It has become common for scholars, activists, and other outside
observers to recommend that international organizations become more
“transparent” in their operations. One important aspect of
these calls for openness involves the idea that deliberations,
discussions, and bargaining in international fora should be subject to
outside scrutiny, and, in particular, to the same level of scrutiny
that is presumed to exist for political bargaining within democratic
states. Members of the public should have the right to either directly
observe international negotiations, or they should have access to
detailed minutes of proceedings. Calls for greater transparency have
been made for a plethora of international organizations, including the
Bretton Woods institutions, the World Trade Organization (WTO), and the
institutions of the European Union (EU).1
While
transparency has attracted much recent attention, the subject is hardly a
new one; after World War I there were frequent demands for greater openness
in international diplomacy, symbolized by Woodrow Wilson's call for
“open covenants of peace, openly arrived at.”
2 The citation is from the first of
Wilson's fourteen points. For discussions of calls for more open
diplomacy during this period, see Nicolson 1939.
The motivation behind demands for
transparency in international bargaining is clear; the more that
citizens know about the actions of government officials, the easier
they will find it to judge whether officials are acting in the public
interest. Ultimately, transparency can make officials more accountable
for their actions. It may also have other benefits. Advocates of
deliberative democracy emphasize that deliberations that occur in
public increase the quality and the legitimacy of decisions taken.
While many normative arguments have been made in favor of making
international organizations more transparent, there has been much less
effort to develop and test positive hypotheses about the concrete
effects of making bargaining public. In this article I develop a
game-theoretic model that shows that while transparency in bargaining
may make governments more accountable, it may also prompt officials to
posture, leading to more frequent breakdowns in negotiations. I also
discuss several alternative and complementary arguments. I then provide
evidence to support four propositions derived from the model, based on
the current dispute whether to make European Council of Ministers
debates public, and based on earlier, eighteenth-century conflicts over
levels of “publicity” for parliamentary debate.
Given that accountability of public officials is a widely shared goal
in democracies, one might be prompted to ask how one could ever argue
in favor of secrecy in government, barring those cases where revelation
of information risks jeopardizing national security or violating
individual rights. In the international context, if transparency allows
national governments to usefully retain control over international
organizations, then since national governments establish such
organizations, one might ask why they do not always insist on
monitoring procedures that ensure as much transparency as possible. In
fact, classic contributions from democratic theory and more recent
contributions by authors using game-theoretic models suggest that
“transparency” or “publicity” (I use the two
terms interchangeably) may have important costs. When government
officials bargain in front of an audience, they may face a greater
incentive to posture by adopting uncompromising bargaining positions.
These positions can be used to send a signal about the alignment of
representatives' preferences with those of their constituents.
Likewise, the presence of an audience may make officials more reluctant
to retreat from initial claims when faced with superior evidence.
In order to explore these issues formally, I develop a game-theoretic
model that compares the outcomes of bargaining behind closed doors with
open-door bargaining where the public observes all actions taken by
officials. In the model, two jurisdictions, A and B, must bargain over
a policy choice, and bargaining for each jurisdiction is conducted by
an agent (the “representative”) on behalf of a principal
(the “public”). The bargaining procedure is kept simple by
assuming that one representative is selected at random to make a
proposal that the other representative must then either accept or
reject. If the proposal is rejected, then the public from each
jurisdiction receives a disagreement payoff. In addition, I assume that
there are two sources of asymmetric information in this game. For one,
the members of the public in jurisdiction A are uncertain whether their
representative shares their policy preferences, or alternatively,
whether their representative is biased. Second, there is uncertainty
about the minimal offer that the public from each side will find
acceptable, and representatives have private information about this
variable. I then consider two alternative bargaining scenarios. Under
open-door bargaining, the public observes who is chosen to make an
offer, it observes the content of any bargaining offer, and it observes
the final outcome of bargaining. Under closed-door bargaining, the
public observes only the final outcome and none of the intermediate
steps.
The model presented here is similar to a two-level game developed by
Putnam, although with one key difference—rather than being
constrained by the need for subsequent ratification of an agreement by
a constituency, in my model bargainers are constrained by the fact that
actions taken during negotiations can influence a constituency's
belief whether a bargainer is biased.3
The setup
for my model draws on a recent contribution by Fingleton and Raith that
does not specifically consider political bargaining.
4 Fingleton and Raith 2002.
While these two authors model a
situation where there are different “types” of bargainers
that vary in their level of expertise (ability to determine the optimal
bargaining offer), I focus on a more political context where there are
different types of representatives who have different preferences over
policy. I adopt this assumption because current debates about the need
for transparency in international organizations, as well as
eighteenth-century disputes about parliamentary transparency, have
centered on the risk that representatives have biased preferences. The
model in this article also differs significantly from Fingleton and
Raith's contribution in that I assume that bargainers care both
about their reputation and about the bargaining outcome (otherwise it
would be meaningless to be biased), whereas Fingleton and Raith assume
that bargainers care only about their reputation. As discussed in the
next section, this has significant implications for predictions about
the desirability of transparency.
In comparing open-door and closed-door bargaining in the model, one
observes two potential effects of moving from one practice to the
other. The first involves an accountability effect, as open-door
bargaining increases the likelihood that both unbiased and biased
representatives will propose a policy that is close to the outcome most
preferred by the public. Were this the only effect of transparency,
then one could conclude that it is unambiguously beneficial for the
public. However, transparency may also have negative consequences.
Uncertainty about disagreement payoffs can create an incentive for
representatives to “posture” by adopting uncompromising
bargaining positions. These uncompromising positions may be adopted to
convince the public that representatives are not caving in to
opposition demands because they are biased. Under these conditions, it
is possible for open-door bargaining to actually make the public of
jurisdiction A worse off if, in equilibrium, both unbiased and biased
representatives posture in this manner.5
This posturing incentive for both types of representatives
is similar to the “political correctness effect” identified
by Morris 2001. Koremenos 2003 has identified an alternative potential cost
of open-door bargaining, suggesting that it can increase susceptibility
of decision-makers to interest group lobbying.
Based on the above model, I present four propositions about the
causes and consequences of transparency. These propositions build on
existing literature that uses rational choice theories to examine the
motivations for actors to choose different forms for international
institutions.6
These propositions also follow
the suggestion by Martin and Simmons that authors consider how international
institutions are simultaneously causes and effects.
7
Finally, my propositions also build on the existing idea from the two-level
games literature that negotiators may use asymmetries of information (such
as holding closed-door sessions) to increase their degree of insulation
from constituencies and thus alter the size of their
“win-set.”
8 Putnam 1988, 445, observes that negotiations may be held
behind closed doors for this purpose. See also Moravcsik 1994; and Milner 1997.
My first proposition suggests that
when representatives are sufficiently concerned about their reputation,
under open-door bargaining they are more likely to make bargaining
proposals close to those most preferred by citizens. However, the
second proposition suggests that when reputational concerns are
significant, open-door bargaining will also prompt representatives to
posture by taking uncompromising positions that may result in
bargaining breakdowns. In a third proposition, I suggest that the
public will prefer open-door bargaining if there is a significant fear
that representatives may be biased. Finally, the model predicts that
both biased and unbiased representatives will prefer closed-door
bargaining.
One interesting feature of recent calls for greater transparency in
international organizations is the extent to which these current
disputes parallel earlier, eighteenth-century conflicts over the degree
of transparency that should prevail in national legislative
institutions. In order to demonstrate that my propositions can apply to
international and domestic bargaining, the third section of this
article reviews historical evidence regarding “publicity”
and the eighteenth-century House of Commons, the French Constituent
Assembly of 1789, and representative assemblies in the early American
republic. In the British and U.S. cases, the initial norm was to hold
deliberations behind closed doors, but this practice gave way, under
external pressure. There is clear evidence for the United States and UK
that demands for transparency appeared during periods of heightened
fears that representatives were biased. In strong contrast, during
periods where fears of bias were less present, the public was more
accepting of closed-door sessions. One can also observe that in the
French Constituent Assembly, where an open-door policy prevailed from
the outset, legislators were more responsive to the opinions expressed
by the public, but there was also a great deal of posturing by
representatives that arguably resulted in several bargaining
breakdowns. All of these observations are consistent with the
theoretical propositions developed in the second section of this
article.
In the fourth section, I examine whether my propositions about
transparency are supported by recent evidence from the European Council
of Ministers. Unlike debates of the European Parliament, Council
deliberations are not open to the public, and before 1993 there was no
procedure for the public to gain access to minutes of Council meetings.
