Sie sind auf Seite 1von 48

Rationality and Society

http://rss.sagepub.com/

Hostage Posting as a Mechanism of Trust : Binding,


Compensation, and Signaling
Werner Raub
Rationality and Society 2004 16: 319
DOI: 10.1177/1043463104044682

The online version of this article can be found at:


http://rss.sagepub.com/content/16/3/319

Published by:

http://www.sagepublications.com

Additional services and information for Rationality and Society can be found at:

Email Alerts: http://rss.sagepub.com/cgi/alerts

Subscriptions: http://rss.sagepub.com/subscriptions

Reprints: http://www.sagepub.com/journalsReprints.nav

Permissions: http://www.sagepub.com/journalsPermissions.nav

Citations: http://rss.sagepub.com/content/16/3/319.refs.html

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


HOSTAGE POSTING AS A MECHANISM
OF TRUST

BINDING, COMPENSATION, AND SIGNALING

Werner Raub

ABSTRACT

We study voluntary hostage posting – pledging a bond – as a com-


mitment mechanism promoting trust, including trust in economic
exchange. A hostage can promote trust by binding the trustee
through reducing his incentives for abusing trust, by providing com-
pensation for the trustor in case trust is abused, and by serving as
a signal for the trustor about unobservable characteristics of the
trustee that are related to the trustee’s opportunities and incentives
for abusing trust. We provide an integrated model that allows for a
simultaneous analysis of how hostages promote trust through bind-
ing, compensation, and signaling. We model hostage posting as a
mechanism of trust using a game with incomplete information
and uncertainty. Our theorems provide conditions for equilibria
such that a hostage is posted by the trustee and induces the trustor
to place trust that is subsequently honored by the trustee. The article
shows that equilibrium selection problems are not severe: the equi-
libria are unique or there are only few other equilibria with less
appealing properties. Hence, the results can be used for predictions
on trust based on hostage posting among rational actors.

KEY WORDS . commitments . cooperation . game theory .


hostages . trust

1. Introduction: Trust Based on Hostages

Consider some examples of trust problems and of hostages as a


mechanism for solving these problems.

Example 1: A guarantee solving the trust problem when buying a


secondhand car. Purchasing a secondhand car is a well-known

Rationality and Society Copyright & 2004 Sage Publications. Vol. 16(3): 319–365.
www.sagepublications.com DOI: 10.1177/1043463104044682

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


320 RATIONALITY AND SOCIETY 16(3)

example of a trust problem in a buyer–seller relation (Akerlof


1970; Dasgupta 1988). The trustor is the buyer. She has to
decide whether or not to buy a used car of unknown quality.
The trustee is the seller. He is better informed on the quality of
the car than the buyer. If the trustor places trust and buys the
car, she may end up with a lemon rather than a peach. She prefers
a peach rather than having no car but would prefer not to own a
car rather than a lemon. The trustee prefers to sell a peach rather
than selling no car at all, but he may have a lemon he could sell
with even higher profit and thus may have an opportunity and
an incentive to abuse trust. One way to solve the problem is for
the seller to offer a guarantee on the car. If the car has a defect
while it is under guarantee, the trustee has to repair the car at
his own cost.
Example 2: A deposit solving the trust problem in car rentals. Car
rentals are associated with trust problems in the sense that the
rent-a-car firm has to trust that the client will not drive recklessly.
In this example, the rent-a-car firm is the trustor. She prefers
placing trust and renting the car to a careful driver who returns
the car in good shape over withholding trust and missing a
client. The client is the trustee. He prefers honoring trust by driv-
ing carefully to having no car at all. However, the client may face
an incentive for reckless driving (‘who cares about a few extra
scratches’). A deposit that the client receives back if he returns
the car in good shape can be used to solve the problem.
Example 3: Employment relations and investments in human capital.
Employers often have to decide on making investments in an
employee that are largely relationship-specific (see Becker 1964
on investments in human capital). For example, the employer
provides training and schooling. These investments have to be
depreciated should the employee decide to quit. Here, the
employer is the trustor, while the employee is the trustee. If the
employer places trust and invests, the employee may have to
decide between honoring trust through a durable relationship
with the firm and abusing trust by accepting an outside offer.
Assuming that the employer has invested in the general human
capital of the employee, it seems likely that the employee will
have opportunities and incentives for abusing trust. After all,
through his increased general human capital the employee
becomes more attractive for other employers who, moreover, do

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 321

not need to recover the costs of training and schooling. One way
of solving this problem is a contract between the employer and the
employee stipulating that the employee has to pay at least part of
the costs of training and schooling if he quits prematurely.
Another way of solving the problem is non-contractual. The
employee moves and acquires real estate close to his job. Thus,
if he quits prematurely and accepts an offer from another and
far away employer, he would have to move again and incur the
financial as well as social costs associated with moving.

