Full Text (Final Version , 39kb)

advertisement
Publicity of debate in committee decision-making: The role of the state of the world
Publicity of debate in committee decisionmaking: The influence of the state of the
world
Lars van der Ham
Erasmus University Rotterdam
Thesis supervisor: O.H. Swank
Abstract
This thesis analyses the behaviour of agents in committee decision-making when faced with a probability distribution of the state
of the world where each state is not as likely to occur as the other one. This thesis shows that when the prior probability of an
outcome of the state of the world is sufficiently high, agents will have an incentive to speak truthfully when deliberation is done
privately, but may vote on a policy action to take that is not in line with the outcome of the message round. When individual
messages and votes are made public, the existence of an informative equilibrium is only possible if the probability of the state of
the world does not deviate too much from a 50-50 per cent level, because a high prior probability of one of the two states results
in players voting in line with this prior, rather than conveying their received signals truthfully.
Introduction
In their paper Publicity of Debate and the Incentive to Dissent: evidence from the US federal reserve
Meade and Stasavage explore the potential costs of transparency in committee decision-making. They
do so by constructing a theoretical model, and by testing it empirically using an original dataset on
deliberations from the principal organ of United States national monetary policy, the Federal Reserve’s
Federal Open Market Committee (FOMC).
The theoretical model by Meade and Stasavage concerns a monetary committee consisting of three
committee members that have to decide on a policy action. The committee members both care about
reaching the correct policy decision and about the level of expertise they have in the eyes of an outside
audience. The correct action depends on the state of the economy, represented by an unobserved
binary state variable. For instance, it could capture whether the output is at or above its potential level.
In their model, Meade and Stasavage assume that each state of the economy is as likely to occur as
the other. Depending on how the state variable is defined, it is in some instances thinkable that this is
not the case. If an economy has had a long period of economic prosperity, it may be more likely that
the state variable indicates a positive state of the world than a negative state of the world. The same
goes for the difference in economic performance between countries. For a given country it may be more
likely to have economic prosperity than for another country.
This study aims to clarify if, and how behaviour of agents differs in a model where the probability of the
state of the world isn’t fixed at a 50-50 probability level, but rather has any other distribution, such as
10-90 or 60-40. The relevance of investigating this question has already been mentioned above. The
difference in cross-country economic performance but also economic performance over time makes it
1
Publicity of debate in committee decision-making: The role of the state of the world
relevant to research if the behaviour of committee members significantly differs when this assumption
is altered.
This thesis is organised as follows: Firstly, relevant literature with regards to committee decision
behaviour will be discussed. Secondly, the model will be constructed and the results under both private
and public deliberation will be analysed. Finally, the results will be discussed and final conclusions will
be made.
Literature review
In their 2008 paper, Meade and Stasavage explore the potential costs of transparency in committee
decision-making. By constructing a theoretical model, they arrive at a couple of propositions, which are
tested empirically using data from the Federal Reserve’s Federal Open Market Committee (FOMC).
Their conclusions are as follows: Firstly, when a known expert speaks first, and when the reputational
concerns of the committee members are sufficiently strong, no committee member will dissent from the
known expert’s opinion. Secondly, when a known expert speaks first it is more likely for committee
members to dissent from the known expert’s opinion if committee deliberation occurs in private. The
incentives of the committee members in this paper are twofold: Agents care about making the correct
policy decision as well as being thought of as having a high expertise in the eyes of an outsider.
In their 2010 paper, Swank, Swank and Visser address two questions: First, what information does a
decision-making committee voluntarily provide to support the adequacy of its decision? And secondly,
how does the committee react to outside pressure to provide more information? Swank, Swank and
Visser arrive at a couple of conclusions: Given that members exchange their information
simultaneously, and care both about reaching the correct policy decision and having a high expertise
in the eyes of an outsider, committee members tend to show a ‘united front’ when committee
deliberation occurs in private, meaning that committee members mimic the most likely opinion and thus
pretend to agree with each other. The reason for this is that disagreement among the committee
members signals a lack of competence. Showing a united front maximizes the expected reputational
payoff for the committee members. Secondly, they show that private decision-making may be inefficient.
