Transcript Ehrmann - Viessmann European Research Centre

Politics and Monetary Policy
Michael Ehrmann and Marcel Fratzscher
European Central Bank
Conference on Central Bank Communication,
Decision-Making and Governance
Waterloo, 28/29 April 2009
The usual disclaimers apply.
1
What the paper is about
“We have continuously asked and asked the European Central
Bank to review its rates to reduce the euro-dollar exchange level.
If this does not happen, I think we should aim to have a political
committee with government representatives to work alongside the
central bank”
2
What the paper is about
“We have continuously asked and asked the European Central
Bank to review its rates to reduce the euro-dollar exchange level.
If this does not happen, I think we should aim to have a political
committee with government representatives to work alongside the
central bank”
Silvio Berlusconi, 08 June 2004
3
What the paper is about
“We have continuously asked and asked the European Central
Bank to review its rates to reduce the euro-dollar exchange level.
If this does not happen, I think we should aim to have a political
committee with government representatives to work alongside the
central bank”
Silvio Berlusconi, 08 June 2004
Background:
– Exchange rate at 1.15, policy rates at 2%
– European elections on June 13, 2004
– Voting support for Forza Italia had just dropped by 5.8%, from a high of
33% in 2002Q1 to 23% in 2004Q1
– GDP growth, 2004Q1: 0.8% in Italy, 1.7% in euro area
4
Introduction
• Kydland and Prescott (1977), Barro and Gordon (1983):
inflationary bias, time inconsistency if monetary policy run by
governments
• Rogoff (1985): appoint a conservative central banker, with greater
weight on price stability than government
• Alesina and Tabellini (2007, 2008): socially optimal vs actual
delegation of tasks to bureaucrats
– Socially optimal to delegate tasks with time inconsistent preferences
– Actual delegation of tasks that are exposed to risk: shift blame
• Given these incentives, politicians are likely to put pressure on
independent central banks, or criticise their actions
5
Introduction
• This paper analyses the political controversies for the ECB
– Is the ECB conservative in Rogoff’s sense?
– Additional disagreement due to different constituencies in a monetary
union?
– Are politicians’ preferences time varying?
• Database containing politicians’ statements about the ECB’s
monetary policy, covering 9 years and heads of governments plus
all ministers of 12 countries
6
Results
• Politicians’ preferred interest rates substantially lower than actual
rates
• ECB is conservative in Rogoff’s sense, i.e. politicians would like to
put more emphasis on growth
• Bulk of controversies arises due to different constituencies
• Time variations matter: preference shift
– If national economic performance is weak
– If the public has generally little trust in the ECB
– Depending on the affiliation of parties in government
7
Overview
• Literature review
• A model of politicians’ preferences
• Data, stylized facts
• What weights on price stability and growth?
• Time variations in politicians’ preferences
• Politicians’ desired interest rates
• Conclusions
8
Literature review
• Normative aspects of delegation of tasks
– Intrinsic motivation differences between politicians and bureaucrats:
Dewatripont, Jewitt and Tirole (1999), Maskin and Tirole (2001), Besley
and Coate (2003), Schultz (2008)
– Socially optimal to delegate if time inconsistency and short-termism are an
issue, or if tasks are technically complex; no delegation in case of
uncertainty about social preferences, or if compensation of losers is
important (Alesina and Tabellini 2007)
• Positive aspects of delegation of tasks (Alesina and Tabellini
2008; Epstein and O’Halloran 1999)
– Politicians decide on delegation, given their incentive structure
– Delegate risky tasks, attempt to shift blame with adverse outcomes, claim
responsibility with positive outcomes
9
Literature review
• Empirical literature
– Is the Fed influenced by Congress? Role of monetary policy in political
business cycles
• Yes: Alesina and Sachs (1988), Grier (1993), Abrams and Iossifov (2006),
Hellerstein 2007)
• No: Wooley (1984), Faust and Irons (1999)
– Measurement of political pressure and its effects
• Havrilesky (1993): number of reports in Wall Street Journal about politicians
