Too Many Cooks? Committees in Monetary Policy Helge Berger and

Too Many Cooks? Committees in
Monetary Policy
Helge Berger and Volker Nitsch
Discussion by Christopher Crowe, IMF
Research Department* for Conference on
Central Bank Communication, DecisionMaking and Governance, Waterloo University
April 28-29 2009
* Does not necessarily reflect the views of the IMF, its management or Executive Board.
Summary of the paper
• Paper’s key contribution is a new dataset
covering >30 countries for up to 40 years
– De facto as well as de jure committee size for
CB MPCs
– Composition of MPC (industrial, regional,
institutional background)
• Principal Use of the Data: is there an
optimal committee size for inflation
control?
Main Results
• U-shaped relationship for inflation: minimized at
8-10 committee members.
– Similar U-shaped relationship for inflation volatility.
– Inverse-U for output growth.
• Higher MPC member turnover associated with
lower inflation (opposite of result with governor
turnover).
• Composition of MPC (e.g. government
representation) has little impact.
My comments
• Very interesting paper on an important topic.
• The dataset could be used to shed light on
several interesting questions relating to
committees, decision-making and central bank
governance…
• …but I’m not wholly convinced that the paper’s
main question is the right one to ask with this
data.
• Plus a few other comments.
Are the main results convincing?
• Why should committee size affect the level of inflation?
– Committee too large: cannot communicate easily; subject to
factionalism or defense of own interests.
– Committee too small: cannot benefit from lower volatility of
mean decision.
– But why higher inflation rather than more volatile inflation or
inflation that’s harder to predict?
– Any plausible model of the inflation bias with committee size as
argument? Or of output growth?
• Omitted variable bias likely large compared to any
genuine effect.
– Easy to think of omitted variables:
• “Institutions” (effectiveness of government sector)
• Heterogeneity (has macro implications, might require large
committee)
• “Culture” (shapes views on decision-making and individual
responsibility, also macro policy and outcomes).
• Regressions have only a minimum set of reasonable controls.
What could be driving the results?
• Inflation bias can be caused by inappropriate
output gap target
– E.g. because government is going for short run
growth
– CBI as remedy: devolve to conservative CB’er or one
with appropriate objectives
• If a government wanted to weaken CBI:
– Have a small committee (less political cover for
tough decisions; easier to replace with more
compliant candidates)
– Have a large committee (ineffective talking shop;
agenda-setting by compliant governor/chairman).
– i.e. committee of reasonable size more
independent.
• But committee size not ultimate causal factor.
What other questions are there?
• Inflation and output volatility
– (paper already looks at variance of inflation)
• Inflation and output predictability (look at
variance of forecast errors)
• Combine with voting behavior (more limited
sample):
– Do larger committees vote differently (e.g. agendasetting by chair?)
– Do different types of MPC member vote differently?
(e.g. Besley, Meads and Surico, 2007; Meade and
Sheets, 2005).
• MPC member turnover (some interesting results
in the paper, I would give these greater
prominence).
Some other comments
• The baseline results could be made more convincing:
– Cluster SE’s by country to capture country-specific patterns of
serial correlation
– Additional controls:
•
•
•
•
measures of governance
Political turnover measures
De jure CBI
Output gap rather than growth minus avg. growth
• Use inflation tax transform π/(1+ π) to reduce impact of
outliers
• It would also be useful to have summary statistics for the
main variables.