The reason for this is that the committee can feign agreement by choosing to implement the policy
action. This is the case because deciding on taking an action can only be a result of agreement among
the committee members, whereas deciding on not implementing the policy action may also be a result
of disagreement.
The inefficiency of decision-making behind closed doors may be remedied by increasing transparency
of decision-making. By imposing that deliberation has to occur in public instead of behind closed doors,
the public can now assess the individual opinions and voting behaviour of the committee members.
Because we assume committee members speak simultaneously, the committee members have an
incentive to speak truthfully, as it increases the probability of reaching the correct policy action as a
committee. A drawback of public deliberation is the fact that committee members now may feel
exposed. Swank, Swank, and Visser (2008) consider that committee members now may organize pre2
Publicity of debate in committee decision-making: The role of the state of the world
meetings, as a costless means to circumvent the exposure imposed on the committee members. In the
short run, transparency may increase transparency. However, committee members will organize premeetings to enable themselves to speak privately, making the formal meetings more scripted.
Gersbach and Hahn (2004) examine the desirability of the publication of the vote transcripts of a
committee of central bankers in a situation where the committee members have different policy
preferences. They find that making the voting behaviour of the individual committee members public is
socially desirable when the appointment process of central bankers is based on considerations of
aggregate welfare. One reason for this is that when voting behaviour is made public, the committee
members can be individually held accountable for their decisions by governments, giving governments
the power to only reappoint central bankers that have the same interests as the public. Secondly, central
bankers with preferences that differ from the preferences of the public may be induced to vote for the
social optimal monetary policy, so they are reappointed and can be part of future decision-making.
Gersbach and Hahn find that central bankers that have the same preferences as the public prefer
transparent decision-making over opaque decision-making. On the other hand, central bankers with
preferences that differ from the preferences of the public prefer opacity over transparency, as opacity
increases the probability for them to be reappointed for a second period. Gersbach and Hahn argue
that their findings may be different in a multi-country setting, such as decision-making in the European
Central Bank. Gersbach and Hahn argue that transparency can cause nationalist voting behaviour,
thereby enabling national governments to reappoint nationalist central bankers, and hence increasing
the share of nationalist central bankers in the committee. Although transparency makes the appointment
process more efficient, nationalist central bankers do not have the incentive to maximize European
welfare, but rather national welfare, hence creating a loss of social welfare.
In their 2008 paper, Gersbach and Hahn consider a case where central bankers differ in their efficiency
to identify which policy decision would lead to a socially optimal outcome and examine the costs and
benefits of the publication of the individual voting behaviour of committee members. They argue that
the publication of voting behaviour is initially harmful to the efficiency of the decision-making, as less
efficient central bankers have an incentive to vote as to maximize their individual utility rather than
maximizing social utility. After reappointment of the central bankers, the social loss will be lower, as
governments can more easily identify the less efficient central bankers. Hence, accountability
increases. Overall, the negative effects of publishing voting transcripts dominate the positive effects.
Opacity is thus preferred over transparency.
A number of phenomena that exist as a result of the desire of committee members to come across as
able decision-makers are explained by Visser and Swank (2007). Their first finding is that committee
members tend to show a ‘united front’ when they care about their reputation. The reason for this is that
competent committee members would agree about the consequences of any policy decision, and so
disagreement signals incompetence. Secondly, they find that reputational concerns have a distorting
effect on the decision-making. This stems from the fact that the policy decision the committee takes has
an effect on the reputation of the individual committee members. A third phenomenon described by
Visser and Swank is the fact that committee members keep their private information to their selves
3
Publicity of debate in committee decision-making: The role of the state of the world
when their attempts to influence the decision process are frustrated as a result of a difference in the
extent committee members care about their reputations. They show that the committee members that
are most concerned with their reputation have the greatest incentive to overstate the positive effects
and to downplay negative effects.