Common Characteristics of the Examples: Trust Problems and


Solving Trust Problems through Hostage Posting
The examples have core features in common. First, a trust problem
is involved. A trustor has to decide between placing and withholding
trust. The trustor knows that she incurs a risk by placing trust. If
trust is placed, the trustee may subsequently abuse rather than
honor trust. If trust is honored, the trustor is better off (at least –
as we shall assume – with some positive probability) than if no
trust is placed. However, if trust is abused, the trustor is worse off
than if she withholds trust. On the other hand, the trustee is at
least likely to have an opportunity and an incentive to abuse trust,
while withheld trust is even worse for the trustee than honoring
trust that has been placed. The trust problem is therefore twofold.
First, by placing trust, the trustor incurs risks, such as trust being
abused. Second, if the trustor decides not to place trust, both trustor
and trustee could have been better off had trust been placed and
honored. Trust is a ‘lubricant’ (Arrow 1974) of exchange and
increases efficiency. Also, however, abusing trust is a typical case
of ‘opportunism’, in the sense of Williamson (1985: 47). Therefore,
the trustor has an incentive to protect herself by withholding trust.
The second common feature of these examples is the mechanism
that could solve the trust problem. In each case, the trustee may
try to solve the problem by voluntarily posting a hostage, in the
sense of pledging a bond. The hostage of the dealer in used cars is
the guarantee. In the car rental example, the deposit is the hostage
of the client. To induce his employer to invest in his training and
schooling, the employee can post a hostage by moving or through
a contract that forces the employee to pay back training costs in
case of a premature quit.

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


322 RATIONALITY AND SOCIETY 16(3)

A hostage can solve the trust problem in three different ways.


First, the hostage reduces the incentives of the trustee to abuse
trust. Selling a lemon becomes less attractive for the seller if he
pays the cost of repairs. Incentives for reckless driving are reduced
because one risks the deposit. Abusing investments of the employer
comes with reimbursement costs or costs of moving for the
employee. Since the trustee’s incentives for abusing trust are
reduced, the trustor may be more willing to place trust.
Second, after hostage posting by the trustee, the trustor may be
more willing to place trust because the hostage includes compen-
sation for the trustor if the risks ‘materialize’ after placement of
trust. Having bought a lemon is less of a problem for the buyer if
the seller pays at least the cost of repairs. The deposit compensates
the rent-a-car firm for at least some of the damage from reckless
driving. If the employee pays some of his training costs in the case
of a premature quit, the damage to the employer from the premature
quit is reduced. Interestingly, however, the employee’s hostage
‘moving and acquiring real estate close to his job’ does not provide
compensation for the employer if the employee quits.
Third and finally, the hostage may serve as a signal for the trustor
about unobservable characteristics of the trustee that are related to
opportunities and incentives for abusing trust. There could be used
car dealers who do not sell lemons, either because they do not have
lemons to sell or because they do not sell them for one reason or
another even if they have them. Providing a guarantee for a used
car may be a signal for the buyer that the dealer does not have
lemons or at least does not sell them. The move of the employee
close to his job may signal that he does not have and does not
expect an outside offer. Assume the trustor does not know a priori
what kind of trustee she faces. If the trustor has reason to conclude
from the fact a trustee has posted a hostage that the trustee has no
opportunity or no incentive to abuse trust, hostage posting will
again solve the trust problem.
Formal specifications of trust problems and hostages as a
mechanism of solving such problems will be provided in due time.
Before we turn to formal model building, however, it is useful to pro-
vide some more intuition. With respect to trust problems, note that
we focus on situations such that one and only one actor, the trustee,
has opportunities and incentives for opportunistic behavior: the
trustor places or withholds trust, the trustee honors trust or behaves
opportunistically by abusing trust. We thus disregard situations

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 323

with two-sided incentives for opportunism such as the Prisoner’s


Dilemma (we return to this issue in the concluding discussion).
Hostage posting is a typical example of a ‘strategic move’ by incur-
ring a commitment in Schelling’s (1960) sense. Schelling accentuated
the use of hostages and other commitments in bargaining contexts.
He showed that a commited actor can frequently induce a bargain-
ing outcome that is favorable for himself and unfavorable for the
partner. Thus, counterintuitively, an actor often has incentives to
incur a commitment and to bind himself voluntarily: ‘the power to
constrain an adversary may depend on the power to bind oneself
[. . .] in bargaining, weakness is often strength, freedom may be free-
dom to capitulate, and to burn bridges behind one may suffice to
undo an opponent’ (Schelling 1960: 22). Other than Schelling, we
analyze how and when commitments such as hostages increase effi-
ciency by allowing for mutually beneficial adjustments. The hostage
modifies subsequent incentives for placing or withholding as well as
honoring or abusing trust. By posting an appropriate hostage, the
trustee incurs a commitment that serves as a safeguard for the trus-
tor if she decides to place trust (Schelling 1960; Williamson 1985).
Williamson (1985: 168) argued that hostages can have ‘ex post’
bonding effects as well as ‘ex ante’ screening effects. This coincides
with our distinction between reducing the incentives for the trustee
to abuse trust through hostage posting and using hostages as signals
for unobservable characteristics of the trustee. Also, Williamson
(1985: 172) highlighted the distinction between the value of the hos-
tage for the actor who posts the hostage and the value of the hostage
for the other party. This coincides with our distinction between
reducing the incentives for the trustee to abuse trust through hostage
posting and compensating the trustor if risks ‘materialize’ after
placement of trust.

