domestic demand declines because of the negative
  wealth effects for oil importing economies
  together with monetary policy reaction
• N.B. real wage drop mitigates the effect of the
  oil price on the domest. firms MC

31
5.1 IMPULSE-RESPONSE OIL PRICE SHOCK

• CA deficits stabilise
• through a RER depreciation and positive Real
  Trade Balance reacts slightly positive (see
  Jimenez-Rodriguez and Sanchez, 2004) if large
  elast. of substitution
• through a RER appreciation inducing lower TOT if
  low elast. of subst.
• If the monetary policy reacts on producer price
  inflation instead of consumption inflation, the
  real interest rate is lower and the domestic
  demand and output are less affected, but at the
  cost of higher and more persistent inflation.

32
5.1 MODEL VALIDATION IRF OIL shock



33
5.1 MODEL VALIDATION IRF OIL shock



34
5.2 STYLISED FACTS
35
5.2 STYLISED FACTS (continued)
36
5.2 STYLISED FACTS (continued)
w markup
37
5.2 STYLISED FACTS (continued)

• this could indicate that a more elaborated
  financial structure allowing more risk-sharing
  and affected by some common shock could help to
  reproduce the observed synchronisation of the
  business cycles

38
5.3 VARIANCE DECOMPOSITION model with HIGH subst.
Nominal
39
5.3 VARIANCE DECOMPOSITION model with HIGH subst.
40
5.3 VARIANCE DECOMPOSITION model with HIGH subst.
41
5.3 VARIANCE DECOMPOSITION model with HIGH subst.
42
5.4 HISTORICAL DECOMPOSITION since 1998

• output and inflation mainly determined by
  domestic shocks, spillover and ROW shocks have
  minor effects
• oil price fluctuations in 1998-99 had 0.5 effect
  on output in US and EA
• UIRP shocks had minor effects on output (larger
  effect on the composition)
• trade balance are influenced by both domestic,
  foreign and row shocks. But overall, the UIRP
  shocks has been main determinant over the last
  five years its influence works only gradually
  over time
• UIRP dominant shock behind USD-EUR exchange rate
  fluctuations

43
5.4 HISTORICAL DECOMPOSITION output EA
44
5.4 HISTORICAL DECOMPOSITION output US
45
5.4 HISTORICAL DECOMPOSITION EURO-USD
46
5.4 HISTORICAL DECOMPOSITION EURO-USD
47
5.4 HISTORICAL DECOMPOSITION EURO-USD
US Trade Balance
48
5.4 HISTORICAL DECOMPOSITION EURO-USD
US Trade Balance
49
Conclusions

• Work in progress...
• Multiple applications/extentions with the model
  are possible
• The success in explaining exchange rate and trade
  balance developments is very modest UIRP
  restrictions are rejected
• Introduce risk premium or imperfect information
  learning
• Relation between productivity markup shocks and
  exchange rate depends on elast. of subst gt
  product innovation firm entry
• Wealth effects of terms of trade are probably
  overestimated
• Introduce imperfect long run pass-through
• Spillover effects are minor co-movement is
  unexplained
• Use positive prior on correlation between country
  shocks
• Introduce more risk sharing to reflect capital
  market integration.

50
(No Transcript)