Finally, Visser and Swank analyse the effects of the voting rule employed by the committee. They argue
that, when information cannot be manipulated, unanimity is the preferred voting rule, because the
person with the fewest reputational concerns is decisive in this case. When information can be
manipulated, the voting rule should balance the benefits from information exchange before the
committee members vote and the costs of making a member decisive whose interests are less aligned
with those of the organisation than the interest of some other member.
The influence of the voting rule in committee decision-making is also discussed by Levy (2007). Her
main result is that when decision-making is done behind closed doors, committee members tend to
conform to pre-existing biases. He also finds that when the voting rule is unbiased, committee members
tend to vote in line with the action that is favoured by the prior. Levy’s results imply that when committees
change their decision-making process from a secretive to a transparent one, they are more likely to
accept reforms or radical decisions. Finally, Levy shows that if the committee uses a particular voting
rule, a secretive decision-making process can be superior to an open decision-making process.
The model
The model used and analysed in this study builds upon the model created by Meade and Stasavage in
2008. In line with the model developed by Meade and Stasavage, the model assumes a three player
monetary policy committee, composed of players A, B and C. Each player has to decide on a binary
policy action π‘Ž ∈ {0,1}. This is in line with decisions actual monetary policy committees have to make,
i.e. deciding to change interest rates or keep them at a fixed level. Each player receives a binary signal
𝑠𝑖 about the state variable πœ” ∈ {0,1}. Contrary to the model used by Meade and Stasavage, this model
assumes the probability of the state variable to be as follows: Pr(πœ” = 0) = 𝛸 and Pr(πœ” = 1) = (1 − 𝛸).
This prior probability is known to all players. The optimal decision of the players is reached when the
action the committee takes corresponds to the outcome of the state of the world, in such a way that
π‘Ž = πœ”. A committee decision is taken when two or more players vote on the same action. The signals
received by the players are independent conditional on the state variable. Player A, being the chairman
of the committee, has a known level of expertise, and receives a signal about the state of the world π‘ π‘Ž .
1
Her signal is accurate with a probability Pr(𝑠𝐴 = πœ”) = πœ… = 𝑝 + (1 − 𝑝). Players B and C also receive a
2
1
signal 𝑠𝑖 . Their signal is accurate with a probability of Pr(𝑠𝐡,𝐢 = πœ”) = 𝜏 = π‘ž + (1 − π‘ž). This means the
2
players receive a signal that is always correct with probability 𝑝 or π‘ž, and that is correct 50 per cent of
times with probability (1 − 𝑝) or (1 − π‘ž). We assume that player A, as the chairman of the committee,
has a higher expertise than player B and C, and therefore assume that πœ… > 𝜏, . We assume that player
B and C together have more information than player A by herself. We also assume that player A
truthfully reveals her signal, so that π‘šπ΄ = 𝑠𝐴 .
4
Publicity of debate in committee decision-making: The role of the state of the world
π΄π‘ π‘ π‘’π‘šπ‘π‘‘π‘–π‘œπ‘› 1. (
𝜏 2
πœ…
) >
1−𝜏
1−κ
The utility of each player not only depends on making the correct policy action, but also depends on an
outsider’s ex post assessment of the player’s expertise. In practice, an outside observer could be a
future employer of the committee member, or more generally, anyone against who the player wants to
come across of as having a high level of expertise. The reputational payoff each player receives when
both the individual votes and the individual messages of the players are made public is defined as
follows: 𝛽 Pr(β„Ž|π‘Ž, π‘šπ‘– , 𝑣𝑖 , πœ”). The reputational payoff is scaled by 𝛽, where 𝛽 > 0. 𝛽 captures the degree
each player cares about its reputation. Given the above information, utility for each player is defined as
follows:
π‘ˆ = π‘Ž + 𝛽 Pr(β„Ž|π‘Ž, π‘šπ‘– , 𝑣𝑖 , πœ”)
if
πœ”=1
π‘ˆ = (1 − π‘Ž) + 𝛽 Pr(β„Ž|π‘Ž, π‘šπ‘– , 𝑣𝑖 , πœ”)
if
πœ”=0
Each player has an incentive to tell the truth with regards to reaching the correct policy decision, which
are the first terms in the utility functions above. It is however possible that the reputational concerns of
a player make the player decide to not reveal their given signal truthfully, but rather send a message
and vote in line with the messages and votes of the other committee members. The timing of the game
is as follows: Firstly, the state of the world πœ” is realized (but not observed) and the private signals of
the players 𝑠𝑖 are realized. We assume that the players know the prior probabilities 𝑋 and (1 − 𝑋). After
that, each player sends a message regarding the state of the world, π‘šπ‘– . A speaks first, followed by B,
followed by C. Following the order of the message round, each player then votes whether to undertake
a policy action π‘Ž. Finally, the state of the world becomes public and each player receives a payoff
determined by the policy action and the probability that she has a high expertise.
