Banco de España Research Conference
Madrid, February 25, 2010
Interactions between Monetary and Fiscal Policies
Keynesian Government Spending
Multipliers and Spillovers
in the Euro Area
Tobias Cwik and Volker Wieland
Goethe University of Frankfurt, CFS and CEPR
‹#›
Heated Debate over Major Policy Experiments
U.S.A.: 2008: tax rebates, 2009: ARRA
American Recovery and Reinvestment Act, $ 787bln
Europe: 2008/9: EERP
The European Economic Recovery Plan, National
plans: for example, in Germany, Konjunkturpaket
1 und 2
 Major discretionary fiscal policy initiatives with
possibly large impact on economic welfare.
 Heated debate in policy circles and academia:
Convictions abound, competitors/opponents are
easily dismissed.
‹#›
Quantitative Analysis and Policy Advice
 Investigate the actual policy experiments, not
standardized shocks.
 Consider competing, empirically-based views
of the structure of the economy (i.e. account for
model uncertainty).
 Account for a realistic range of scenarios
(monetary-fiscal interactions, zero bound
effect, implementation lags, …).
 Policy advice should be robust! Develop
decision criteria that account for competing
model and realistic scenarios.
‹#›
Some Questions
1. How big is the actual € area stimulus? Which
countries?
2. What GDP effects should we expect in the
euro area? Is there a Keynesian multiplier
effect or crowding out?
 Account for model uncertainty and realistic
scenarios.
3. What are the spillover effects between
countries in the union?
 We make use of a model archive that offers a
new platform for a comparative approach to
model-based policy analysis.
‹#›
1. How big is the Euro area stimulus?
Total Package in %GDP: 2009: 1.01 2010: 0.85
Expenditures in %GDP: 2010: 0.58 2010: 0.22
German stimulus 50 % of EU 11
German gov. expenditures: 43% of EU 11
German package %GDP: 3.37% (09: 1.44% 10: 1.93% )
‹#›
EU 11 Stimuli
‹#›
‹#›
2. What is the GDP effect of the
Stimulus? Multipliers?
 Focus on government spending which
promises the largest multiplier.
 Model uncertainty: consider wide range of
competing Keynesian-style models,
identical monetary policy reaction.
 Alternative scenarios:
(1) Actual spending plan (Immediate
implementation with slight phase-in)
(2) Implementation delay.
(3) Zero bound: notional interest rate target
negative  fiscal stimulus boosts output,
but output increase does not immediately
‹#›
lead to policy tightening.
Model Uncertainty & Robustness
 New-Keynesian DSGE models:
- Smets and Wouters 2003, (ECB), Euro Area
- Laxton&Pesenti 2003 (IMF), Euro Area +Cz.Rep.
- EU- Euro area model 2009 (EU-QUEST) (35%
liquidity-constrained consumers)
 New-Keynesian
- Taylor G-7 model, 1993
 Old-Keynesian
- ECB Area Wide model, 2004
‹#›
New-Keynesian DSGE Models of
IMF and EU Researchers
‹#›
NK-Models
 The increase in GDP quickly produces a
permanent contraction in private sector
saving and consumption.
 Households anticipate that government debt
incurred needs to be paid off with interest
by raising taxes in the future. (SW-SmallIMF assume lump-sum/ non-distortionary
taxes)
 Interest rates rise. Crowding out of
investment and consumption. No multiplier.
‹#›
Taylor and ECB-AWM Models
‹#›
Scenario (2): Implementation Lag
‹#›
GDP Effect with Implementation Lag
‹#›
Consumption and Investment
‹#›
Consumption and Investment
‹#›
Scenario (3): Zero Bound and
Monetary-Fiscal Policy Interaction
 So far interest rates are set according to
Gerdesmeier-Roffia (ECB) rule used also in
Kuester-Wieland (JEEA, June, 2010)
 Recession, nominal rate bounded at zero, central
bank want to lower further  notional negative
interest rate target  as long as it is below zero,
fiscal stimulus boosts GDP without triggering policy
tightening.
‹#›
Zero Interest Rate Floor
 Cash is a zero-interest bearing asset
 Technically:
 nonlinear RE model
 rational anticipation that zero interest floor
becomes binding for 1 year
simulation: keep interest rate constant for 1
year, then return to Gerdesmeier-Roffia
rule.
Endogenous recession and zero bound (see
example later).
‹#›
Zero Bound effective in 2009
‹#›
Decision Criteria: Robustness under
Model Uncertainty
 How to weigh the models?
 Bayesian probabilities, equal weights,
Minimax, ambiguity aversion (see also
Kuester-Wieland 2010)
 Here a simple criterion:
 cumulative 2-year or 4-year multiplier
negative or positive?
‹#›
Cumulative Effect
(output net of government spending)
 Crowding out dominates.
‹#›
3. Spillover Effects
 German stimulus greater, almost like ARRA
for 2 years.
 Do the other countries benefit, as
supporters of fiscal stimulus predicted?