Trust Based on Embeddedness and Hostages as a Commitment


Device Provided by Institutional Embeddedness
Sociologists often argue that trust depends on the embeddedness
(see Granovetter 1985, 1992) of a trust relation in a social context.
Different dimensions of embeddedness can be distinguished (Buskens
and Raub 2002). Dyadic embeddedness refers to repeated encounters
between trustor and trustee. Network embeddedness refers to rela-
tions of trustor and trustee with third parties. Two basic mechanisms
through which dyadic embeddedness and network embeddedness

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


324 RATIONALITY AND SOCIETY 16(3)

promote trust are learning and control. Learning is based on past


interactions. Given dyadic embeddedness, the trustor can learn
from her own previous encounters with the trustee about the trustee’s
trustworthiness as well as his competence (e.g. Kreps et al. 1982;
Granovetter 1985; Coleman 1990). Given network embeddedness
in the sense that the trustor has access to other trustors of the trustee,
the trustor can learn from these other trustors about the trustee (Burt
and Knez 1995; Buskens 2002). Control is based on the prospect of
future interactions. Given dyadic embeddedness and the prospect
that trustor and trustee will meet again, the trustee has to take
into account that the trustor’s trust may be conditional (Taylor
1987 [1976]; Axelrod 1984). If the trustee behaves in a trustworthy
manner today, the trustor may reward this by placing trust again
in the future. Conversely, if the trustee abuses trust today, the trus-
tor may punish the trustee by withholding trust in future encounters.
Given network embeddedness, trustworthy behavior today may be
rewarded through the trustful behavior of other trustees in the
future. Abusing trust today may be punished in the sense that other
trustors will withhold trust in future interactions with the trustee
(Raub and Weesie 1990; Buskens 2002). If the long-term losses
due to untrustworthy behavior today are higher than the short-
term gains from abusing trust, the trustor and third parties can con-
trol the trustee and induce the trustee to honor trust by placing trust
conditionally.
Dyadic embeddedness and network embeddedness are sometimes
but not always sufficient for the solution of trust problems. Extreme
cases are one-shot encounters and no exchange of information
between trustors. Also, previous encounters of the trustor or of
third parties with the trustee may provide insufficient information
on the trustee’s trustworthiness and competence. The prospect of
future sanctions may be insufficient to deter the trustee from abusing
trust today, certainly so if abusing trust today is a ‘golden opportu-
nity’ and particularly attractive. Hostages can be used to solve trust
problems even if learning and control do not suffice to promote
trust. Hostage posting becomes feasible under institutional embed-
dedness of the trust problem (Weesie and Raub 1996). We assume
an institutional context as given that provides opportunities for
the trustee to post a hostage ex ante, before the trustor decides to
place or withhold trust. The context providing the opportunity to
post a hostage is an exogenous condition in this analysis. By using
this opportunity and posting a hostage, the trustee creates a private

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 325

institution for his interaction with the trustor. As Coleman (1990)


put it, the trustee creates a ‘constructed social environment’ that
promotes trust. Thus, institutional embeddedness provides oppor-
tunities for private ordering (Williamson 1985; Macaulay 1986) of
relations. The private institution itself – the hostage – is endogenous.
We study conditions in which these private institutions result from
individually rational equilibrium behavior and are thus self-
enforcing (see Schotter 1981 and Calvert 1995 for the distinction
between institutions as exogenous constraints and as outcomes of
equilibrium behavior). Therefore, we do not assume that an external
third party forces the trustee to incur a commitment by hostage post-
ing but address the deeper question concerning the conditions such
that the trustee posts a hostage voluntarily and without external
coercion. Notice that institutional embeddedness includes but is
not restricted to the legal infrastructure of social relations and the
enforcement of contractual agreements on hostages through the
law. Our examples include hostages that are posted through a con-
tractual agreement like the car dealer’s guarantee, the deposit in
the car-rental example, and the contract stipulating that the
employee compensates the employer for some of the training costs
in the case of a premature quit. In other cases, however, hostages
are posted informally and the enforcement of the commitments
incurred through hostage posting is not provided by law. The
employee who moves close to his job posts a hostage in a non-
contractual way.
We introduce a game-theoretic model of hostage posting as a
mechanism of trust, specifying the conditions such that hostage
posting promotes trust by binding the trustee, compensating the
trustor, or signaling unobservable properties of the trustee. We pro-
vide an integrated model allowing for a simultaneous analysis of the
binding, compensation, and signaling function of hostages. We
derive necessary and sufficient conditions for equilibria such that a
hostage is posted and trust is placed. We also tackle the equilibrium
selection problem, showing that there are only few – if any – other
equilibria and that these have less appealing properties. Previous
research analyzed different functions of hostages in separate
models. Raub and Keren (1993) as well as Weesie and Raub (1996)
provide a model of trust by hostage posting based on binding the
trustee. They neglect the signaling function of hostages. Also,
while the model used in their analyses addresses compensation of
the trustor through hostage posting, the basic assumption ensuring

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


326 RATIONALITY AND SOCIETY 16(3)

that compensation becomes relevant for promoting trust is the


assumption that the trustor’s behavior is not only equilibrium beha-
vior but also maximin-behavior. The model developed here allows
for studying the compensation function of hostages in a more parsi-
monious way, based on equilibrium behavior only. Rational signal-
ing through hostage posting is excluded in the model studied by
Snijders (1996; see also Snijders and Buskens 2001). Snijders (2000)
offers a model that includes binding and signaling as well as another
model that includes binding and compensation, but not a simple
integrated model including binding, compensation, and signaling.
Finally, Raub and Weesie (2000) offer a model that includes binding
and signaling but neglects compensation. Also, they only offer suffi-
cient conditions for the existence of equilibria such that hostages are
posted and induce trust but do not address equilibrium selection
problems.
The next section provides a formal model of trust problems that
is more complex than standard models as it includes uncertainty
as well as incomplete information of the trustor. Subsequently,
we extend the model by including an option for the trustee to post
a hostage. Thus, we consider a model of hostage posting for a trust
game with uncertainty and incomplete information. The following
section provides the analysis of the model, offering theorems that
specify conditions for rational hostage posting and trust based on
hostage posting. In the concluding section, we summarize the
results, outline testable predictions, and sketch a research agenda.