In what follows, we identify the possible informative equilibria of this game. An informative equilibrium
exists when players A and B send truthful messages, i.e. messages in line with their individual signals,
and player C sends a truthful message when truthfully revealing her signal means that other players
change their beliefs about which state is more likely to be correct.
Results under private deliberation
With each state as likely as the other, given that committee members care about reaching the correct
policy action and about an outsider’s assessment of the committee members’ expertise, all committee
members have an incentive to reveal their signal truthfully, because truthfully conveying a signal means
more information for the committee to base its decision on. The committee will decide on implementing
the policy action π‘Ž only when the posterior probability Pr(πœ” = 1) is higher than the posterior probability
Pr(πœ” = 0).
As individual messages and votes will not be made public, and because there is no trade-off between
reputational and policy incentives, the committee will act ‘collegial’. The reason for this is that the payoff
5
Publicity of debate in committee decision-making: The role of the state of the world
for each player depends on the outcome of the committee as a whole. Therefore, each player truthfully
conveys her signal to the other committee members in the message round.
Because the probability of the state of the world is not necessarily 50-50, the committee members have
to condition their votes not only on the signals conveyed in the message round, but also on the prior
probability of the state of the world. We analyse the following situation: Player A receives a signal that
the state of the world is equal to 0, and player B and C receive a signal that the state of the world is
equal to 1. In a situation where the probability of the state of the world is 50-50, it would now be more
likely that the state of the world is equal to one, because the accuracy of the signals of player B and C
together is higher than the accuracy of player A alone. In this case however, the probability that the
state of the world is equal to 1, given the signals of the players, depends on 𝑋 as well.
The posterior probabilities of πœ”, given the signals of the players are as follows:
Pr(πœ” = 1|𝑠𝐴 = 0, 𝑆𝐡,𝐢 = 1) =
Pr(πœ” = 0|𝑠𝐴 = 0, 𝑆𝐡,𝐢 = 1) =
1
(1 − 𝛸) (1 − 𝑝)𝜏 2
2
1
1
(1 − 𝛸) (1 − 𝑝)𝜏 2 + π›Έπœ… ( (1 − π‘ž))
2
2
1
π›Έπœ… ( (1 − π‘ž))
2
2
2
2
1
1
π›Έπœ… ( (1 − π‘ž)) + (1 − 𝛸) (1 − 𝑝)𝜏 2
2
2
The committee decides on the policy action to take given the probabilities above. Mathematically it
2
1
follows that Pr(πœ” = 1|𝑠𝐴 = 0, 𝑆𝐡,𝐢 = 1) > Pr(πœ” = 0|𝑠𝐴 , 𝑆𝐡,𝐢 = 1) if
1−𝑋
𝑋
>
πœ…(2(1−π‘ž))
1
(1−𝑝)𝜏2
2
. If the probability that
πœ” = 0 is sufficiently high, the expression above will not hold and the committee will vote for π‘Ž = 0, even