arguing for more or less restrictive monetary policy; affects actual interest rates
• Maier, Sturm and de Haan (2002): no effect for Bundesbank
• Gersl (2006): no effect for Czech National Bank
• Maier and Bezoen (2004): index for ECB, until 02/2002
10
A model of politicians’ preferences
• Based on politicians’ statements
– 20 October 2006: “German Economics Minister Michael Glos said Friday
that the European Central Bank is in no hurry to raise interest rates from a
German perspective, pointing to the country's low inflation rates.” Source:
Dow Jones International News
– 01 February 1999: “German Finance Minister Oskar Lafontaine again
hinted Monday at the desirability of an ECB rate cut to stimulate growth
and employment in the euro-zone.” Source: Market News International
– 27 February 2007: “Austrian Chancellor Alfred Gusenbauer said the
European Central Bank (ECB) might have to adapt its interest rate policy
to match the current economic situation.” Source: Reuters News
11
A model of politicians’ preferences
• Preferred interest rates a function of inflation and growth in the
euro area and the national economies
itP   P   P , EA tEA   P ,C ( tC   tEA )   P , EA y tEA   P ,C ( y tC  y tEA )   tP
 P, EA  0;  P,C  0;  P, EA  0;  P,C  0
12
A model of politicians’ preferences
• Preferred interest rates a function of inflation and growth in the
euro area and the national economies
itP   P   P , EA tEA   P ,C ( tC   tEA )   P , EA y tEA   P ,C ( y tC  y tEA )   tP
 P, EA  0;  P,C  0;  P, EA  0;  P,C  0
• This need not be the actual interest rate; abstracting from a
possible monetary policy rule, we model
itact   act   act  tEA   act y tEA   tact
 act  0;  act  0
13
A model of politicians’ preferences
• Preferred interest rates a function of inflation and growth in the
euro area and the national economies
itP   P   P , EA tEA   P ,C ( tC   tEA )   P , EA y tEA   P ,C ( y tC  y tEA )   tP
 P, EA  0;  P,C  0;  P, EA  0;  P,C  0
• This need not be the actual interest rate; abstracting from a
possible monetary policy rule, we model
itact   act   act  tEA   act y tEA   tact
 act  0;  act  0
• Rogoff tells us that
 act  P , EA
 P , EA
act


14
A model of politicians’ preferences
• The interest rate gap is then
itact  itP  ( act   P )  (  act   P , EA ) tEA   P,C ( tC   tEA )
 ( act   P , EA ) ytEA   P ,C ( ytC  ytEA )   t
15
A model of politicians’ preferences
• The interest rate gap is then
itact  itP  ( act   P )  (  act   P , EA ) tEA   P,C ( tC   tEA )
 ( act   P , EA ) ytEA   P ,C ( ytC  ytEA )   t
– Rogoff implies
 act   P, EA  0 and / or  act   P, EA  0
– Additional weight on national variables implies
 P ,C  0;  P,C  0
16
A model of politicians’ preferences
• The interest rate gap is then
itact  itP  ( act   P )  (  act   P , EA ) tEA   P,C ( tC   tEA )
 ( act   P , EA ) ytEA   P ,C ( ytC  ytEA )   t
– Rogoff implies
 act   P, EA  0 and / or  act   P, EA  0
– Additional weight on national variables implies
 P ,C  0;  P,C  0
• Need to run a proxy regression due to unobservability of
preferred interest rate
– assume that political pressure is increasing in the gap
ppt  a  b1 tEA  b 2 ( tC   tEA )  c1 y tEA  c 2 ( y tC  y tEA )   t
17
A model of politicians’ preferences
• Time variations: preference shift in periods xt
itP   P   P , EA tEA   P ,C ( tC   tEA )   P , EA, x tEA xt   P ,C , x ( tC   tEA ) xt
  P , EA ytEA   P,C ( ytC  ytEA )   P , EA, x ytEA xt   P ,C , x ( ytC  ytEA ) xt  xt   tP
18
A model of politicians’ preferences
• Time variations: preference shift in periods xt
itP   P   P , EA tEA   P ,C ( tC   tEA )   P , EA, x tEA xt   P ,C , x ( tC   tEA ) xt
  P , EA ytEA   P,C ( ytC  ytEA )   P , EA, x ytEA xt   P ,C , x ( ytC  ytEA ) xt  xt   tP
– Preference shift towards growth, away from inflation
 P , EA, x  0;  P , EA, x  0
– Preference shift towards national variables
 P ,C , x  0
– What dominates for national inflation?
 P ,C , x  ?
19
A model of politicians’ preferences
• Time variations: preference shift in periods xt
itP   P   P , EA tEA   P ,C ( tC   tEA )   P , EA, x tEA xt   P ,C , x ( tC   tEA ) xt
  P , EA ytEA   P,C ( ytC  ytEA )   P , EA, x ytEA xt   P ,C , x ( ytC  ytEA ) xt  xt   tP
– Preference shift towards growth, away from inflation
 P , EA, x  0;  P , EA, x  0
– Preference shift towards national variables
 P ,C , x  0
– What dominates for national inflation?
 P ,C , x  ?