In 1995, The Guardian newspaper launched a lawsuit with the
European Court of First Instance to obtain access to records of
meetings of the Council. This suit was opposed by a majority of EU
governments. Much like eighteenth-century demands for openness, the
current dispute in the EU has emerged during a period of increased
fears that representatives may be biased. In a context of rising
concerns about Europe's “democratic deficit,”
proponents of transparency have insisted that the public gain access to
Council deliberations for reasons of accountability. The EU governments
that oppose such a move have argued that open-door bargaining would
lead to clear inefficiencies, as governments bargaining before an
audience would find it more difficult to strike compromises. This
situation contrasts strongly with that which prevailed in Europe during
the mid-1980s around the time of the signature of the Single European
Act. During this earlier period there were fewer fears of a disjuncture
between the preferences of representatives and the public over policy,
and one observes few criticisms of Council secrecy. The fact that
demands for transparency have emerged during a period of perceived bias
is consistent with my theoretical propositions, as are the observed
consequences of Council secrecy, which facilitates compromise but
weakens accountability. Several recent studies of the EU Council of
Ministers support this contention. The fourth section draws on recent
literature on committee decision making within the EU to argue that a
similar pattern is seen within the committees that report to the
Council of Ministers, and in particular the Committee of Permanent
Representatives (COREPER), and the Economic and Financial Committee
(EFC).
In the remainder of the article I first develop my theoretical
propositions about the causes and consequences of transparency in the
second section, followed by an empirical application to
eighteenth-century parliamentary debates in the third section, and to
the current dispute about transparency in EU Council of Ministers
deliberations in the fourth section. The final section concludes.
Causes and Consequences of Transparency
The most commonly cited benefit of transparency in government is that
it makes it easier to hold officials accountable for their actions. As
an early advocate, Jeremy Bentham argued that “publicity”
in a parliamentary context would “constrain members of the
assembly to perform their duty” and that it would also
“secure the confidence of the people, and their assent to the
measures of the legislature.”9
In the absence of such publicity, it may be difficult for electors to judge
whether a representative has taken their interest in consideration when
bargaining over policy, or alternatively, whether unseen actions by
lobby groups are dominating outcomes. In economic terms this would be
referred to as a problem of moral hazard, the risk that representatives
will pursue private goals over those of their electors. The classic
strategy for reducing moral hazard is to make the actions of an agent
(in this case an elected official) more observable.
10 Many recent writers about public affairs make the implicit assumption
that transparency in government is unambiguously beneficial, precisely
because it can make officials more accountable, and it can help
facilitate emergence of a consensus over policies. As a result, one
might conclude that, except in rare cases, transparency should be the
rule. Bentham suggested that “publicity” should only be
limited in cases where it would jeopardize national security, where
there would be a risk of violating privacy, or if it would
“inflict too severe a punishment upon the guilty.”11
This unbounded optimism about the effect
of public opinion was not shared by a number of subsequent critics. John
Stuart Mill argued that Bentham's political propositions went too far
in “riveting the yoke of public opinion closer and closer round the
necks of all public functionaries,” thus excluding the possibility
that a representative might use his or her own reason in making a
decision.
12 Alexis de Tocqueville took a similar view.
13 As argued by Habermas, Mill's and Tocqueville's
critiques of Bentham were part of a more general move, beginning in the
nineteenth century, to conceptualize public opinion as a force that could
have negative and positive consequences.
14 See
Habermas 1962. Criticisms of transparency in
government have existed at least since Thomas Hobbes 1651
[1998], chap. 19. I would like to thank an anonymous
referee for this point.
Observers have identified at least two potential negative
consequences of transparency: a risk of pandering by public officials
and a risk of posturing during bargaining. Though I focus primarily on
the latter phenomena, it is worth considering both here, because they
are closely related. Pandering can be defined as a situation where
elected representatives choose policies based on voter opinion, even if
representatives themselves believe that voters are incorrectly informed
about their true interests (one might prefer to use the term
“responsiveness” given that pandering has quite negative
connotations in common usage). More specifically, in recent discussions
of pandering it is suggested that voters may be uncertain about the
link between policies and outcomes.15
This could apply to situations as diverse as
tax-cut proposals, calls for global environmental regulation, or
monetary policy decisions. If the public is indeed incompletely
informed about the consequences of a given policy choice, then the
temptation for an individual representative to pander might be higher
the greater the extent to which the public is able to directly observe
the representative's actions. Prat has presented a model with a
quite general result showing that when a principal cannot observe the
consequences of actions taken by an agent, then it may be optimal to
also keep the actions themselves secret.
16 Posturing is a phenomenon related to pandering that refers to the
incentive for representatives to adopt uncompromising positions during
negotiations, to demonstrate to their constituents that they are
effective or committed bargainers. The problem with posturing is that
it can provoke a breakdown in bargaining that has a negative impact for
all concerned. Fingleton and Raith have recently developed a model that
examines how posturing may occur if two agents each bargain for a
principal, if the agents have private information about the minimal
offer acceptable to the other side, and if the agents vary in their
ability to determine this minimal offer.17
Fingleton and Raith 2002.
Below I present a model that draws
on their contribution, but where I assume that the public is instead
uncertain about a representative's preferences, and where I also
assume that representatives care both about the policy outcome and
about their reputation. As argued in the introduction, this assumption
that agents vary in terms of policy preferences is better suited for
addressing both classic debates about publicity in parliamentary
settings and recent debates about transparency in international
institutions. The assumption also results in a different prediction
from Fingleton and Raith regarding the preferability of open-door
bargaining, as discussed under Proposition (3) below.
Formalizing the Problem
I consider a scenario where two jurisdictions, A and B, must bargain
over a unidimensional policy choice [0,1]. Members of the
public from jurisdiction A prefer a policy of 0 while the public from
jurisdiction B prefers 1. Given a policy choice x, the public
from jurisdiction A receives utility UA =
(1 − x) and the public from jurisdiction B receives
utility UB = x. Policy is chosen
through delegated bargaining whereby a representative negotiates on
behalf of each jurisdiction. The bargaining game is kept simple by
assuming that one of the two representatives is selected at random to
make a proposal o that the other representative must then
either accept or reject.18
The assumption
that proposal power is random here is a convenient way of introducing
uncertainty into the outcome of the game. All of the results reported
here would still hold if proposal power was not random, provided there
was some other possibility for a random event beyond players'
control to influence the bargaining outcome.
If the policy is
accepted it is implemented; if the policy is rejected then a
disagreement payoff
d applies (with
d ≤ 0.5). The
disagreement payoff determines the minimal acceptable proposal for the
public from each jurisdiction, as shown in
Figure
1.
Preferred and minimum acceptable policies
In a one-stage bargaining game with complete information one would
normally expect B's representative,
RB, to propose A's minimum acceptable
policy (1 − d), and one would expect
RA to propose d. In order to make
the issue of open-door versus closed-door bargaining relevant, I add
two further assumptions to the model: there is a risk that the
representative for jurisdiction A is biased, and there is asymmetric
information about the value of the disagreement payoff d.
To consider the potential effect of a bias for representatives, I
assume that with probability p, representative
RA shares the preferences of the public
from jurisdiction A, and with probability (1 − p), he or
she is biased and instead prefers a policy outcome x = 1.
Representative RA derives utility both
from policy and from the public's ex post assessment of
the probability that he or she is unbiased, given the bargaining
outcome and the bargaining proposal
Pr(u|x,o), as expressed in equations
(1) or (2) below. The parameter α determines the relative weight
that RA places on achieving the desired
bargaining outcome, versus creating a reputation for being unbiased.
Making the payoff for an agent conditional on the principal's
ex post assessment of the agent's type is a common
specification in the literature on “career concerns” in
principal-agent relationships.19
Following the canonical paper by Holmström 1982. See Ottaviani and Sorensen 2003 for a recent review of the literature on
agents with “career concerns.”
In the political
context considered here, one could motivate the assumption by
suggesting that the probability that the representative will be
retained by the public for some future (unspecified) periods of play is
an increasing function of the posterior
Pr(
u|
x,
o).
20 All of the theoretical results presented below would also
hold if I adopted the alternative strategy of specifying an explicit
two-period model where the public would decide based on the first
period bargaining offers and outcome whether to retain representatives,
provided that there was some uncertainty on the part of representatives
about the actions that would lead to their replacement.
The
representative from the second jurisdiction,
RB is assumed for simplicity to always
share the preferences of the public from jurisdiction B.
To consider the effect of asymmetric information about the
disagreement payoff, I also assume that representatives
RA and RB
each receive a private signal about the utility that the public from
the other jurisdiction receives from disagreement. The utilities from
disagreement are assumed to be either d with probability 0.5
(and d < 0.5), or 0.5 with probability 0.5, implying that
the best deal that can be achieved is an even split. This signal
s ∈ {d,0.5} is accurate with probability
q with 1 > q > 0.5. The idea behind these
additional assumptions is that during the course of any negotiation,
bargainers are likely to form more certain beliefs about the minimal
offer acceptable to their opponent. Odell has provided abundant
evidence that participants in international economic negotiations have
incomplete information about disagreement payoffs of other parties.21
Odell 2000. He
also argues convincingly that bargainers may be uncertain of their own
disagreement payoffs.
Because these signals are private, they
cannot be directly observed by the public either under closed-door or
open-door bargaining. The signals represent the intuition of individual
bargainers. Crucially, I also assume that these signals cannot be
credibly transmitted by a representative to the public. Because some
representatives are biased, there would be a clear incentive to
misrepresent signals if they were communicated. The bargaining game
proceeds in the following stages.
- Nature determines whether RA is
biased, and it determines the disagreement payoffs.
- RA and RB
each receive a signal about the other jurisdiction's disagreement
payoff.
- RA or RB
is randomly selected to propose.