though player B and C received a signal 𝑠𝐡,𝐢 = 1, and knowing that the signal of player B and C together
is more accurate than the signal of player A. Given that 𝑠𝐴 = 1 and 𝑠𝐡,𝐢 = 0, the posterior probabilities
for πœ” = 0 and πœ” = 1 are as follows:
Pr(πœ” = 0|𝑠𝐴 = 1, 𝑠𝐡,𝐢 = 0) =
Pr(πœ” = 1|𝑠𝐴 = 1, 𝑠𝐡,𝐢 = 0) =
1
𝑋 (1 − 𝑝)𝜏 2
2
1
1
𝑋 (1 − 𝑝)𝜏 2 + (1 − 𝑋)πœ… ( (1 − π‘ž))
2
2
1
(1 − 𝑋)πœ… ( (1 − π‘ž))
2
2
2
2
1
1
(1 − 𝑋)πœ… ( (1 − π‘ž)) + 𝑋 (1 − 𝑝)𝜏 2
2
2
The committee will only decide to take a policy action, i.e. π‘Ž = 1 when the posterior probability
Pr(πœ” = 1|𝑠𝐴 = 1, 𝑠𝐡,𝐢 = 0) > Pr(πœ” = 0|𝑠𝐴 = 1, 𝑠𝐡,𝐢 = 0). It follows that this is the case when
1−𝑋
𝑋
>
6
Publicity of debate in committee decision-making: The role of the state of the world
1
(1−𝑝)𝜏2
2
2
1
π‘˜( (1−π‘ž))
2
. Given that Pr(πœ” = 1) = (1 − 𝑋) is sufficiently high, the committee will decide to take a policy
action, even when player B and C received a signal 𝑠𝐡,𝐢 = 0.
Results under public deliberation
In this section we will discuss the existence of informative equilibria when both the individual messages
and votes of the committee members will be made public. In this situation, the committee members may
be confronted with an incentive to be untruthful about the signal they received, depending on the
probability of the state of the world 𝑋, depending on their reputational concerns, captured by 𝛽, and
depending on the messages sent by the other committee members. In an informative equilibrium, all
players convey their signal truthfully in the message round. As seen in the analysis of the committee
members’ behaviour under private deliberation, it is possible that players vote on taking a policy action,
even when two of three players receive a signal that the state of the world is equal to 0, depending on
the prior probabilities 𝑋 and (1 − 𝑋).
Given the priors above, the question is whether players have an incentive to reveal their signals
truthfully in the message round, given that the individual messages and votes of the committee
members are made public.
We first consider the behaviour of player C, assuming that player A and B convey their signals truthfully
in the message round. We analyse a situation where player C receives a signal that is in line with the
message sent by player B, but is different from the message sent by player A. In this case, with the
posterior probabilities defined as πœ™0,1 = Pr(πœ” = 𝑠𝐢 |π‘šπ΄ , π‘šπ΅ , 𝑠𝐢 ), utility for player C is defined as follows:
π‘ˆπΆπ‘‘π‘Ÿπ‘’π‘’ = πœ™0,1 + 𝛽[Ο•0,1 Pr(β„Ž|π‘šπΆ = πœ”) + (1 − πœ™) Pr(β„Ž|π‘šπΆ ≠ πœ”)]
π‘“π‘Žπ‘™π‘ π‘’
π‘ˆπΆ
= (1 − πœ™0,1 ) + 𝛽[(1 − Ο•0,1 ) Pr(β„Ž|π‘šπΆ = πœ”) + πœ™0,1 Pr(β„Ž|π‘šπΆ ≠ πœ”)]
1
Where, given π‘šπ΄ = 1, π‘šπ΅ = 0 and 𝑠𝐢 = 0, πœ™0 =
𝑠𝐢 = 1, πœ™1 =
1
2
(1−𝑋) (1−𝑝)𝜏2
1
2
1
2
(1−𝑋) (1−𝑝)𝜏2 +π‘‹πœ…( (1−π‘ž))
2
𝑋2(1−𝑝)𝜏2
1
2
1
2
𝑋 (1−𝑝)𝜏2 +(1−𝑋)πœ…( (1−π‘ž))
2
and given π‘šπ΄ = 0, π‘šπ΅ = 1 and
.