2. A Trust Game with Uncertainty and Incomplete Information

We consider trust problems in the sense that the trustee has incen-
tives to abuse trust and the trustor has something to lose if trust
is abused (Coleman 1990: chap. 5). A well-known model of trust
problems is the standard trust game as introduced by Dasgupta
(1988: 59–61) and Kreps (1990: 100–1). The game is played by two
players. Player 1 is the trustor and player 2 is the trustee. The trustor
moves first – Coleman stresses the importance of this feature: there is
a time lag between the action of the trustor and the action of the
trustee – and chooses between placing trust and withholding trust.
We denote the placement of trust by C1 (with C indicating ‘coopera-
tion’) and withholding trust by D1 (with D indicating ‘defection’).
The game ends if trust is not placed, with payoffs Pi (i ¼ 1; 2) for

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 327

trustor and trustee. If trust is placed, the trustee chooses between


honoring (C2 ) and abusing (D2 ) trust. If trust is honored, trustor
and trustee receive Ri > Pi . If trust is abused, the trustor receives
S1 < P1 , while the trustee receives T2 > R2 . The trustee thus has
an incentive to abuse trust. Note that these assumptions capture
Coleman’s argument that placing trust involves a risk. The trustor
is better off if trust is placed and honored than if she withholds
trust. On the other hand, if the trustee abuses trust, the trustor is
worse off compared to the situation when trust is not placed.
While the standard trust game represents basic features of trust
problems, we wish to introduce more complexity with respect to
information assumptions. The standard trust game employs rather
strong assumptions regarding the players’ information. In particu-
lar, there is no uncertainty for the players in the trust game and
they are assumed to be completely informed.
In the first step, uncertainty is introduced. In the standard trust
game, the trustor always receives the favorable payoff R1 if trust is
placed and honored: if the trustee honors trust and does not
behave opportunistically, it is certain that the gain that could be
associated with placing trust is realized. However, even if the trustee
honors trust, ‘things may go wrong’ due to unfavorable external
contingencies. For example, the dealer in used cars intends to sell
a peach but the car turns out to be in fact a lemon. The driver
who rents a car drives carefully but, unfortunately, becomes the
victim of a crash with another driver who is reckless. The employee
in whose training and schooling the employer invests does not quit
opportunistically but falls ill.
In all these scenarios, the trustor suffers even though the trustee
honored trust. While it is far from sure that ‘something goes
wrong’, the possibility that things go wrong should not be excluded.
It will turn out that such a possibility has implications for trust based
on hostage posting. We model the possibility that things can go
wrong by assuming a chance event (a ‘move of Nature’) after trust
has been placed and honored. The trust game with uncertainty dif-
fers from the standard trust game in that the trustor receives the
low payoff S1 after trust has been placed and honored with prob-
ability  and receives the high payoff R1 after trust has been
placed and honored with probability 1  . The expected payoff
of the trustor after placing trust that is honored by the trustee is thus

R1 ¼ S1 þ ð1  ÞR1 : ð1Þ

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


328 RATIONALITY AND SOCIETY 16(3)

In the trust game with uncertainty,  models the probability that


unfavorable contingencies obtain. If  > 0, the trustor faces uncer-
tainty about her payoff after trust has been placed and honored.
Obviously,  ¼ 0 is the special case of the standard trust game.
We assume 0   < 1, excluding the less interesting special case in
which the trustor never benefits from placing trust. For simplicity,
we retain the assumption that the trustee always receives R2 after
trust has been placed and honored (see Snijders 2000, Güth and
Ockenfels 2002, or Barrera and Buskens 2003 for alternative ways
of including uncertainty in a trust game). We thus assume, for
example, that the profit associated with unintentionally selling a
lemon is the same for the dealer as his profit from selling a peach.
Uncertainty for the trustee’s payoff R2 after trust has been placed
and honored (an assumption used in Snijders 2000 and Barrera
and Buskens 2003) could be easily included in the model. The impor-
tant feature is not that R2 ¼ R2 but only that T2 > R2 > P2 so that
the trustee has an incentive to abuse trust, while he also has an incen-
tive to induce the trustor not to withhold trust.
We wish to relax still an additional information assumption for
the trust game. In the standard trust game, the trustor is completely
informed of the possible actions as well as the incentives of the
trustee and knows for sure that she is confronted with a trustee
who has opportunities and incentives to abuse trust. A more realistic
– but likewise more complex – assumption seems to be that the
trustee has opportunities and incentives to abuse trust not with
certainty, but only some positive probability. Also, the trustor has
incomplete information about the trustee: she only knows the prob-
ability that the trustee has opportunities and incentives to abuse
trust but does not know the actual behavioral alternatives and incen-
tives of the trustee.
A setting with incomplete information seems to be a more realistic
model than a game with complete information for our examples of
trust problems. It seems reasonable to assume that there are some
car dealers who either do not have lemons or do not sell them
even if they have them. For some drivers it may be out of the ques-
tion to drive recklessly. An employee may or may not have an
opportunity and an incentive for an opportunistic quit.
We model the assumptions that the trustee may or may not have
opportunities and incentives to abuse trust and that the trustor does
not know the opportunities and incentives of the trustee for sure in
a game with incomplete information (see Rasmusen 1994 for an