- The proposal is implemented if the nonproposing representative
accepts. Otherwise disagreement payoffs apply.
- Under open-door bargaining, the public observes the proposal
o, as well as the bargaining outcome x. Under
closed-door bargaining the public observes only x.
- RA receives a payoff based on the
bargaining outcome and on the posterior of his or her type:
Pr(u|x,o)
Given the above assumptions, representative
RA has an incentive to make a bargaining
proposal that will result in a policy outcome close to his or her ideal
point, but RA may also have an incentive
to use the proposal to signal that he or she is unbiased. In what
follows, I consider how these incentives vary depending on whether
bargaining takes placed behind closed-doors or in an open-door setting.
The key conclusion will be that open-door bargaining presents a
trade-off. It may discipline biased representatives by prompting them
to make a bargaining proposal closer to the preferred outcome of the
public from jurisdiction A, but it may also prompt both biased and
unbiased representatives to posture.
Open-door bargaining
I begin by considering perfect Bayesian equilibria of the bargaining
game under open-door bargaining (all proofs are presented in the
Appendix).22
I assume that
representatives are restricted to pure strategies, and that they may
offer either d, 0.5, or 1 − d in equilibrium.
This latter restriction simplifies the exposition of the model while
not altering any of the results or propositions.
Under
open-door bargaining, the public observes the bargaining outcome, and
it also directly observes which representative is selected to make a
proposal, as well as the content of the proposal. The public's
inference about
RA's type, given the
outcome and the bargaining proposal, can be expanded as follows using
Bayes's rule.
From the public's point of view, the ideal outcome of the game
would involve some mechanism that ensures that both the biased and the
unbiased representatives pursue the following bargaining strategy:
propose d if their private signal is d, and propose
0.5 if their private signal is 0.5. Unfortunately for the public, it
can never be an equilibrium for both types to play this
“truthful” strategy. If both the unbiased and the biased
representative played the “truthful” strategy, then the
probability for the public of observing either a proposal of 0.5 or a
proposal of d would be the same regardless of whether
RA is biased. This implies that
Pr(u|o = 0.5) =
Pr(u|o = d). Under these conditions,
biased representatives will have an incentive to ignore their signal
and always propose 0.5.
If there is no “truthful” equilibrium, the next question
is what alternative equilibria are possible. Below I show that while
there are some ranges of parameters for which multiple equilibria
exist, and where biased and unbiased representatives may make different
proposals, as long as representatives care sufficiently about their
reputation, there is a unique “posturing equilibrium” of
the bargaining game. In this posturing equilibrium both an unbiased and
biased representative RA will propose
d regardless of their private signal to avoid appearing
biased.23
Given that representative
RB is by assumption known to be unbiased,
the equilibrium behavior of RB is particularly
straightforward and is presented in the Appendix.
This result
makes intuitive sense; transparency should have the biggest impact on
representatives who are concerned about their reputation.
When reputational concerns are weak, both unbiased and biased
representatives may have an incentive to posture, but it is also
possible for two separating equilibria to exist where the two types
pursue different strategies. In the first separating equilibrium,
biased representatives always propose (1 − d), and
unbiased representatives play the truthful strategy. When biased
representatives care above all about policy, rather than reputation,
they will have an incentive to make the closest proposal possible to
their preferred bargaining outcome, even if this results in full
revelation of their type. They will have an incentive to pursue this
strategy in equilibrium as long as
.
Even if the disagreement payoff is zero, this equilibrium can only
exist when biased representatives place at least twice as much weight
on the bargaining outcome as on their reputational payoff. In the
second separating equilibrium, unbiased representatives play the
truthful strategy, while biased representatives always propose 0.5.
This equilibrium can also only exist if reputational concerns are weak,
because otherwise unbiased representatives who receive a signal
s = 0.5 will have an incentive to deviate by proposing
d, so that the public fully learns their type (remembering
that biased representatives never propose d in this
equilibrium). In addition, this second equilibrium cannot exist if
reputational concerns are very weak, or else biased representatives
will have an incentive to propose (1 − d) even if this
fully reveals their type. As shown in the Appendix, this implies that
the second equilibrium can only exist when the public's prior
belief that the representative is unbiased is relatively high
.
When reputational concerns are sufficiently strong, there will be a
unique equilibrium of the bargaining game where both the biased and the
unbiased representative posture by always proposing d
irrespective of their private signal. When representatives give
significant weight to reputational concerns (α is low), they will
have an incentive to use their bargaining proposal to signal that they
are unbiased. A biased representative will find it increasingly costly
to pursue a strategy of proposing (1 − d) or 0.5. An
unbiased representative will find it increasingly costly to propose
0.5, even if he or she believes that not proposing 0.5 will result in
disagreement, because a proposal o = 0.5 may be taken as an
indication that he or she is biased. The end result is that the only
possible equilibrium strategy is for both the biased and unbiased
representative to always propose d. This equilibrium has the
advantage for the public of disciplining biased representatives, but it
has the disadvantage of also ensuring that unbiased representatives
fail to make use of their private information.
As shown in the Appendix, at a minimum, the posturing equilibrium is
unique whenever p < 2/3 and α < 2/3.
24 Based on the two inequalities (14) and
(15).
As the private information of representatives becomes
less accurate, and as the disagreement payoff grows larger, the range
of parameter values for which posturing is the unique equilibrium
expands. Because neither 0.5 nor (1 −
d) are proposed in
the posturing equilibrium, the equilibrium also includes a
specification of the “out of equilibrium belief” that the
public attaches to a proposal
.
The two conditions for existence of the posturing equilibrium, which are
presented in expressions (4) and (5), involve the unbiased
representative's incentive to offer 0.5 rather than d, and
the biased representative's incentive to offer (1 − d)
rather than d.
One can conclude that unless reputational concerns are weak,
open-door bargaining will produce posturing behavior on the part of
representatives.
Closed-door bargaining
Equilibrium outcomes change significantly when representatives
bargain in secrecy. Under closed-door bargaining the public observes
the final outcome of the bargaining process, but it does not observe
who was selected to propose, nor does it observe the content of the
proposal.25
I also assume for simplicity
that under closed-door bargaining the public cannot distinguish between
an outcome where a proposal of (1 − d) is accepted, in
which case the public receives utility d, and a breakdown in
bargaining, in which case the public also receives utility d.
None of the results presented depend on this assumption.
As a
consequence, the public must establish inferences about
RA's type based exclusively on the
bargaining outcome Pr(
u|
x). This greater
difficulty in inferring a representative's type has two
contrasting effects; biased representatives will now have less of an
incentive to mimic the actions of unbiased representatives, but
unbiased representatives will also have less of an incentive to ignore
their private signal. This presents a clear trade-off for the
public.
Under closed-door bargaining, there is in fact a unique perfect
bayesian equilibrium in pure strategies where a biased representative
always proposes (1 − d) and the unbiased representative
plays the truthful strategy. This pair of strategies constitutes an
equilibrium for a greater range of parameter values than is the case
under open-door bargaining, because a biased representative will now
incur a lower reputational cost from proposing (1 − d).
When the public observes an outcome (1 − d), it will
revise its belief that the representative is biased downwards
Pr(u|x = 1 − d) <
p, but the posterior will not be zero, because it is also
possible for the outcome (1 − d) to occur with an
unbiased representative if RB is selected
to propose. For the public this separating equilibrium is
unsatisfactory to the extent that a biased representative always
proposes an unfavorable policy, but the equilibrium is advantageous to
the extent that an unbiased representative uses private information
efficiently, rather than posturing.
As discussed in the Appendix, one can rule out three alternative
possibilities for equilibria under closed-door bargaining: both
representatives playing truthfully, both posturing, and the biased
representative always proposing 0.5 while the unbiased representative
plays truthfully. In any of these three equilibria the posterior
probability that the public attaches to the outcome (1 −
d) would be simply p. This is because the outcome (1
− d) could be observed if
RB was selected to propose, and it would
be equally probable with either a biased or an unbiased representative.
As a result, the biased representative could always propose (1 −
d) and achieve an improvement in the bargaining outcome
without any loss of reputation.
Based on the possible equilibrium outcomes under open-door and
closed-door bargaining one can make the following two propositions
about the effect of open-door bargaining on bargaining outcomes.
Proposition 1: Transparency ensures that representatives
will make proposals close to the preferred outcomes of citizens,
provided reputational concerns are sufficiently strong.
As long as representatives are sufficiently concerned about their
reputation, under open-door bargaining one will observe a unique
equilibrium where both biased and unbiased representatives always
propose d. When one compares this with the unique
pure-strategy equilibrium under closed-door bargaining, one can observe
that the expected proposal under transparency will always be closer to
the public's preferred outcome.
Proposition 2: Transparency will prompt representatives to
posture, provided reputational concerns are sufficiently strong, and
the public perceives a risk of bias.
Under open-door bargaining the posturing equilibrium will be the
unique equilibrium as long as reputational concerns are sufficiently
strong. In any empirical investigation one should expect to see more
uncompromising positions taken during open-door bargaining, greater
polarization of debate, and more frequent breakdowns in bargaining than
would otherwise be the case.
