Transcript Document

Convergence of Government
Bond Yields in the Euro Zone:
The Role of Policy Harmonization
Denise Côté and Christopher Graham
International Department
28 April 2006
Motivation
• Since the early 1980s, long-term government
bond yields in the euro zone have declined, in
line with those in other industrialized countries
• By the time the euro was introduced in January
1999, bond yields across the euro zone countries
had largely converged to that of Germany (the
euro zone's largest economy)
Motivation
Why does convergence of national yields to a
stable level with reduced risk matter?
• Reduces various risks => less uncertainty
regarding the value of funds over time
• Cheaper access to debt financing
• Contributes to financial stability
• Stimulates investment and output within
converging countries
Motivation
• What factors can drive convergence of long-term
bond yields to a stable level with reduced risk and
maintain it over the long term?
• One explanation, by the ECB:
– Convergence driven by anticipation of the introduction
of the euro and corresponding elimination of exchange
rate risk (ECB Monthly Bulletin November 2003)
Motivation
Our goal:
• To examine how monetary and fiscal
policies adopted on the path to EMU,
including the introduction of the euro,
contributed to the convergence of national
long-term government bond yields in the
euro zone
Outline
1. Institutional Background
2. Our Approach
3. Empirical Analysis
• Panel analysis
• Currency risk
4. Conclusions
5. Policy Implications
Institutional Background
•
Maastricht Convergence Criteria (1993):
1) General government deficit to GDP ≤ 3%
2) Gross general government debt to GDP ≤ 60%
3) Inflation ≤ 1.5 ppt above average of 3 best
performing countries in terms of price stability
4) Long-term interest rate ≤ 2 ppt above average of 3
best performing countries in terms of price stability
5) Exchange Rate Mechanism respected for at least
2 years prior to adoption of euro (±15%)
Our Approach
• Assess convergence of long-term
government bond yields resulting from:
• Increased harmonization of monetary and
fiscal policies on the road to EMU
• The introduction of the common currency itself
Analytical Approach
• Empirical Analysis:
• Cointegration and panel estimation techniques
applied over 1980Q1 - 2002Q4
• Estimate 10-year government bond yields using a set
of long-term determinants:
expected inflation, government balance,
government debt and world interest rate
• Apply to a pool of nine EMU countries
• Compare to 2 control-groups: EU3 (UK, Denmark
and Sweden) and OECD4 (Australia, Canada,
Norway and Switzerland)
Euro zone Estimates
Parameter
German
Yield as
World Yield
US Yield As
World Yield
Expected
Inflation
0.83
0.83
Fiscal
Balance
(% of GDP)
-0.18
-0.17
World
Interest Rate
0.72
0.23
Adjustment
Speed (ECM)
-0.080
(6.09)
-0.060
(5.00)
• All estimated parameters are of
the expected sign, except the
debt-to-GDP ratio
• Fiscal balance parameter
consistent with Orr, Edey &
Kennedy (1995), Brook (2003)
• Parameter on world interest rate
three times larger when real
German yield is used (Knot &
de Haan [1995])
• Speed of convergence to longrun equilibrium is faster when
using the German yield
EU3 Estimates
German
Yield as
World Yield
US Yield As
World Yield
Expected
Inflation
0.76
0.85
Fiscal Balance
(% of GDP)
-0.23
-0.11
Fiscal Debt
(% of GDP)
0.05
0.04
World
Interest Rate
0.82
0.36
Adjustment
Speed (ECM)
-0.059
(2.82)
-0.042
(2.01)
Parameter
• Parameters qualitatively the
same as for the euro zone
• Debt ratio is now of
expected sign
• Adjustment slower than for
euro zone countries
OECD4 Estimates
German
Yield as
World Yield
US Yield As