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 329

accessible introduction to games with incomplete information). In


the Trust Game with Uncertainty and Incomplete Information
(TGUI) the trustor can meet two possible ‘types’ of trustees. The
first type is ‘unreliable’ in the sense that he has opportunities and
incentives to abuse trust. If the trustor places trust, such a trustee
can either honor (C2) or abuse trust (D2) and abusing trust yields
a higher payoff (T2) than honoring trust (R2). The trustor again
receives P1 after abused trust and – due to the uncertainty involved
in the game – expected payoff R1 ¼ S1 þ ð1  ÞR1 after honored
trust. A second type of trustee is ‘reliable’ in the sense that he has no
opportunities for abusing trust. If the trustor meets such a trustee,
the trustee ‘automatically’ honors trust and the game ends with
payoff R2 for the trustee and expected payoff R1 for the trustor
after trust has been placed. A chance move of Nature at the begin-
ning of TGUI determines the type of trustee playing the game.
With probability , Nature ‘chooses’ an unreliable trustee and
with probability 1   a reliable one. The probability  of meeting
an unreliable trustee is known to both players. The outcome of
Nature’s initial move and hence the trustee’s actual type is known
to the trustee himself but is unobservable for the trustor. We thus
assume that the potential buyer knows the proportion of dealers
who do not sell lemons but cannot tell the type of a specific
dealer. Similarly, the rent-a-car firm is assumed to know the pro-
portion of careful drivers while being unable to identify a specific
driver’s type. The employer is assumed to be able to derive the prob-
ability for the lack of opportunities and incentives for a premature
quit of the employee from characteristics of the labor market as
well as characteristics of the employee himself. Note that TGUI is
not only a game with incomplete information (the outcome of a
move of Nature at the beginning of the game is not observed by
all players) but also a game with asymmetric information (the infor-
mation of the trustor about the initial move of Nature differs from
the information of the trustee). If  ¼ 1, the trust game with uncer-
tainty results from TGUI; the trustee is always unreliable and the
trustor is completely informed of the trustee’s type. For  ¼ 0, we
get a trust game with incomplete information but without uncer-
tainty. For  ¼ 1 as well as  ¼ 0, we get the standard trust game.
Figure 1 provides the extensive form of TGUI. The reliable trustee
who has no opportunities for abusing trust does not move at all in
the tree in Figure 1. This is equivalent to using a tree such that the

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


330 RATIONALITY AND SOCIETY 16(3)

Figure 1. Extensive form of the Trust Game with Uncertainty and Incomplete
Information (TGUI) with payoffs for player 1 ðR1 > P1 > S1 Þ and player 2
ðT2 > R2 > P2 Þ; 0   < 1; maxðð S1 þ ð1  ÞR1  P1 Þ=
ð S1 þ ð1  ÞR1  S1 Þ; 0Þ <   1; dotted line indicates information set.

only move of the reliable trustee after placement of trust is to honor


trust (C2).
We use standard game-theoretic assumptions for analyzing trust
and, subsequently, trust based on hostage posting. Therefore, pay-
offs represent utilities. Thus, an actor’s distributional concerns such
as altruism, egalitarianism, envy, etc. or other ‘utility arguments’
which imply that his or her ‘objective outcomes’ (for example, the
actor’s money) differ from ‘effective outcomes’ (the actor’s utility)
are included in the payoffs (see Snijders 1996 for an analysis of the
trust game that explicitly distinguishes between objective and effec-
tive outcomes, following the suggestion by Kelley and Thibaut
1978). We further assume that the structure of the game (the exten-
sive form of the game as summarized in the game tree, including the

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 331

payoffs for each player at each end node) is common knowledge


(Rasmusen 1994: 44). Also, we assume throughout that the game
is played noncooperatively in the standard sense that players are
unable to make enforceable agreements or enforceable one-sided
commitments, except agreements and commitments explicitly
modeled as moves in the extensive form. Assuming a noncooperative
game does not mean that we assume or imply that rational players
will never cooperate in the sense of placing and honoring trust. On
the contrary, the assumption of a noncooperative game is employed
precisely because we wish to derive conditions such that rational
players incur agreements and commitments that promote trust.
Thus, we employ the Nash program (Nash 1951) of explicitly
modeling pre-play behavior such as bargaining, contracting, and
commitments on how to play the game as moves in the extensive
form of an extended noncooperative game and to derive cooperative
behavior as equilibrium behavior of the extended game.
Consider equilibrium behavior in TGUI. In equilibrium, the un-
reliable trustee will always abuse trust because T2 > R2 . The trustor
chooses to withhold trust in equilibrium if the payoff P1 she receives
after withholding trust, irrespective of the trustee’s type, is higher
than her expected payoff after placing trust. This is the case if
and only if P1 > S1 þ ð1  Þ½ S1 þ ð1  ÞR1 , which can be
rewritten as
 > ðR1  P1 Þ=ðR1  S1 Þ: ð2Þ
Thus, TGUI has a unique equilibrium such that trust is not placed if
and only if the probability  of meeting an unreliable trustee is suffi-
ciently high. The critical value for the probability of an unreliable
trustee is specified in (2). If  < ðR1  P1 Þ=ðR1  S1 Þ, the unique
equilibrium of TGUI implies that trust is placed and abused by
the unreliable trustee. We assume that  fulfills (2) because we
wish to avoid a trivial ‘solution’ of trust problems based on the
assumption that the probability of meeting a trustee who has oppor-
tunities and incentives to abuse trust is simply too small. It is easily
verified that ðR1  P1 Þ=ðR1  S1 Þ < 1 so that (2) is indeed fulfilled
for sufficiently large . Obviously, in the special cases of complete
information of the trustor on the trustee’s type ð ¼ 1) or in the
trust game without uncertainty ð ¼ 0Þ, there is likewise a unique
equilibrium such that trust is not placed but would be abused if it
were placed.