Causes of transparency
In addition to propositions about the consequences of transparency,
one can also develop hypotheses about the causes of transparency, using
a simple thought experiment. One could extend the model by adding an
initial stage where the public from jurisdiction A must decide whether
to have open-door or closed-door bargaining. If this were the case,
then transparency would prevail whenever the expected utility for the
public from open-door bargaining exceeds the expected utility from
closed-door bargaining. When one considers those cases where
reputational concerns are sufficiently strong to make posturing the
unique equilibrium outcome under transparency, it becomes clear that
this trade-off will depend in large part on p, the prior
probability that the public attaches to their representative being
unbiased. The public's expected utility in the posturing
equilibrium will be simply 0.5. Expected utility in the posturing
equilibrium does not depend on p, because both biased and
unbiased representatives behave identically. In contrast, expected
utility in the unique equilibrium under closed-door bargaining depends
directly on p. The higher the probability that the
representative is unbiased, the more the public believes that
closed-door bargaining will result in the representative using private
information efficiently. The lower the probability, the greater the
risk that the representative makes a proposal o = 1 −
d. The public will prefer open-door bargaining whenever the
inequality in expression (6) is satisfied.
This simplifies to the expression below:
Proposition 3: The public will prefer open-door bargaining
if there is a significant risk of bias and reputational concerns are
sufficiently strong.
The inverse of Proposition 3 also holds when reputational concerns
are sufficiently strong, as the public will prefer closed-door
bargaining when there is little risk that representatives are biased.
Proposition 3 would also hold if I had specified an explicit two-period
model. In such a model there would be an additional advantage of the
separating equilibrium in that it would allow the public to use
first-period behavior to sort between biased and unbiased
representatives and to retain only a representative it thought was
unbiased. This would not be possible in the posturing equilibrium as
both types of representative behave identically. As a result, in a
two-period model, the critical value of p below which the
public would prefer open-door bargaining would be somewhat lower.
However, one would still observe that in cases where the risk of bias
is sufficiently strong, the public will prefer transparency. Finally,
Proposition 3 provides a different result from that obtained if one
instead assumed that representatives varied in terms of bargaining
skill, and that representatives did not care about the bargaining
outcome, as in Fingleton and Raith's model.26
Fingleton and Raith 2002.
In their model, skilled bargainers
always use their private information efficiently, regardless of whether
bargaining occurs in public, and unskilled bargainers have a greater
tendency to posture when there is open-door bargaining. As a result,
closed-door bargaining will always provide a better expected policy
outcome for the public, and one does not observe the same trade-off
presented here.
Though principals will prefer open-door bargaining whenever there is
a significant expectation that their representative may be biased,
representatives may themselves have different preferences with regard
to the choice of bargaining institutions. An unbiased representative
will always do better under closed-door bargaining if posturing would
otherwise prevail under transparency. Closed-door bargaining allows
unbiased representatives to use their private information efficiently,
and it also allows them to gain a reputation by distinguishing
themselves to a degree from
RA-biased. Biased representatives
will also earn higher expected utility from closed-door bargaining,
because it will allow them to propose a more preferable policy from
their point of view, even if this does entail some cost in terms of
reputation. This leads to a fourth and final proposition.
Proposition 4: Representatives will prefer closed-door
bargaining if reputational concerns are sufficiently strong.
As a corollary to Propositions 3 and 4, if a conflict emerges between
the public and a representative over the setting for bargaining, then
it seems logical that during any such dispute, the public would be
expected to emphasize the advantages of transparency for holding
representatives accountable, while representatives could be expected to
emphasize the risk that transparency will lead to more frequent
breakdowns in negotiations.
Alternative and Complementary Arguments
The game theoretic model developed in this article has provided
propositions about the causes and consequences of open-door bargaining.
Though the model captures a number of the trade-offs about
“publicity” emphasized in democratic theory, there are
several alternative observations that also deserve consideration. The
first of these concerns the effect of transparency in a context where
bargaining involves argument between parties who may be uncertain about
the outcome of different policies. From a normative standpoint, it has
been argued that public deliberation has the advantage of building
societal consensus. From a positive standpoint, scholars have observed
that when debate takes place in public, participants have an incentive
to avoid appeals to self-interest. On the downside it has also been
observed that participants in public debate may be less willing to
retreat from initial opinions. I argue here that these predictions
generally complement, rather than contradict, the predictions of my
model.
In many, if not most political negotiations, participants do not
restrict themselves to making the sort of bargaining propositions
described in the model above; they also make truth claims about the
effect of different policy choices on outcomes. As a result, most
political negotiations inevitably involve both an element of bargaining
and an element of deliberation. The work of Habermas has been
particularly influential in suggesting that public deliberation helps
to build societal consensus and legitimacy for policy choices. In the
parliamentary context, he has argued specifically for “publicity
requirements that keep institutionalized opinion- and will-formation
open to the informal circulation of general political
communication.”27
His subsequent work suggests
that such conclusions could also be extended to the EU context.
28 However, Habermas has also indicated
that he does not believe that public deliberation should necessarily be
the norm in all governmental institutions.
29 The
arguments made by Habermas point to important potential benefits of
public debate that are not captured in the model presented above, but
these arguments do not contradict the basic insights from Propositions
1 to 4. There is a potential tension, however, to the extent that the
model presented in this article predicts that representatives will have
less incentive to reveal private information when discussions are
public, while the arguments of Habermas emphasize the improved quality
of deliberation when it takes place in public.
If negotiations involve an element of deliberation, then the presence
of an audience may also produce a further effect that Elster refers to
as the “civilizing force of hypocrisy.”30
When deliberation takes place in
public, he suggests actors face strong incentives to base their
arguments on claims regarding the general interest, rather than on
selfish appeals. Though Elster's main concern has been with
deliberation in national settings, recently Risse has used this same
proposition to suggest that the establishment of publicity in
international negotiations will also have a “civilizing”
force on actors.
31 This “civilizing” force
is a further potential benefit of public debate that does not emerge
from the game-theoretic model developed above, which is limited to a
bargaining context where preferences of the public over outcomes are
known and fixed.
Though public deliberation may have benefits in terms of building
consensus and in prompting officials to make claims based on general
interest, it can also have clear disadvantages. A number of authors
have suggested that the presence of an audience may make officials more
reluctant to retreat from initially stated positions when confronted
with persuasive counter-arguments.32
If deliberation occurs in public,
then officials may also be more reluctant to venture opinions that
deviate from “received wisdom.” This additional consequence
of open-door deliberation is closely related to the problem of
posturing in open-door bargaining that is presented in this article.
Eighteenth-Century Disputes over
“Publicity”
Though today one takes it for granted that parliamentary debate
should occur in public, when modern forms of representative democracy
first emerged, the more common expectation was that deliberations of
any legislative body should take place in secret. In Great Britain, the
House of Commons banned publication of its debates, and following this
example, colonial and revolutionary assemblies in North America
deliberated behind closed doors. The proceedings of the U.S. Senate
also initially took place in secret. By the end of the eighteenth
century, this norm of closed-door proceedings gave way in response to
demands that the public should be better informed about the actions of
its representatives. The issue of closed-door versus open-door
proceedings in eighteenth-century assemblies has interested a number of
different democratic theorists and historians.33
Among contributions by democratic theorists, see Habermas
1962; Manin 1997;
and Elster 1991. For discussions of the issue
by historians see Rossiter 1966; Stewart
1969; Schwoerer 1977; and Siebert 1952.
This section briefly considers the
debate about parliamentary “publicity” in the United
States, the UK, and France. For the United States and the UK, there is
strong evidence that public demands for transparency emerged during
periods where there were substantial fears that legislators were
biased. In strong contrast, in periods where there were fewer fears of
bias, the public was much more accepting of closed-door legislative
sessions. This is consistent with Proposition 3. Evidence about the
type of arguments made by those favorable and those opposed to openness
also support my theoretical model. Finally, in this section I also
consider the consequences of publicity, arguing that eighteenth century
experience provides clear evidence of a trade-off, with public
bargaining increasing both accountability and the risk of posturing.
Explaining Opposition to and Support for Publicity
The English House of Commons
An early dispute about publicity of House of Commons sessions took
place during the Convention Parliament of 1689. This was the Parliament
convened immediately after the Glorious Revolution. A number of Whig
members of Parliament (MPs) favored dropping the historic ban on
publishing debates, but a majority of MPs opposed any such change,
suggesting that this would restrict the freedom of parliamentary
debate. What is particularly striking about the Convention
Parliament's ban on publication of its proceedings is that there
was no public protest at this measure. Based on extensive research of
newspapers from the period, Schwoerer concludes that there were a large
number of leaked reports of parliamentary debates, but she was unable
to find any instances where the authors of these publications protested
against the official ban.34
One potential reason why the
British public in 1689 accepted parliamentary secrecy is that this was
not a period where there was a particularly strong fear of a bias on
the part of representatives. While Whig majorities in the Commons
would, beginning in the 1720s, be increasingly criticized as corrupt
and unrepresentative oligarchies, there was far less expression of such
sentiment during this earlier period of Whig rule.