 Need at least a 3-country model. We‘ll use
Taylor (1993), G-7.
 Impact of German stimulus on France and
Italy
‹#›
Spillovers of German Package
 No spillover, or even negative.
 What effects are present?
‹#›
Spillovers of German Package
 3 Effects:
Direct positive demand effect on Italian and
French exports. Significant in empirical
export demand equation.
Indirect negative effect 1: upward pressure
on euro area interest rates.
Indirect negative effect 2: euro real
appreciation.
‹#›
Counterfactual: Flexible Exchange Rates
between Germany, France and Italy
 Positive spillover once exchange rate effect
reinforces it.
‹#›
Conclusions
 Myth: „German package small“
 Spending multipliers: Confirms US analysis
with multiple New Keynesian models,
 Concentrated in 2 years, slightly greater
multipliers.
 Implementation lags mean effect in 09 Q1-2
negative.
 1-year constant rate increases multiplier little.
 2-4 years, signficant crowding-out.
 Intra-euro area direct-demand spillovers
dominated by negative interest-rate and
exchange-rate effects.
‹#›
Appendix 1: What if ARRA applied in € Area?
‹#›
Counterfactual - What if ARRA
implemented in Euro Area?
‹#›
Appendix 2: Recession and zero bound
 Baseline scenario:
Simulate SW model with actual US data up
to and including 2009:Q1.
 then project forward from 09:Q1 onwards
with and without fiscal stimulus.
Compute difference.
with Fed following Taylor rule, the zero
bound is not binding when simulating the
SW 07 model. Use SW rule instead.
‹#›
09:Q1 Deep recession and zero bound
 Counterfactual:
 consider deeper recessions.
1.5x the 09:Q1 shock, and 2x the 09:Q1
shock.
The first scenario is similar to CCTW 1 year
constrained.
‹#›
‹#›
‹#›
Projection and Realization: The
Beginning of the Recovery
 09:Q1:
Blue dashed: with
ARRA spending.
Black: no ARRA
spending
Red dashed: Actual
Realizations Q2-Q4
‹#›
‹#›
‹#›
‹#›
‹#›
‹#›
‹#›
Appendix 3: Government Spending Shocks:
Evidence from Reduced-Form Regressions
 Ramey (2009) says:
“literature remains divided on central
questions such as whether the GDP effect
is greater than unity”
“studies using VAR techniques in which
identification is achieved by assuming that
government spending is pre-determined
within the quarter typically find a larger
effect of government spending on GDP,
while studies using the Ramey-Shapiro “war
dates” indicate a smaller GDP effect and
‹#›
crowding-out of consumption.”
Range of Estimates
 VAR-based estimates:
 Gali et al (2007), up to 1.7
 Mountford & Uhlig (2009), below unity
 Other approaches to identification:
 Ramey (2009), 0.6-0.8 on defense
spending when WWII is excluded, with
WWII closer to unity.
 Barro & Redlick (2009), 0.6-0.7, defense
spending including WWII, closer to 1 if
unempoyment > 12%, non-defense
spending multiplier insignificant.
‹#›
Anticipations Matter.
‹#›
‹#›
Appendix 4: Eine Idee zur Exitstrategie
Ein Konsumstimulus der etwas anderen Art.
Problem bisher: Staatsausgabenerhöhung
verdrängt privaten Konsum und
Investitionen.
Alternative: Ankündigung eines zukünftigen
Sparpakets.
Beispiel – USA American Savings Pact
(ASPA = -2.5 x ARRA start in 2010:Q3)
‹#›
Gespart: 2.5 mal Ausgaben unter ARRA
ARRA versus ASPA ("American Stability PAct")
SW model (2007), Taylor rule after 1-year constant interest rate
1,2
ARRA Spending
0,7
GDP under ARRA
GDP under ASPA
0,2
-0,3
-0,8
2016q1
2015q3
2015q1
2014q3
2014q1
2013q3
2013q1
2012q3
2012q1
2011q3
2011q1
2010q1
Quarters
2009q3
-1,8
2010q3
-1,3
2009q1
Percent of GDP
ASPA Saving
‹#›
Consumption: ARRA versus ASPA
SW model (2007), Taylor rule after 1-year constant interest rate
Consumption under ASPA
Consumption under ARRA
2016q1
2015q3
2015q1
2014q3
2014q1
2013q3
2013q1
2012q3
2012q1
2011q3
2011q1
2010q1
Quarters
2009q3
-0,4
-0,6
2010q3
0,8
0,6
0,4
0,2
0
-0,2
2009q1
Percent of GDP
1,2
1
‹#›
Appendix 5: Ein Blick auf die Daten
 Was hat sich zwischen 09:Q1 und 09:Q2
geändert?
 Können wir daraus Schlüsse ziehen,
bezüglich der Wirkung verschiedener
Maßnahmen?
‹#›
U.S.A.: Quartalswachstum
(siehe auch Wall Street Journal (U.S.), 17.9.09, Cogan, Taylor, Wieland.)