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


332 RATIONALITY AND SOCIETY 16(3)

Since P2 < R2 , the trustee prefers honored trust to withheld trust


in TGUI. Since P1 < R1 , the trustor is better off if trust is placed
and honored and favorable contingencies obtain than if trust is with-
held. However, since uncertainty is involved, Pareto-(sub)optimality
of the ‘no trust’-equilibrium of TGUI depends on the probability
of unfavorable contingencies. The trustor prefers honored trust
to withholding trust and the ‘no trust’-equilibrium is Pareto-
suboptimal if P1 < R1 ¼ S1 þ ð1  ÞR1 . This is the case if and
only if
 < ðR1  P1 Þ=ðR1  S1 Þ: ð3Þ
We see that TGUI is a social dilemma in the sense that individually
rational equilibrium behavior implies a ‘collectively irrational’
(Rapoport 1974) outcome if the probability  of unfavorable con-
tingencies is small enough. It follows immediately that the equili-
brium of the game is Pareto-suboptimal for the special case of no
uncertainty ð ¼ 0Þ, and for the standard trust game without un-
certainty and with complete information ð  ¼ 0 and  ¼ 1Þ. We
will see that the conditions for solving trust problems through
hostage posting depend in important ways on whether or not (3)
holds. Note also that ðR1  P1 Þ=ðR1  S1 Þ > 0 if and only if (3) is
fulfilled. Therefore, the trustor withholds trust irrespective of the
probability  of an encounter with an unreliable trustee if the uncer-
tainty associated with placement of trust is too large. Subsequently,
we assume maxððR1  P1 Þ=ðR1  S1 Þ; 0Þ <   1, excluding the less
interesting possibility that the trustee is surely reliable as well as the
case that the probability of meeting an unreliable trustee is ‘too
small’.
Several features of TGUI merit discussion. First, the labels ‘un-
reliable’ and ‘reliable’ trustee are based on a rather specific interpre-
tation of ‘(un)reliability’. The meaning of unreliability is ambiguous
in the context of a trust game. On the one hand, ‘unreliability’ may
mean that the trustee has opportunities and incentives for abusing
trust because he can choose between C2 and D2 and because
T2 > R2 . ‘Unreliability’ in this sense refers to the structure of the
interaction situation for the trustee. On the other hand, ‘unreli-
ability’ may mean that the trustee would actually abuse placement
of trust by the trustor by choosing D2. In this sense, ‘unreliability’
refers to behavior. Here, we use ‘unreliability’ as a characteristic
of the structure of the interaction situation. When we consider
trust based on hostage posting, we wish to show that an agent

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 333

who is unreliable in the structural sense of having opportunities and


incentives to abuse trust may well honor trust and thus be reliable in
the behavioral sense; under suitable conditions, posting a hostage
is a means for an otherwise unreliable agent to commit himself to
honoring trust.
Second, it is instructive to relate condition (2) to Coleman’s dis-
cussion of trust problems. Coleman (1990: 99) provides a condition
for placing and, conversely, withholding trust that became influen-
tial in the sociological literature. This condition says that trust is
not placed if the probability that the trustee is trustworthy divided
by the probability that he is not trustworthy is less than the loss
from abused trust divided by the gains from honored trust. Using
our notation, Coleman’s condition is thus
ð1  Þ= < ðP1  S1 Þ=ðR1  P1 Þ:
Coleman neglects a strategic (i.e. game-theoretic) analysis of trust
problems. He analyzes the trustor’s decision as a game against
Nature rather than against an incentive-guided partner. Therefore,
Coleman’s theoretical justification of his condition is somewhat
dubious: it seems strange to assume rational and strategic behavior
of the trustor but to neglect or ‘assume away’ rational and strategic
behavior of the trustee (see Hardin 2002: 6, for a similar argument).
Note, however, that Coleman’s condition is equivalent to condition
(2). Hence, our analysis provides a ‘rational reconstruction’ of
Coleman’s condition for the placement of trust.
Third, we use a very specific and simple assumption on incomplete
information of the trustor in TGUI. Alternative assumptions are
conceivable. For example, one could imagine that a reliable trustee
has opportunities to abuse trust but has no incentives to do so. This
would be the case if the trustee has internalized norms and values
inducing sufficient ‘internal sanctions’ should he abuse trust and
behave opportunistically. Then, honoring trust would be associated
with a higher ‘net utility’ than abusing trust (see Camerer and
Weigelt 1988; Dasgupta 1988; Bacharach and Gambetta 2001; and
Güth and Ockenfels 2002 for such models of trust games with
incomplete information). This alternative scenario would lead to
similar results. We prefer our conceptualization because it seems
to fit better with a rational choice approach by focusing on the
possible lack of opportunities to abuse trust rather than the possi-
bility that the trustee is ‘(over)socialized’: we keep our assumptions
about individuals and their preferences as simple as possible, while