• Again: estimated by means of proxy regression
20
Data, stylized facts: political pressure
• Extract politicians’ statements
– Factiva (14,000 sources)
– 1999-2007
– Heads of government, ministers of 12 euro area countries
– Avoid double counting (in contrast to Havrilesky)
• Classify and code politicians’ statements
– Statements about future interest rate decisions and those about mandate,
independence, past decisions, exchange rate
– Code as
Ct
 1

  0
 1

call for lower rates
neutral statement
call for higher rates
21
Data, stylized facts: political pressure
• Issues
– Search in English; consistency check with other languages gave few
differences (newswires important!)
– Exclusion of local government officials, opposition parties, lobby groups,
trade unions, international institutions
– Classification and coding subjective
• Done independently according to content analysis (Holsti 1969)
• Rather plain language, coding comparatively uncontroversial
– An independent role of the media?
• Aggregation into 299 quarterly observations per political party,
ranging from -1 to 11
22
Data, stylized facts: political pressure
Comments on future interest rate decisions
Lower rates
Country
Total
Austria
Belgium
Spain
Finland
France
Germany
Greece
Ireland
Italy
Luxembourg
Netherlands
Portugal
Neutral
28
42
22
1
83
92
4
2
24
11
1
7
317
Higher rates
2
4
7
3
7
8
5
1
2
8
4
0
51
0
0
8
0
0
2
0
2
1
0
3
0
16
Total
30
46
37
4
90
102
9
5
27
19
8
7
384
– Strong asymmetry, only 5% of statements calling for higher rates
– Considerable cross-country variation
– National performance seems to matter, e.g. Spain
23
2
0
5
3
10
interest rate
4
15
5
20
Data, stylized facts: political pressure
1999q3
2001q3
2003q3
.
political pressure
2005q3
2007q3
interest rate
– Time variations; high pressure in 2001-2003, 2005/2006
– Discussion since French presidential election campaign reflected in
comments on mandate
24
0
5
10
15
Data, stylized facts: political pressure
1999q3
2001q3
2003q3
.
high GDP
2005q3
2007q3
low GDP
– Economic performance matters: more pressure from low-growth countries
25
Data: political economy variables
• Periods when politicians might shift preferences towards growth,
or more likely to pressure ECB
– Pre-election period (Drazen 2000); dummy for election quarter
– Low voting support; rate of decline of voting support
– Low public trust in ECB; dummy equal one if share of respondents in
Eurobarometer survey mistrusting the ECB is above sample mean
– Excessive deficit procedure; dummy when procedure in place
– Negative growth differential (Alesina and Cukierman 1990, Rogoff and
Sibert 1988); dummy if national growth below euro area
– Left-wing party (Hibbs 1977, Powell and Whitten 1993, Swank 1993);
dummy based on Chapel Hill Party Dataset 2002
26
What weights on price stability and growth?
Hypothesis
Benchmark
All
comments
Pre-2007
No fixed
effects
Ordered
probit
Negative
binomial
Euro area macro variables:
 t EA
b1>0
-0.526
-0.507
-0.454
-0.373
-0.309
-0.460*
y t EA
c1<0
-0.143*
-0.099
-0.184**
-0.173**
-0.151**
-0.186***
Country-specific macro differences:
-0.444***
 t C  t EA
b2<0
-0.512***
-0.248
-0.482***
-0.386***
-0.476***
-0.105**
-0.081
-0.091***
-0.043*
-0.161**
y t C -y t EA
c2<0
Party fixed effects
Observations
R-squared
-0.073*
Yes
299
0.27
Yes
299
0.28
Yes
257
0.29
No
299
0.12
Yes
299
0.14
Yes
299
ppz ,t  az  b1 tEA  b2 ( zC,t   tEA )  c1 ytEA  c 2 ( yzC,t  ytEA )   z ,t
– Larger desired weight on GDP growth
– Larger desired focus on national economy
– Robust to adding unemployment or the exchange rate
27
What weights on price stability and growth?
Hypothesis
Benchmark All comments
No fixed
effects
Ordered
probit
Incidence of
comments
Negative
binomial
Euro area macro variables:
 t EA
b1>0
EA
c <0
-0.526
-0.507
-0.373
-0.309
-0.157
-0.460*
-0.143*
-0.099
-0.173**
-0.151**
-0.211***
-0.186***
Country-specific macro differences:
2
-0.444***
b <0
 t C  t EA
-0.512***
-0.482***
-0.386***
-0.326*
-0.476***
-0.105**
-0.091***
-0.043*
-0.023
-0.161**
yt
1
y t C -y t EA
c2<0
Party fixed effects
Observations
R-squared
-0.073*
Yes
299
0.27
Yes
299
0.28
No
299
0.12
Yes
299
0.14
Yes
299
0.24
Yes
299
ppz ,t  az  b1 tEA  b2 ( zC,t   tEA )  c1 ytEA  c 2 ( yzC,t  ytEA )   z ,t
– Larger desired weight on GDP growth
– Larger desired focus on national economy
– Robust to adding unemployment or the exchange rate
28
Time variations
Hypothesis
Pre-election
period
Low voting Low public
support trust in ECB
Political economy variables & interaction terms:
1.132**
 t EA * x t
b3>0
Excessive
deficit
procedure
Negative
growth
differential
Left-wing
party
-0.016
-0.550
-0.094
-0.350
0.018
y t EA * x t
c3<0
-0.332
0.001
-0.080
-0.518*
-0.062
-0.148
( t C  t EA ) * x t
b4=?