Beiträge zu Quartalswachstumraten (annualisiert)
BIP
2009 Q1
2009 Q2
-6,4
-1
0,44
-0,69
-1,13
-8,98
-3,2
5,78
Bauten (nicht Wohn.)
-2,28
-0,59
1,69
Ausruestungsinv.
-3,01
-0,56
2,45
Lagerhaltungsaender.
-2,36
-1,39
0,97
Wohnbauten
-1,33
-0,66
0,67
2,64
1,6
-1,04
Staatsausgaben
-0,52
1,27
1,79
Verteidigung
-0,27
0,67
0,94
Nichtverteidig. (Bund)
-0,06
0,15
0,21
Ausgaben (Staaten,Lokal)
-0,19
0,44
0,63
Konsum
Investitionen
Nettoexporte
Differenz
5,4
‹#›
€-Zone: Quartalswachstum
(nicht annualisiert)
Beiträge zur Quartalswachstumsrate in
Vorjahrespreisen
08 Q3
08 Q4
09 Q1
09 Q2
Differenz
09 Q1 vs Q2
BIP Insgsamt
-0,4
-1,8
-2,5
-0,1
Inländische Nachfrage
0,17
-0,7
-2,05
-0,79
Investition (ohne Lagerh.)
-0,29
-0,74
-1,12
-0,26
0,86
Privater Verbrauch
-0,01
-0,25
-0,29
0,1
0,39
0,1
0,12
0,14
0,1
-0,04
0,38
0,18
-0,77
-0,72
0,05
-0,52
-1,11
-0,41
0,67
Staatlicher Verbrauch
Lagerhaltungsänderung
Nettoexporte
2,4
1,26
1,08
‹#›
Beitrag zum BIP Wachstum
(Quartalsraten, annualisiert)
Deutschland
08 Q2
08 Q3
08 Q4
09 Q1
09 Q2
Bruttoinlandsprodukt
-2,3
-1,3
-9,8
-14,1
1,3
15,4
Konsum
-0,9
0,8
-0,5
2,2
2,0
-0,2
-1,4
0,7
-1,0
1,5
1,7
0,2
0,5
0,1
0,4
0,8
0,3
-0,4
-4,1
3,5
-0,9
-6,0
-6,9
-0,9
Anlageinvestitionen
-1,8
0,5
-1,5
-5,9
0,6
6,4
Ausrüstungsinv.
-0,3
0,5
-1,2
-5,9
-0,1
5,8
Bauinvestitionen
-1,7
-0,2
-0,4
0,1
0,6
0,5
Sonst. Anlagen
0,1
0,1
0,1
0,0
0,1
0,1
Vorratsveränderungen
-2,2
3,1
0,7
-0,2
-7,5
-7,3
0,3
-0,2
-15,3
-18,8
-2,0
16,8
1,0
-0,9
-14,5
-19,5
-1,4
18,1
-0,7
0,7
-0,7
0,7
-0,6
-1,3
-2,4
5,5
-7,1
-8,4
-7,8
0,6
-2,7
5,8
-6,4
-8,8
-6,8
2,0
0,3
-0,3
-0,7
0,4
-1,0
-1,4
‹#›
Privater Konsum
Staat. Konsum
Investitionen
Exporte
Waren
Dienstleistungen
Importe
Waren
Dienstleistungen
2009 Q2-Q1
Deutschland: Investitionen (Staat vs Privat)
Beitrag zum BIP Wachstum
(Quartalsraten, annualisiert)
08 Q2
08 Q3
08 Q4
09 Q1
09 Q2
-4,1
3,5
-0,9
-6,0
-6,9
-0,9
-1,8
0,5
-1,5
-5,9
0,6
6,4
-0,3
0,5
-1,2
-5,9
-0,1
5,8
0,0
-0,1
0,0
0,1
0,0
-0,1
-0,3
0,6
-1,2
-6,0
-0,1
5,9
0,4
-0,6
0,0
Fahrzeuge
-0,6
1,1
-1,3
Bauinvestitionen
-1,7
-0,2
-0,4
0,1
0,6
Bau Staat
-0,3
0,1
0,2
-0,2
0,2
0,4
Bau Nichtstaat Wohn.
-1,1
-0,2
0,0
0,3
-0,1
-0,3
Bau Nichtst. Nichtw.
-0,4
0,0
-0,5
0,1
0,5
0,4
0,1
0,1
0,1
0,0
0,1
Sonst.Anlag. Staat
0,0
0,0
0,0
0,0
0,0
0,0
S.Anlag. nichtStaat
0,1
0,1
0,1
0,0
0,1
0,1
-2,2
3,1
0,7
-0,2
-7,5
Investitionen
Anlageinvestitionen
Ausrüstungsinv.
Ausrüst. Staat
Ausrüst Nicht-Staat
Maschinen und Gerät
Sonst. Anlagen
Vorratsveränderungen
Differenz
0,5
0,1
-7,3
‹#›