35 In sum,
the 1689 episode is consistent with Proposition 3 from the theoretical
model; during a period where there was relatively little fear that
representatives might be biased, the public accepted parliamentary
secrecy.
The next dispute over publishing House of Commons deliberations
occurred in a radically changed political context, as fears that the
Commons majority was biased helped feed public demands for greater
information about proceedings within Parliament. In 1738, prompted by
the fact that clandestine reports of proceedings regularly appeared in
newspapers, several MPs presented a motion, which eventually received
majority support, calling for the Commons to reaffirm its traditional
prohibition of the publication of its debates. Some supporters of the
motion, and most notably the leader of the Commons majority, Robert
Walpole, argued that it was necessary because of the danger that a
partisan press would misrepresent the content of speeches.36
Others, including some of Walpole's
traditional opponents, argued that opposition would restrict freedom of
debate precisely because it would increase accountability.
37 It is useful
to remember the political context underlying this dispute. By 1738,
Walpole had been a leader of the ruling Whig Party for twenty years,
and his rule was subject to increasing criticism. Historians of the
period have referred to a growing public perception that Walpole's
Commons majority was an oligarchy held together by corruption financed
by “monied interests.”
38 Speck suggests that “the aristocracy and gentry,
together with merchants, plutocrats and leading professional men,
amalgamated to form a narrow oligarchy which controlled all the levers
of power, at the centre and in the provinces.” Speck 1977, 163.
By 1738 a number of
anti-Walpole newspapers had appeared, including the two best-selling
British dailies of the period,
The Craftsman and
The
Gentleman's Magazine, each of which regularly reported
parliamentary debates.
39 In stark contrast to
the 1689 episode, there is also clear evidence that the parliamentary
effort to maintain secrecy in 1738 generated substantial public
opposition. This is consistent with Proposition 3; as the British
public grew more fearful that representatives were biased, it grew more
intolerant of parliamentary secrecy.
The question of publishing parliamentary debates arose again in 1771
when the House of Commons attempted to prosecute several printers who
had published accounts of its proceedings. As with the Walpole episode
of 1738, this attempt by the Commons to reassert its right to secrecy
and the criticism of the secrecy rule that this unleashed in the press,
both occurred during a period where there was a growing sentiment that
the House of Commons majority did not faithfully represent the British
public. Such fears were driven in part by the fact that an archaic
design of constituencies allowed some MPs to be elected by borough
constituencies where corruption was rife. The parliamentary reformer
John Wilkes argued that “disenfranchising the mean, venal and
dependent boroughs would be laying the axe to the root of corruption
and treasury influence, as well as aristocratical tyranny.”40
Speech of John Wilkes in 1776 in favor of
parliamentary reform in Williams 1965.
Proponents of constituency reform
like Wilkes also advocated making Commons proceedings public, precisely
because this would make it easier to hold potentially corrupt
representatives to account.
41 Within the House of Commons, MPs continued to suggest that secrecy
was necessary to preserve freedom of debate. Countering the idea that
the Commons should simply reflect the preferences of the people,
Charles Fox referred to the danger of pandering by suggesting that
“[w]e are bound to promote their true interests in
preference to the dearest desires of their hearts.”42
The printers' case of 1771 ended with the Commons
voting to create a special committee that would be charged with making
recommendations on the case. However, after 1771 publication of
clandestine reports of Commons proceedings became increasingly
widespread.
43 Overall, the evidence from the above three episodes supports
Propositions 3 and 4. Calls for greater transparency in 1738 and 1771
were driven by a growing sentiment that the Commons majority of the
time was a corrupt oligarchy, dominated by financial interests. In
1689, one observes a much different pattern; during a period where
there was less perception of bias, parliamentary secrecy was not
challenged by the public.
Assemblies in the early U.S
For the United States there is clear evidence that increased fears
that legislators were biased led to a shift in public attitudes after
1787, from accepting legislative secrecy to demanding transparency. The
Continental Congress, the Constitutional Convention of 1787, and, in
fact, all state assemblies during the colonial period met in secret. As
had been the case in the British House of Commons, supporters of the
Constitutional Convention's strict secrecy rule argued that it was
necessary to preserve freedom of debate, and hence the quality of
decisions taken. In later years, James Madison provided a particularly
succinct formulation of this argument.
Had the members committed themselves publicly at first, they
would have afterwards supposed consistency required them to maintain their
ground, whereas by secret discussion no man felt himself obliged to retain
his opinions any longer than he was satisfied of their propriety and truth,
and was open to the force of argument.44
Farrand 1911, vol. 3, 478. This passage has been
previously cited by Elster 1991.
The case of the Constitutional Convention is particularly
interesting, because there is solid evidence that the U.S. public
broadly accepted the secrecy of its proceedings. Writing about the
Convention's secrecy rule, the constitutional historian Rossiter
observed “The remarkable thing about this rule is not that it was
so readily adopted, but that it was so rigidly observed by the
delegates and so uncomplainingly accepted by the press and
public.”45
Rossiter provides evidence from
several newspaper accounts that did not challenge the secrecy rule. A
letter from Madison to Thomas Jefferson provides further support, as
Madison observed regarding the Convention's closed-door sessions
that “The public mind is very impatient for ye event, and various
reports are circulating which tend to inflame Curiosity. I do not learn
however that any discontent is expressed at the concealment.”
46 Letter from James Madison to Thomas
Jefferson, 18 July 1787, reported in Farrand 1911, vol. 3, 60. Jefferson, who was in Paris
during the summer of 1787, later wrote to John Adams criticizing the
precedent set by the Convention's secrecy rule, but his opinion
seems to have been atypical. Thomas Jefferson to John Adams, 30 August
1787.
When the time came for individual states to ratify the
Constitution, the fact that the document had been negotiated in secret
did not generate significant opposition.
47 Apart from Patrick Henry's protest against secrecy
during the Virginia ratifying convention, Elliot's
Debates (Elliot 1836), which extensively
documents the ratification debate in individual state conventions,
records no widespread protest at the fact that the Constitution had been
negotiated in secret. For Patrick Henry's speech, see vol. 3,
169–70.
By the early 1790s in the United States there was considerably more
opposition to closed-door legislative sessions than had been the case
in 1787. During the First Congress, the House of Representatives had
opened its debates to the public, but the members of the Senate kept
with tradition by voting to keep their proceedings secret. Closed-door
Senate proceedings soon came under widespread attack. A number of state
legislatures, which had the responsibility of electing senators, called
for the Senate to publish its proceedings as a means of ensuring
accountability of individual senators.48
The
secrecy of Senate debates also came under severe criticism in
newspapers.
49 Faced with substantial outside
pressure, the Senate finally voted in 1794 to open its debates to the
public.
One obvious explanation for the public's shift between accepting
secrecy in 1787 and opposing it a few years later is that by the early
1790s many observers believed that the Federalist administration was
biased in favor of financial speculators and commercial interests. The
financial policies of Alexander Hamilton, involving increased taxation,
creation of a national bank, and full repayment of national debt had
helped create a split between what would become the Federalist Party
and an opposition party, often called the Jeffersonian Republicans. The
study by Stewart demonstrates that opposition to the Federalists led to
the creation of a number of newspapers devoted to criticizing the
administration. In these newspapers, criticism of the Federalist
administration for being unduly influenced by speculators went hand in
hand with calls for openness in government, and in particular the
abolition of secret Senate sessions.50
See Stewart 1969, chaps. II and
XII.
Similarly, McPherson demonstrates that the reason why
southern state legislatures, which were predominantly controlled by
Jeffersonian Republicans, argued for an abolition to Senate secrecy is
that they sought to establish greater control over potentially biased
senators in a Senate dominated by a Federalist majority.
51The French Constituent Assembly of 1789
Unlike in the American and British cases, publicity of legislative
debate was a guiding principle of French representative assemblies from
the first days of the 1789 revolution.52
At the
outset of the revolution, King Louis XVI had attempted to decree that
deliberations would occur in secret to “preserve greater freedom
of action” for the deputies.
53 French Chamber of Deputies, 23 June 1789, in Archivas
Parlementaires n.d., 144.
Apart from a few rearguard attempts
to enforce secrecy of deliberations, the vast majority of French
deputies to the Constituent National Assembly appear to have favored
the principle of publicity. It also seems clear that publicity was seen
as a means of ensuring that decisions within the Constituent Assembly
were not subject to undue influence from the court at Versailles. This
is consistent with Proposition 3 from the theoretical model; a
potential bias involving lobbying pressure by the court led to demands
for proceedings to occur in public.
Consequences of Publicity in Eighteenth-Century
Assemblies
One can examine the consequences of publicity (or secrecy) in
eighteenth-century parliaments by drawing comparisons both across
countries and over time. Given limited space, I consider three specific
comparisons involving the UK, the United States, and France.
In a fascinating article, Elster has contrasted the pattern of debate
at the U.S. Constitutional Convention of 1787, which held all of its
proceedings in secret, with the French Constituent Assembly of 1789,
which also served a constitutional convention, but where all debates
took place in front of an audience.54
Elster
argues that the proceedings in Paris were marked by much more posturing
than occurred in Philadelphia, precisely because the debates took place
in front of an audience. If the presence of an audience for French
Constituent Assembly debates prompted deputies to posture, observers
have suggested the exact opposite for the U.S. Constitutional
Convention. Rossiter claimed that secrecy helped facilitate the
compromises during the Convention that led to a constitution being
successfully adopted.