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


334 RATIONALITY AND SOCIETY 16(3)

complexity is introduced into the model via assumptions on their


interaction situation, such as assumptions on feasible actions and
restrictions (see Wippler and Lindenberg 1987; Coleman 1990). Still
another variant of a trust game with incomplete information is one
in which the trustor is confronted with a trustee who is ‘malicious’
in the sense that his payoff function is such that he tries to minimize
the trustor’s payoff (see Raub and Keren 1993 and Snijders 2000).
This approach would also yield results similar to those presented
below. Finally, a substantively different approach to modeling
trust problems with incomplete information is to assume that the
trustor lacks information on the probability  of unfavorable con-
tingencies rather than the trustee’s opportunities or incentives for
abusing trust (see Barrera and Buskens 2003 for such a model).
Fourth, note that 1   can be interpreted as the probability that
the trustee is competent in the sense that if trust is placed and the
trustee honors trust by trying to realize a good outcome for the trus-
tor, the good outcome is indeed realized: 1   is the probability
that the dealer in used cars is able to distinguish between lemons
and peaches. It is often argued in the literature on trust (e.g.
Barber 1983) that placing trust can be problematic due to incentives
for the trustee to abuse trust as well as due to uncertainty of the
trustor about the competencies and abilities of the trustee. Trust
in competencies and abilities is sometimes labeled ‘confidence’. My
model thus covers problematic trust due to incentives as well as
due to uncertainty about competencies – placing trust depends on
trust as well as the confidence of the trustor.
Fifth, one could obviously increase the complexity of this model
by assuming a chance event not only after trust has been placed
and honored but also after trust has been placed and abused. In
such a scenario, ‘things may go right’ for the trustor even if the
trustee abuses trust due to favorable external contingencies. The
intuition is that the possibility of such favorable contingencies
would facilitate trust. We abstain from an explicit analysis of this
case.

3. Modeling Hostage Posting as a Mechanism of Trust

We now introduce a model of hostage posting that allows derivation


of conditions such that rational trustors and trustees are able to
solve their trust problem. We define a Hostage Game HTGUI

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 335

that is an extended version of the trust game with uncertainty and


incomplete information.
Like in TGUI, Nature moves first in HTGUI and determines the
trustee’s type. With probability , the trustee is unreliable and has
incentives and opportunities for abusing placement of trust; with
probability 1, the trustee has no opportunities for abusing trust
(he is reliable). Nature’s move is observed by the trustee but the trus-
tor cannot observe the trustee’s type and is only informed of the
probabilities for either type of trustee. Subsequently, the trustee
moves and chooses between posting or not posting a hostage. We
denote hostage posting by H þ , while H  denotes the decision not
to post a hostage. The trustor is informed on the trustee’s hostage
posting decision. Afterwards, the trustor decides to place (C1 ) or
to withhold trust (D1 ). The game ends after D1 . If the trustee
happens to be of the reliable type who cannot abuse trust, the
game also ends after C1 (more precisely, the game ends after the reli-
able trustee’s ‘trivial’ move C2 and the subsequent chance move of
Nature that determines whether or not ‘things go wrong’ although
the trustee honors trust). If the trustor places trust and the trustee
is unreliable, the trustee chooses between honoring (C2 ) and abusing
trust (D2 ) and the game ends (again, if trust is honored, after the
chance move of Nature). Note that when the trustor moves, she
can condition her placement of trust on hostage posting.
We assume that the trustee loses his hostage if and only if he posts
it, the trustor places trust, and the trustee either abuses trust or
honors trust, but unfavorable contingencies obtain and ‘things
go wrong’ even though the trustee abstains from opportunistic
behavior. We therefore consider hostage posting as a mechanism
to mitigate risks from opportunistic behavior of the trustee as well
as risks from uncertainty. For example, the guarantee provided by
the dealer covers repairs of lemons he sold intentionally as well as
unintentionally. The hostage has value K2 > 0 for the trustee. For
simplicity, we assume that the value of the hostage for the reliable
trustee is the same as the value for the unreliable trustee. Hostage
posting is associated with transaction costs for the trustee. We con-
sider transaction costs in Williamson’s (1985: 20–2, 388) sense, that
is, the costs of drafting, negotiating, setting up, and running the
hostage arrangement. Examples are opportunity costs due to the
temporary impossibility of using the hostage, enforcement costs
from hiring a lawyer who monitors the hostage arrangement,
the risk that a hostage posted under the control of the partner is