Player C will only decide to be truthful if she thinks the state of the world is more likely to be in line with
1
the signal she received, and will be untruthful otherwise. This will be the case if
1
π‘šπ΄ = 1 and π‘šπ΅ = 0, and if
1−𝑋
𝑋
>
πœ…(2(1−π‘ž))
1
(1−𝑝)𝜏2
2
𝑋
1−𝑋
>
πœ…(2(1−π‘ž))
1
(1−𝑝)𝜏2
2
2
given
2
given π‘šπ΄ = 0 and π‘šπ΅ = 1. We have seen in the chapter
above that agents have an incentive to be truthful about their signal when deliberation occurs privately.
Since we assume here that players A and B will convey their messages truthfully, the condition stated
above is equal to the one for player C under private deliberation.
7
Publicity of debate in committee decision-making: The role of the state of the world
We next consider the incentives of player B to reveal her message truthfully when she receives a signal
that is different from the signal sent by A. Maximizing her utility is a trade-off between the policy payoff
and the reputational payoff and depends both on Χ and 𝛽. Player B will decide to convey her message
truthfully only when Pr(πœ” = 𝑠𝐡 |π‘šπ΄ , 𝑠𝐡 ) > Pr(πœ” ≠ 𝑠𝐡 |π‘šπ΄ , 𝑠𝐡 ), and will otherwise mimic the message sent
by player A.
Anticipating that player C will convey her message truthfully when the priors are sufficiently high (see
above), and assuming that π‘šπ΄ = 1 and 𝑠𝐡 = 0, the utility for player B is given for the situations where
player B conveys her signal truthfully and where player B decides to be untruthful.
When we express the posterior probabilities in πœ™, this results in the following utility functions.
π‘ˆπ΅π‘‘π‘Ÿπ‘’π‘’ = πœ™0,1 + 𝛽[πœ™0,1 Pr(β„Ž|π‘šπ΅ = πœ”) + (1 − πœ™0,1 ) Pr(β„Ž|π‘šπ΅ ≠ πœ”)]
π‘“π‘Žπ‘™π‘ π‘’
π‘ˆπ΅
= (1 − πœ™0,1 ) + 𝛽[(1 − πœ™0,1 ) Pr(β„Ž|π‘šπ΅ = πœ”) + πœ™0,1 Pr(β„Ž|π‘šπ΅ ≠ πœ”)]
Where, given π‘šπ΄ = 1 and 𝑠𝐡 = 0, πœ™0 =
1
2
1
1
(1−𝑋) (1−𝑝)𝜏+π‘‹πœ… (1−π‘ž)
2
2
(1−𝑋) (1−𝑝)𝜏
1
𝑋2(1−𝑝)𝜏
1
1
𝑋 (1−𝑝)𝜏+(1−𝑋)πœ… (1−π‘ž)
2
2
, and given π‘šπ΄ = 0 and 𝑠𝐡 = 1, πœ™1 =
. Player B will convey her message truthfully if she conceives that the state of the
world is more likely to be equal to the signal she received. This will be the case if
π‘šπ΄ = 1 and 𝑠𝐡 = 0, and if
1−𝑋
𝑋
>
πœ…(1−π‘ž)
𝜏(1−𝑝)
𝑋
1−𝑋
>
πœ…(1−π‘ž)
𝜏(1−𝑝)
given
given π‘šπ΄ = 0 and 𝑠𝐡 = 1. When these expressions do not hold,
player B has an incentive to be untruthful about her signal to maximize her payoff.
With the probability of both states of the world equal, player A will always have an incentive to reveal
her message truthfully, as this maximizes both her policy and her reputational payoff. When this prior
probability is not equal, player A will have to condition her message on this probability. Player A
maximizes her utility by sending a message about the state of the world that she thinks is most likely to
be the correct one. In the eyes of player A, Pr(πœ” = 𝑠𝐴 ) > Pr(πœ” ≠ 𝑠𝐴 ) when
0, and when
1−𝑋
𝑋
1
>2
𝑋
1−𝑋
1
>2
(1−𝑝)
πœ…
, given that 𝑠𝐴 =
(1−𝑝)
πœ…
, given that 𝑠𝐴 = 1. If these expressions are not fulfilled, player A will have no
incentive to reveal her signal truthfully in the message round, and consequently, there will be no
informative equilibrium.