55 Toward the end of his life, Madison
offered the opinion that the Constitution would never have been
successfully adopted if the debates had been held in public.
56 Madison as reported in an interview by
Jared Sparks (1830), cited in Farrand 1911,
vol. 3, 478.
One can also consider the effects of transparency by comparing events
in legislative assemblies before and after the establishment of
open-door deliberations. With the British House of Commons, as noted
above, publicity gradually became tolerated, even if the House made no
formal decision allowing reporting of its debates. Writing in the
mid-nineteenth century, Mill observed that the shift to publicity had
bound members of Parliament more closely to their constituents.57
Finally, one might also argue that the establishment of publicity
played a role in the evolution of the U.S. Senate after 1794. The study
by Swift shows how the move to open-door debates was one of a series of
institutional changes that helped transform the Senate into a
legislative body that was more directly linked to constituents.58
Current Disputes About Transparency in the
EU
Recently, a debate has emerged about the secrecy of European Council
of Ministers proceedings, and the opposing sides in this dispute have
staked out positions that closely parallel those taken by
eighteenth-century opponents and proponents of “publicity”
in national parliaments. To its critics, the Council is at once the
most powerful and the least transparent institution of the EU. Council
proceedings are not public, and there is no generalized rule allowing
access to transcripts or complete minutes of its meetings. Moreover,
there is a similar lack of access to minutes from the numerous
committees of officials that make recommendations to the Council, such
as COREPER and the EFC. Criticisms regarding secrecy can also be
leveled against the European Council, which is the regular meeting of
EU heads of state and government.59
For
discussions of the Council of Ministers and the committees that report
to it, see Lewis 2003a, 2003b, 2000, 1998; Verdun 2000;
Jacobsson and Vifell 2003; and Wallace 2002. For discussions of the EU's democratic
deficit more generally, see Hix 1999;
Habermas 1998; and Moravcsik 1994, 2002.
This
critical view of current Council practice is not shared by all
observers. Moravcsik argues that concerns about the lack of
transparency of EU institutions such as the Council are overstated.
60 Moreover, a majority of EU
governments have claimed that secrecy is necessary to facilitate
efforts to reach compromises within the Council of Ministers.
Background to the Dispute
In 1993, the European Commission and the Council of Ministers
established a “Code of Conduct” that specified that the
European public should have “the widest possible access to
documents.” Under this code the Council also amended its internal
rules to clarify exactly when such access would be granted. In so doing
the Council gave itself wide latitude to keep any document secret by
stating that “access to a Council document may be refused in
order to protect the confidentiality of the Council's
proceedings.”
Using the new procedures, John Carvel of The Guardian
newspaper requested a number of documents from the Justice Council, the
Agriculture Council, and the Social Affairs Council, but only the
latter accepted the request, and it was later learned that the Social
Affairs Council had only agreed to release the documents as a result of
an administrative error. In the letter refusing access to other
documents, the Council noted that granting the access Carvel requested
would “fail to protect the confidentiality of its
proceedings,” and that “the documents in question contain
confidential information relating to the position taken by the members
of the Council during its deliberations.”61
Council letter of 17 May 1994 cited in Judgment of the
Court of First Instance.
Making previously confidential
information public was precisely the motivation behind the request,
however, and so Carvel and
The Guardian took the case to the
European Court of First Instance. This legal action was supported by
the governments of the Netherlands and of Denmark, and it also received
support from the European Parliament. A majority of EU governments
opposed the action.
The Court of First Instance ultimately found against the Council, but
because of the content of its judgment, the case did relatively little
to reduce the secrecy of Council deliberations. The Court found that in
deciding to refuse access to certain documents, the Council had not
exercised the proper discretion necessary, as stated in the
Council's “Code of Conduct.” The Council's
refusal to grant access to The Guardian was instead a flat
rejection with little explanation or justification. However, after this
judgment, the Council still retained the right to refuse access to any
internal document, as long as it provided evidence that it
“exercised proper discretion” when deciding to preserve
secrecy.
Since the 1994 case, the Council has made some effort to become more
open, with increased access to documents, and a move to allow portions
of several meetings per year to be viewed on closed circuit television.
In June 2002, EU heads of state and government responded to increased
demands for transparency by declaring that certain additional
deliberations of the Council of Ministers should be made public. This
concerns meetings where the Council is acting in accordance with the
procedure for codecision with the European Parliament. However, the
majority of Council discussions continue to take place behind closed
doors, and in particular those sessions that might prove controversial.
So, for example, a portion of a ministers of finance meeting of July
2003 was televised, but the televised portion was restricted to a broad
discussion of the Council presidency's economic goals, whereas a
more critical discussion of the Economic and Monetary Union's
Stability Pact enforcement was kept secret.62
Council of the European Union 2003.
In addition, the work of COREPER,
which prepares and frequently adopts Council legislation, also remains
confidential. Secrecy also continues to prevail on other committees
that report to the Council, such as the EFC.
63Causes of Council Secrecy
When one considers the positions taken by those who favor and those
who oppose making EU Council of Ministers deliberations public, one
observes that they conform closely to the predictions of Propositions 3
and 4. Calls for transparency have emerged during a period of increased
concern about a disjuncture between the preferences of EU citizens and
representatives on the Council. Proponents of transparency have
emphasized the need for transparency to increase accountability, while
those who favor secrecy have emphasized the efficiency benefits of
closed-door bargaining.
Considering the proponents of closed-door sessions first, the 1994
case launched by The Guardian provides a particularly
interesting opportunity to examine the motivations behind opposition to
open-door bargaining, because the Council of Ministers was forced to
provide a formal defense of its procedures. In its rejoinder to the
Court of First Instance, the Council defended its desire for secrecy in
the following terms,
The Council normally works through a process of negotiation
and compromise, in the course of which its members freely express their
national preoccupations and positions. If agreement is to be reached, they
will frequently be called upon to move from those positions, perhaps to the
extent of abandoning their national instructions on a particular point or
points. This process, vital to the adoption of Community legislation,
would be compromised if delegations were constantly mindful of the fact that
the positions they were taking, as recorded in Council minutes, could at any
time be made public through the granting of access to these documents,
independently of a positive Council decision.64
Emphasis in original. Council of the European Union 1994. A related statement is also cited by Fingleton
and Raith 2002.
The Council's defense closely follows the logic of the
theoretical model described above, as it is based on the idea that open
negotiations might make national representatives less likely to move
away from their initial positions, triggering a greater incidence of
bargaining breakdowns. One might ask whether this is an obvious result;
in a democratic age what other arguments could the Council make in
favor of secrecy? In fact, a plausible alternative strategy would have
been to argue that opening up Council debates would risk revealing
information to third parties. This would be particularly relevant in
the foreign policy domain, where the Council debates actions to take
with respect to nonmember states. It is also possible that the Council
might have relied more heavily on the simple argument that the Council
is a meeting between ministers, and cabinet meetings between ministers
at the national level invariably take place in secret.
Turning to the proponents of transparency, one sees that the pattern
of support for making EU Council of Ministers public closely resembles
that observed for the eighteenth-century cases discussed above.
Widespread protests about the Council's lack of transparency have
emerged since the beginning of the 1990s, during a period of heightened
concerns that Council representatives may not be faithfully
representing the preferences of their national electors. This has also
been a period where the EU has acquired an expanded role over policy,
making any potential disjuncture between citizen and representative
preferences more salient. In contrast, during earlier periods, like the
mid-1980s, one observes both that European public opinion was more
favorable to Council decisions, and during these earlier periods
Council secrecy provoked little public criticism. In what follows I
draw on evidence from public opinion polls in particular to support
these claims, which are consistent with Proposition 3.
I have already provided evidence that recently there have been
demands for making the Council of Ministers more transparent. One
potential explanation for these demands is that since the early 1990s
the European public has felt poorly represented by EU institutions, and
in particular by the Council of Ministers. A Eurobarometer survey
conducted in 1992, revealed that 71 percent of EU citizens felt there
was “insufficient democratic influence in EC decision
making.”65
Eurobarometer 38,
European Commission, December 1992, 71.
Subsequent surveys
have revealed that when citizens are asked whether they trust specific
EU institutions, the Council of Ministers regularly scores the lowest
among major EU bodies. In a survey conducted in 2000, 53 percent of
respondents reported that they “tend to trust” the European
Parliament, and 47 percent trusted the Court of Justice, but only 39
percent reported trusting the Council of Ministers.
66 Eurobarometer 54, European Commission,
November–December 2000, 55.
This perception of a problem
of insufficient representation within the EU is exacerbated by
increased pessimism that certain policies decided upon at the EU level,
such as internal economic liberalization, will provide a benefit to EU
citizens. In a 2001 Eurobarometer poll, only 24 percent of respondents
thought that economic liberalization within the EU would improve their
personal financial position, and only 26 percent thought it would
improve their quality of life.
67 Eurobarometer 55.1, “Europeans, Globalisation, and
Liberalisation,” spring 2001, 34–49.