World Yield
Expected
Inflation
0.87
0.91
Fiscal Balance
(% of GDP)
-0.06
-0.05
World
Interest Rate
0.49
0. 33
Adjustment
Speed (ECM)
-0.048
(3.10)
-0.037
(2.48)
Parameter
• Parameters qualitatively similar
to those for euro zone and EU3
• Impact of fiscal balance
slightly reduced
• Debt ratio not significant
• Adjustment slower than for
euro zone, but similar to EU3
Summary of Panel Results
Summary of results for euro zone, EU3 and OECD4
• In the long-run, 10-year bond yields driven by:
• Expected inflation
• Developments in larger country's bond yields
• To a lesser extent, effects of persistent changes
in general government fiscal balances (debt
ratio for EU3)
• Results robust (expected inflation, ECM lags & Y-gap)
Summary of Panel Results
Results suggest:
• Convergence driven by policy harmonization
(especially monetary policy)
• Not confined to members of the common
currency or common market
Currency Risk
• Plot two corporate bonds
– Same issuer  same default risk
– Different countries  different currency and
liquidity risks
• Currency risk declined gradually and had
essentially disappeared before adoption of euro
• Liquidity risk remains and is very small
Conclusions
• Harmonization of monetary and fiscal policies greatly
contributed to convergence of long-term government
bond yields in the euro zone by prompting
convergence of their long-run determinants
• Convergence in the euro zone cannot be attributed
primarily to the strict introduction of the euro, since
EU3 and OECD4 also experienced a similar
convergence
Policy Implications
• Long-term estimates imply decreasing trend for
euro zone yields
• Current average yield is below trend for
cyclical reasons
• Given current policy, trend should remain low
• Mitigates financial risk
Convergence of Government
Bond Yields in the Euro Area: The
Role of Policy Harmonization
Denise Côté and Christopher Graham
International Department
28 April 2006
Empirical Analysis
• General long-run specification, based on theory:
RLt = a1ecpit + a2gbalt + a3gdebtt + a4rrlwt + ut
(1)
ecpi = expected inflation
gbal = General government fiscal balance as % of
nominal GDP (“+”=surplus, “-”=deficit)
gdebt = Gross general government debt as % of nominal GDP
rrlw = U.S. or German 10-year government real bond
yield (measure of the world real yield)
Estimation of Panel Error Correction Model
Euro zone Countries
Sample: 1980Q1-2002Q4, 9 countries, 819 Observations
First Step: Estimates of Long-Run
Relationship
2nd Step: Estimation of Error-Correction
Models, Error-Correction Term (g)
.04 + .81ecpi - .21gbal - .01gdebt +.23rrlus
(18.54) (52.24) (10.26) (3.54)
(7.09)
-0.0606
(5.02)
.04 + .83ecpi - .17gbal +.23rrlus
(23.98) (61.87) (10.02) (7.25)
-0.0596
(5.00)
.02 + .81ecpi - .21gbal -.01gdebt +.70rrlgy
(6.83) (52.73) (10.66) (2.90) (13.74)
-0.0811
(6.04)
.02 + .83ecpi - .18gbal +.72rrlgy
(6.49) (64.09) (11.04) (14.08)
-0.0804
(6.09)
Estimation of Panel Error Correction Model
EU3 Countries
Sample: 1980Q1-2002Q4, 3 countries, 273 Observations
Estimates of Long-Run Relationship using
panel data
Error-Correction Term (g)
.01 + .85ecpi - .11gbal + .04gdebt +.36rrlus
(1.80) (22.16) (3.09) (3.01)
(5.80)
-0.0416
(2.01)
-.01 + .76ecpi - .23gbal +.05gdebt +.82rrlgy
(1.50) (20.07) (6.32) (4.70)
(8.24)
-0.0585
(2.82)
Estimation of Panel Error Correction Model
OECD4 Countries
Sample: 1980Q1-2002Q4, 4 countries, 364 Observations
Estimates of Long-Run Relationship using
panel data
Error-Correction Term (g)
.03 + .91ecpi - .05gbal + .01gdebt +.33rrlus
(6.97) (26.18) (1.97) (1.25)
(6.34)
-0.0371
(2.48)
.02 + .87ecpi - .06gbal +.01gdebt +.49rrlgy
(4.51) (24.65) (2.72) (1.01)
(5.83)
-0.0483
(3.10)