0.290
-0.005
0.038
0.016
-0.507*
0.216
(y t C -y t EA ) * x t
c4<0
0.139**
0.005
-0.344**
-0.175
-0.235*
-0.432**
d=0
-1.874
0.035
1.203
1.061
1.553
-0.095
xt
Party fixed effects
Observations
R-squared
Yes
299
0.27
Yes
299
0.27
Yes
299
0.30
Yes
299
0.28
Yes
299
0.31
Yes
299
0.35
ppz ,t  az  b1 tEA  b2 ( zC,t   tEA )  b3 tEA xz ,t  b4 ( zC,t   tEA ) xt  c1 ytEA  c 2 ( yzC,t  ytEA )  c3 ytEA xz ,t  c 4 ( yzC,t  ytEA ) xz ,t  dxz ,t   z ,t
– Preference shift
• If national economy performs poorly
• If the public has generally little trust in the ECB
• Depending on affiliation of parties in government
29
Time variations, robustness: negative binomial
Hypothesis
Pre-election
period
Low voting Low public
support trust in ECB
Excessive
deficit
procedure
Negative
growth
differential
Left-wing
party
Political economy variables & interaction terms:
1.927**
 t EA * x t
b3>0
0.007
-0.410
0.310
-0.326
0.016
y t EA * x t
c3<0
-0.461*
0.002
0.047
-0.225
0.081
-0.020
( t C  t EA ) * x t
b4=?
0.190
0.005
0.048
0.044
-0.235
0.182
(y t C -y t EA ) * x t
c4<0
0.268*
0.007
-0.307***
0.025
-0.370*
-0.267**
d=0
-3.562*
-0.013
0.706
-0.301
1.021
-0.170
xt
Party fixed effects
Observations
Yes
299
Yes
299
Yes
299
Yes
299
Yes
299
Yes
299
ppz ,t  az  b1 tEA  b2 ( zC,t   tEA )  b3 tEA xz ,t  b4 ( zC,t   tEA ) xt  c1 ytEA  c 2 ( yzC,t  ytEA )  c3 ytEA xz ,t  c 4 ( yzC,t  ytEA ) xz ,t  dxz ,t   z ,t
– Broadly robust to using only calls for lower rates, all comments, country
fixed effects, ordered probit model
30
Politicians’ desired interest rates
• Extracting politicians’ preferred interest rates
– Assume
itP  itact if ppt  0
– Estimate preferred rate for all neutral speakers
itact   zP   P, EA tEA   P,C ( zC,t   tEA )   P, EA ytEA   P,C ( yztC  ytEA )   zP,t
– Use estimated parameters to predict preferred interest rates for the nonneutral speakers (assuming model constancy)
31
Politicians’ desired interest rates
Hypothesis
Benchmark
Euro area macro variables:
 t EA
 P,EA>0
0.518***
y t EA
P,EA>0
0.437***
Country-specific macro differences:
 t C  t EA
 P,C>0
0.327***
y t C -y t EA
P,C>0
-0.007
0.759*
Const.
Party fixed effects
Observations
R-squared
Yes
131
0.610
itact   zP   P, EA tEA   P,C ( zC,t   tEA )   P, EA ytEA   P,C ( yztC  ytEA )   zP,t
32
Politicians’ desired interest rates
• Preferred interest rates rise by
– 50 basis points in response to a 1% increase in euro area inflation
– 40 basis points in response to a 1% increase in euro area growth
– 30 basis points in response to a 1% increase in domestic inflation
33
Politicians’ desired interest rates
• Plot for the preferred interest rate of the average politician
1
2
3
4
5
– Mostly desire for more accomodation, partially substantial gaps
1999q3
2001q3
2003q3
2005q3
2007q3
.
actual
preferred
implied
34
Conclusions
• Politicians’ preferred interest rates substantially lower than actual
rates
• ECB puts more emphasis on price stability than politicians, i.e. is
a Rogoff-type of central bank
• Lots of controversy arises due to different constituencies
• Time variations in politicians’ preferences
• Independence of ECB important to shield it from
– Time varying demands by politicians
– National concerns of politicians
35