When one instead considers the EU during the mid-1980s, there is much
less evidence of either demands for transparency of the Council of
Ministers or fears that representatives might have been biased. The
best evidence that making Council of Ministers discussions public was
not a salient issue may be the simple fact that the Council did not
even deem it necessary to have a formal policy on
“transparency” until 1993. Further evidence is provided by
the fact that before the 1990s there was not significant discussion in
major European newspapers of the potential need to make Council of
Ministers discussions public. This greater acceptance of Council
secrecy appears to have coincided with lower levels of mistrust of EU
institutions, and less evidence of pessimism about EU policies like
internal economic liberalization. The nature of available data makes it
difficult to prove this statement conclusively. It would be extremely
useful to have access to Eurobarometer poll data from the mid-1980s on
“trust in EU institutions,” but Eurobarometer surveyors
only began asking questions on this issue during the 1990s, after
mistrust emerged as a perceived problem.68
Eurobarometer 38, conducted in 1992, was the first
Eurobarometer survey to ask whether citizens felt they had sufficient
influence over EC decision making. Eurobarometer 41, conducted in 1994,
was the first survey to ask whether citizens believed that decisions
taken by specific EU institutions were in their interest.
Nonetheless, there is public opinion evidence from the 1980s that
suggests that the policy of internal economic liberalization embodied
in the Single European Act (SEA) of 1986 had substantial public support
at the time. In 1986, soon after the negotiation of the SEA, a
Eurobarometer poll asked whether citizens believed that the
liberalization of flows of people, goods, and capital implied by the
act was a good thing. Among those who replied, 83 percent agreed at
least to some extent.
69 Eurobarometer 25,
European Commission, June 1986, 20–22.
Results of a poll
from 1987 show similarly strong support.
70 Eurobarometer 28, European Commission, December 1987,
36–40.
Though the wording of questions differs between
the survey on economic liberalization from 2001 and these earlier
surveys, given the dramatically different responses, it would seem
difficult to contest the fact that there is now a much larger
disjuncture between the preferences of EU citizens and the policies
decided by the EU Council with regard to liberalization, than was the
case in 1987.
Consequences of Council Secrecy
One of the implications of secrecy for the EU Council of Ministers is
that there is less information available about Council deliberations
than is the case for more public institutions within the EU, such as
the European Parliament. This makes it difficult to ascertain with
certainty whether the closed-door deliberations of the Council are
indeed characterized by the lower degree of posturing, as well as by
greater departures from public opinion that would be consistent with
Propositions 1 and 2. It is nonetheless significant that numerous
studies of the Council of Ministers have made reference to its
“culture of compromise” and its club-like atmosphere.
Authors have in fact provided descriptions of Council bargaining that
clearly conform to the idea that closed-door bargaining facilitates
compromise.
Among scholars who specialize in the workings of the Council of
Ministers and its subsidiary bodies, Lewis has argued that COREPER has
become a highly efficient bargaining forum precisely because it works
in secret.71
Wallace makes this argument about
the workings of the Council of Ministers more generally, suggesting
that the Council's “club-like” atmosphere helps
facilitate agreements, while she notes that this method of functioning
has unfortunately depended on a lack of transparency for
proceedings.
72 In another contribution, Lewis
describes several other features of behavior within COREPER that
strongly support the idea that closed-door bargaining helps facilitate
compromise.
73 For one, he observes that it is
common for national representatives to COREPER to depart from original
instructions by their governments. It was for this reason that the
German representative to COREPER jokingly suggested that he was often
referred to as the “Permanent Traitor” in his home capital,
rather than the “Permanent Representative.” These
observations support the Council's own deposition in
The
Guardian case, suggesting that secrecy was necessary precisely to
allow national representatives to depart from instructions. Second, in
a case study of the 1994 Local Elections Directive, Lewis provides
convincing evidence that EU governments sought to keep the discussion
of this directive limited to the secretive environment of COREPER,
precisely because it was feared that publicity would inhibit
governments from compromising.
74 Beyond facilitating compromise, it has also been argued that the
secrecy of certain committees that report to the European Council, and
most notably COREPER and the EFC, encourages participants to engage in
free deliberation about the advantages and disadvantages of different
policies. This deliberation is seen as potentially leading to shifts in
preferences and higher quality decisions. Jacobsson and Vifell report
that the highly secretive EFC has provided an environment within which
representatives can discuss politically sensitive policy options
openly.75
Jacobsson and Vifell 2003. See also Lewis 2003b on deliberation within COREPER and Joerges
and Neyer 1997 on deliberation as a method of
governance in the EU.
It also seems clear that in committees
such as COREPER and the EFC, participants have a greater tendency to
make unprepared remarks, rather than relying on the sort of
pre-prepared statements that ensure accuracy but which can stifle
debate. In contrast, in more open committees there tends to be far less
real debate. In their survey on the extent of deliberation in EU
bodies, Jacobsson and Vifell arrive at the interesting conclusion that
“the more closed the forum, the more openness in
discussion.” In the introduction to an edited volume on decision
making in different European institutions, Checkel makes a similar
proposition, suggesting that discussions based on persuasion will be
more likely to occur in “less politicized and more insulated
in-camera settings.”
76 While scholars suggest that secrecy helps increase the efficiency of
the Council of Ministers in terms of reaching agreements, they also
emphasize that secrecy of Council proceedings inevitably breeds
suspicion about the agreements that are struck. Wallace suggests that
the closed-door nature of proceedings has “the potential to
estrange other national politicians and national parliaments from the
EU process.”77
Likewise, Verdun has argued
that the secrecy of the EFC, which reports to the Council of Ministers,
undermines democratic accountability.
78 These
arguments parallel the more general claims made by Habermas about
secrecy undermining public confidence in EU institutions.
79 The risk of secrecy undermining pubic confidence in EU decisions was
graphically illustrated when it was revealed during the 2002 French
presidential campaign that the prime minister, Lionel Jospin, and the
president, Jacques Chirac, had agreed at a European Council meeting to
a declaration that all EU countries should increase the average
retirement age by five years.80
“L'Europe demande une
accélération de la réforme des
retraites,”Le Monde, 4 April 2002.
Independent
of whether one sees this measure as desirable, the declaration
contradicted Chirac's and Jospin's positions on the issue as
stated to the French national press. A recent episode involving a
documentary film on the Council provides further clear evidence that
Council representatives often take public positions that differ from
their positions as privately expressed in the Council. With the
agreement of the president of the EU Council, in December 2002, a
Danish film crew shot a documentary on the Copenhagen Council summit
that included scenes with ministers having frank exchanges about
enlargement questions in particular, and without apparent knowledge
that the film would subsequently be aired on Danish TV. When the film
was broadcast in April 2003, it provoked a storm of controversy, most
notably for Germany's foreign minister, Joschka Fischer, who had
made negative comments about potential Turkish membership in the EU,
despite having supported Turkey's application in public.
81 “Fury Sweeps Capitals of Europe as TV
Shows What Summit Leaders Really Said,” Financial Times,
19 April 2003, 7.
More generally, one could suggest that the
secrecy of Council of Ministers proceedings makes it easier for
national representatives to blame certain policy decisions on
“Brussels” even when they themselves have participated in,
and even supported such decisions.
In sum, the recent dispute over secrecy of the EU Council of
Ministers illustrates well the trade-offs from transparency. Keeping
proceedings behind closed doors appears to have fostered a
“culture of compromise,” but this has come at the cost of a
perceived lack of accountability. This perception has been particularly
problematic in a period where there is increased suspicion that
national representatives may not be faithfully defending the views of
their constituents during Council proceedings.
Conclusions
I have argued in this article that the issue of transparency in
political bargaining presents a trade-off for society. Bargaining that
takes place in public helps ensure that representatives propose
policies preferred by their constituents. But open-door bargaining also
encourages representatives to posture by adopting overly aggressive
bargaining positions that increase risks of breakdown in negotiations.
As a result, a shift to open-door bargaining may increase
accountability while nonetheless resulting in undesirable outcomes for
constituents. I have suggested that when there is a strong belief that
representatives may not share the policy preferences of their
constituents, then transparency will be preferable, because the
expected benefit from constraining the actions of biased officials will
exceed the expected costs of prompting unbiased officials to posture.
Empirical evidence from current debates about making the EU Council of
Ministers public, as well as eighteenth-century debates about
“publicity” for national parliaments supports these
propositions. Finally, I expand on these conclusions by making the
following three points about international institutions.
First and most directly, authors who emphasize the importance of
establishing “global governance” and
“transparency” in international institutions should
recognize that openness can involve costs as well as benefits. Most of
the existing literature has focused almost exclusively on the latter.
As a result, in the EU case, rather than speaking of a
“democratic deficit,” as far as transparency is concerned
it may make more sense to follow Moravcsik by suggesting there is a
“democratic dilemma.”82
When
advocates of transparency argue that international institutions should
adhere to the same norms of transparency that are presumed to prevail
within domestic institutions, they often overlook the fact that the
degree of openness in domestic institutions actually varies quite
significantly, and in ways that many would see as being desirable. So,
for example, parliamentary discussions now take place in public, but
jury deliberations and court deliberations take place behind closed
doors. There is no reason to believe that there should not be a similar
diversity in levels of transparency in international institutions.