Given the conclusions reached above, the possibility of reaching an informative equilibrium when
players B and C receive a message that is different from the message received by player A depends
on the expression players B and C face in their decisions: The right-hand side of the expression B faces
is higher for all relevant values of 𝑝 and π‘ž than the right-hand side of the expression player C faces.
This means that player B has a higher degree of uncertainty regarding the state of the world and is thus
more inclined to send a message in line with the prior. More importantly, when B decides to convey her
signal truthfully, so will player C.
8
Publicity of debate in committee decision-making: The role of the state of the world
Intuitively, this makes sense. The committee member with the highest degree of certainty regarding the
state of the world is, as a result of being informed by the other committee members before expressing
her own opinion, player C. If player B is convinced that it is more probable that the state of the world is
in line with her own message, and the signal of player C is the same as the signal of player B, player C
should be all the more convinced that her signal is in line with the actual state of the world.
To illustrate why the results discussed above make sense, we discuss a trivial situation, namely the one
where the probability Χ is equal to 0 or 1. In this case, when deliberation occurs behind closed doors,
the committee members are indifferent between conveying their messages truthfully, as the outcome
of the decision-making process is already clear; not a single strategy exists for any of the committee
members to get a payoff that exceeds the one when the committee makes the correct policy action. In
a situation where deliberation occurs publicly, a prior probability of 0 or 1 means that none of the
conditions that make an informative equilibrium possible hold. Hence, all players will vote in line with
the prior.
Discussion and conclusion
In this thesis, the role of the state of the world in the context of publicity of debate in committee decisionmaking was analysed. A three-member economic policy committee was considered in which the agents
care about making the correct policy decision and about the opinion of an outsider regarding the
expertise of the agents. It is shown that, when confronted with a probability distribution of the state of
the world that is not equal to 50-50, agents individually condition their behaviour on this distribution.
When deliberation takes place behind closed doors, and individual messages and votes are not
consequently made public, agents have an incentive to be truthful about their messages within the
committee, but may choose to vote in a way that is not in line with the signals that the players received
when the prior probability for one of the two states of the world is sufficiently high. When individual
messages and votes are made public, the likelihood of an informative equilibrium is lower when the
probability distribution of the state of the world is not equal to 50-50. The further away the distribution
is from this level, the more agents will have an incentive to send messages in line with this prior
probability, instead of revealing their messages truthfully. The probability distribution of the state of the
world is shown to have an impact on the behaviour of committee members in both public and private
deliberation. As there are no costs for the committee members to be truthful in a closed deliberation
process, an informative equilibrium is more likely to be reached in a private decision-making process
than in a public decision-making process.
References
Gersbach, H., & Hahn, V. (2004). Voting transparency, conflicting interests, and the appointment of
central bankers. Economics & Politics, 16(3), 321-345.
Gersbach, H., & Hahn, V. (2008). Should the individual voting records of central bankers be published?
Social Choice and Welfare, 30(4), 655-683.
9
Publicity of debate in committee decision-making: The role of the state of the world
Levy, G. (2007). Decision Making in Committees: Transparency, Reputation, and Voting Rules. The
American Economic Review, Vol. 97, No. 1, 150-168.
Meade, E. E., & Stasavage, D. (2008). Publicity of debate and the incentive to dissent: Evidence from
the us federal reserve. The Economic Journal, 118(528), pp. 695-717.
Swank, J., Swank, O. H., & Visser, B. (2008). How committees of experts interact with the outside world:
some theory, and evidence from the FOMC. Journal of the European Economic Association,
6(2-3), pp. 478-486.
Visser, B., & Swank, O. H. (2007). On Committees of Experts. Quarterly Journal Of Economics,
forthcoming.
10
Download