A second point is that the optimal design of international
institutions may change over time. Rather than deciding once and for
all whether it is better to have EU cooperation be based on closed-door
institutions, and rather than deciding whether practices such as the
“club model” of multilateral cooperation are optimal in an
absolute sense, it may be more accurate to suggest that these secretive
forms of international cooperation were desirable at one time, but they
are suboptimal today, in an era where an increasing number of citizens
perceive a bias on the part of their representatives. One might use
evidence on growing perceptions of the EU's “democratic
deficit” to support this contention, although in this article I
have not demonstrated explicitly that the costs of Council of Ministers
secrecy now outweigh the benefits (I have argued simply that there are
both costs and benefits and perceptions of the costs have grown more
acute). If the optimal level of transparency in bargaining changes over
time, it would also be worth ascertaining whether there are clear cases
where institutions have become less transparent, and this has been
accepted by the public. While the model presented here implies that
this could occur, empirical observation suggests that moves in the
direction of greater transparency may be much more frequent.
A third and final point concerns the literature on deliberation.
Advocates of deliberative democracy have emphasized that public
deliberation improves the quality and the legitimacy of decisions
taken. While the model I have presented here is one of bargaining,
where there is no uncertainty about the link between policies and
outcomes, my propositions nonetheless raise an important question about
the benefits of holding deliberations in public. To the extent that
deliberation depends on the willingness of participants to accurately
reveal private information about policies, then this might actually be
less likely to occur in a public context. Much as reputational concerns
give the bargainers in my model an incentive to ignore their private
information about the optimal offer, participants in an open-door
deliberative proceeding might refrain from accurately revealing their
private information for exactly the same reasons.83
See Stasavage 2004 for a model
of deliberation that compares public and private settings.
It
is particularly interesting to note here that observers of EU
committees have suggested that those committees that are the most
secretive tend to be those that also are characterized by the most
actual deliberation. Future research could explore this question more
extensively.
Appendix
In order to present formal proofs of Propositions 1 to 4, it is
necessary to first demonstrate conditions for existence and uniqueness
of the different equilibria under open-door and closed-door bargaining.
As stated in the text, I restrict representatives to four possible pure
strategies {truthful, d, 0.5, 1 − d}. Any
equilibrium must consist of a set of offers
{oA-unbiased,
oA-biased,
oB}and a belief
for each offer
and outcome
that do not occur in equilibrium. Given that
RB is unbiased by assumption,
RB's equilibrium behavior is
straightforward. Both under open-door and closed-door bargaining, in
any equilibrium where a biased RA makes an
offer other than (1 − d),
RB will always play the truthful strategy.
In any equilibrium where a biased RA
proposes (1 − d), RB will
play the truthful strategy unless the prior belief that
RA is unbiased is low p <
.
Otherwise, RB will always propose (1
− d). In what follows I consider the possible
equilibrium offers of RA. For a biased
representative RA the equilibrium offer
o that leads to an equilibrium outcome x must satisfy
the following inequality where
is an alternative offer and
is an alternative outcome. The relevant expression for the unbiased
representative follows the same form.
Existence of Equilibria under Open-Door Bargaining
Truthful and pooling equilibria
Conditions for existence of a “truthful” equilibrium and
of a “posturing” equilibrium have already been presented in
the text. Alternative pooling equilibria, which would include pooling
on 0.5 or pooling on (1 − d), can be ruled out using the
intuitive criterion of Cho and Kreps.
84
Consider a strategy profile of pooling on 0.5. If the unbiased
representative received a signal
d, he or she would receive an
expected policy gain by deviating and proposing
d. As a
result, the unbiased representative would deviate unless there was an
out of equilibrium belief for the action that satisfied
.
This is an implausible belief, given that that a biased representative
could never gain by proposing d. One can rule out an equilibrium
of pooling on (1 − d) using a similar argument.
Separating equilibria
Given the four possible strategies {d, 0.5, (1 −
d),t} (where t stands for
“truthful”), there are twelve possible candidates for
separating equilibria in pure strategies. All but two of these can be
quickly ruled out. The biased representative could deviate from any of
the following strategy profiles and gain in terms of policy
(reputation) without losing in terms of reputation (policy):
Likewise, the unbiased representative would gain in terms of policy
without losing in terms of reputation by deviating from {(1 −
d, 0.5),(1 − d, d)}.
This leaves two possible separating equilibria. In the first of
these, the biased representative always proposes (1 − d)
while the unbiased representative plays the truthful strategy. It is
trivial to show that an unbiased representative could never gain from
deviating from this equilibrium. A biased representative will not
deviate by offering
,
as long as α((1 − d) − 0.5) ≥ 1 − α,
which implies that this equilibrium exists as long as
.
In the second possible separating equilibrium, the unbiased
representative plays truthfully, while the biased representative always
offers 0.5. The biased representative will not deviate from this
equilibrium by offering (1 − d) as long as the following
inequality is satisfied.
By the intuitive criterion, it is implausible that the out of
equilibrium belief
− d) could be anything other than 0, because an unbiased
type would never have an incentive to make this offer.
The unbiased representative would not deviate from offering
o = 0.5 (if the representative received a signal 0.5) by
offering d as long as the following inequality is satisfied (a
biased representative would never deviate by offering d if the
following inequality was satisfied).
One can conclude that this separating equilibrium will exist when
Irrespective of the value of α, the above inequality can only be
satisfied when
.
One can conclude that posturing is the unique equilibrium under open-door
bargaining when either of the two conditions in expressions (14) or (15) is
satisfied and thus neither of the separating equilibria exist. This implies
that at a minimum, the posturing equilibrium is unique whenever p
< 2/3 and when α < 2/3.
Existence of Equilibria under Closed-Door Bargaining
Truthful and pooling equilibria
One can rule out a truthful equilibrium, or any pooling equilibrium,
under closed-door bargaining using the following reasoning. In any
truthful or pooling equilibrium, the biased and unbiased
representatives make the same proposals with the same probabilities,
and as a result Pr(u|x) = p. The
posterior for any offer that is made by RB
but which is not made by either the biased or unbiased representative
will also be p. Given that RB
will always propose o = (1 − d) with positive
probability, a biased representative will always deviate from any
strategy other than proposing (1 − d).
Separating equilibria
One can use similar reasoning to rule out any separating equilibrium
where the biased representative makes a proposal other than (1 −
d). With any of the three alternative strategies to proposing
,
given that RB proposes (1 − d)
with positive probability, and given that an unbiased representative would
never propose (1 − d) in equilibrium, this implies that the
following condition must be satisfied for
to constitute an equilibrium proposal
.
This implies that
must also hold, but there is no possible pair of strategies that could satisfy
this requirement, without prompting one of the two types to deviate.
In the unique pure-strategy equilibrium under closed-door bargaining,
the unbiased representative plays the truthful strategy and the biased
representative proposes (1 − d). The biased
representative will not deviate from this equilibrium by offering
d, instead of (1 − d), if the following
inequality is satisfied (this applies as long as
RB plays the truthful strategy, otherwise
different values apply for Pr(x = 1 −
d|u) and Pr(x = 1 −
d|b).
The biased representative would not deviate by offering 0.5 as long as
An unbiased representative would not deviate by offering d
if the signal was s = 0.5 as long as
Proof of Proposition 1 Posturing is the unique
equilibrium under open-door bargaining when reputational concerns are
sufficiently strong to satisfy either expression (14) or (15). Under
these conditions, when compared with the unique equilibrium under
closed-door bargaining, a shift from closed-door to open-door will
always result in the biased representative shifting from proposing (1
− d) to proposing d. An unbiased representative
will shift from the truthful strategy to always proposing d.
Proposition 1 will also hold for those parameter values for which a
separating equilibrium exists under open-door bargaining where the
biased type proposes 0.5 and the unbiased type plays truthfully.
Proof of Proposition 2 Proposition 2 is proved by
demonstrating the conditions for uniqueness of the posturing
equilibrium (that is, when either expression (14) or (15) is
satisfied).
Proof of Proposition 3 When the inequality in
expression (14) holds, posturing will be the unique equilibrium under
open-door bargaining, and, as demonstrated in expressions (6) and (7)
in the text, this equilibrium will provide higher expected utility for
the public when compared with the unique equilibrium under closed-door
bargaining. When expression (15) holds, posturing is the unique
equilibrium under open-door bargaining, and the public will earn higher
expected utility from closed-door bargaining.
Proof of Proposition 4 Whenever the posturing
equilibrium is the unique equilibrium under open-door bargaining, an
unbiased representative will always obtain a more preferable policy
outcome under closed-door bargaining, as long as (q(1 −
d) + (1 − q)d) > 0.5, which must
hold given that, by assumption, 0.5 < q < 1. The
unbiased representative will also obtain a higher reputational payoff
under closed-door bargaining as long as the following inequality is
satisfied. This inequality holds for all possible parameter values of
p and q.
A biased representative will prefer closed-door bargaining as long as
the following inequality is satisfied:
This inequality is satisfied for all parameter values for which the
two equilibria exist.