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


336 RATIONALITY AND SOCIETY 16(3)

unexpectedly not returned later on, and costs that result from
reduced flexibility to respond to changes in exogenous conditions
(see also the concluding discussion). These costs arise if and only
if the trustee decides to post the hostage and do not depend on
how the game develops after hostage posting. Hence, these costs
are not only due if the trustee loses his hostage (in this case, one
could consider the transaction costs simply as an ingredient of the
value K2 of the hostage for the trustee) but also if trust is placed
and honored after hostage posting as well as if no trust is placed
after hostage posting. As an example, consider the car dealer’s trans-
action costs associated with providing a guarantee. Here, a core
element of his costs are the dealer’s (increased) costs of running a
garage so that repairs of cars under guarantee can be carried out.
Alternative assumptions on transaction costs are conceivable. For
example, Snijders (2000) employs the assumption that costs of
hostage posting arise if and only if the hostage is ‘accepted’ by the
trustor in the sense that she places trust (thus, no transaction costs
arise for the trustee if a hostage is posted but no trust is placed).
Our assumption on the transaction costs of hostage posting some-
what simplifies the analysis of the model, while using Snijders’s
assumption would not yield substantively different results.
A core assumption of our model is that the transaction costs of
hostage posting can differ between the two types of trustees. On
the one hand, allowing for differences in transaction costs seems
realistic when considering the examples. The costs of running a
garage that are related to the repair of cars under guarantee
depend on whether or not the dealer in used cars has lemons in
house. In the example of the employee who posts a hostage by
moving close to his job, an important ingredient of the transaction
costs involved in posting the hostage is due to undermining the
employee’s bargaining position vis-à-vis an alternative employer
offering a new position. These transaction costs emerge by definition
for an employee who is unreliable in the sense of our model, while a
reliable trustee does not have to incur these costs. While the assump-
tion that transaction costs of hostage posting can differ between dif-
ferent types of trustees seems to be empirically plausible, differences
in transaction costs are also interesting from our theoretical per-
spective of trust based on hostage posting. After all, one expects
from signaling theory (Spence 1974) that the signaling function of
hostages depends on differences in signaling costs for different
types of trustees. Transaction costs associated with hostage posting

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 337

can be conceived as signaling costs of hostage posting and, therefore,


the question arises if differences in transaction costs for different
types of trustees can affect the signaling function of hostages. We
denote the transaction costs of hostage posting for an unreliable
trustee with u , those for a reliable trustee with r , and assume
that transaction costs are positive for both types (u ; r > 0).
What happens if the trustee posts a hostage that he loses because
he abuses trust after hostage posting? We wish to analyze the
compensation function of hostages. Therefore we assume that a
lost hostage is given to the trustor for whom the hostage has value
K1. The case K1 ¼ 0 can be interpreted as the situation in which a
lost hostage is not given to the trustor (see Weesie and Raub 1996:
214–16 for a detailed discussion of ‘hostage institutions’, that is,
exogenous rules of the hostage game that determine under what con-
ditions players lose a hostage and what happens with a lost hostage).
Note that our examples differ with respect to the value of the
hostage for the trustor. The guarantee in the example of buying a
used car is a relatively valuable hostage for the trustor. The value
of the hostage ‘moving close to one’s job’, while being high for the
employee as trustee who posts the hostage, is low for the employer
as trustor. We neglect an explicit analysis of the effects of the pos-
sible costs of hostage posting for the trustor. In the example of
buying a used car, the buyer’s costs associated with bringing a car
with a defect to the dealer’s garage and her costs associated with
not being able to use a car that is under repair can be assumed to
be an ingredient of K1. Accounting for the effects of an increase in
the price of the car if the dealer offers a guarantee would require
the assumption that the trustor’s payoff R1 also depends on whether
or not a hostage has been posted. Implications of such an assump-
tion will be addressed in the concluding discussion.
We can now define the payoff function for HTGUI. We assume
that each player’s payoff in HTGUI is additive in the player’s
payoff at the end of the corresponding trust game with uncertainty
and incomplete information TGUI, in the value of a hostage lost
or received, and in the transaction costs of hostage posting. For
example, the payoff of an unreliable trustee is T2  K2  u if he
posts a hostage and subsequently abuses trust that has been placed
by the trustor, while the payoff of the trustor would be S1 þ K1 .
The expected payoff of an unreliable trustee after hostage posting
and honored trust would be ðR2  K2  u Þ þ ð1  ÞðR2  u Þ,
with ðS1 þ K1 Þ þ ð1  ÞR1 as the corresponding payoff of the

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


338 RATIONALITY AND SOCIETY 16(3)

trustor, and ðR2  K2  r Þ þ ð1  ÞðR2  r ) as the expected


payoff of the reliable trustee after hostage posting and trust placed
by the trustor. HTGUI is thus defined by (R1 , P1 , S1 , T2 , R2 , P2 ,
K1 , K2 , r , u , , ). The extensive form of the game is as shown in
Figure 2.

4. Analysis of the Model: Theorems on Trust Based on Hostage


Posting

We use a standard characterization of individually rational behavior


in games with incomplete information and consider perfect Bayesian
equilibria of HTGUI (see Rasmusen 1994: 52, 146–52, for a discus-
sion of the concept). Roughly we thus consider strategy combi-
nations such that each player’s strategy maximizes the player’s
expected payoff given the other players’ strategies and such that
the trustor uses her observation of the trustee’s hostage posting deci-
sion to update her beliefs about his type in a rational way, namely
according to Bayes’s rule (and where, technically speaking, out-of-
equilibrium behavior leaves the beliefs of the trustor unchanged).
We present generic properties of HTGUI. Consider two outcomes
of the game such that in one outcome the trustee has posted a
hostage while in the other outcome he did not post a hostage.
Then, generically, the trustee is never indifferent between these two
outcomes. Thus, for example, the trustee is not indifferent between
the outcome such that no hostage has been posted and trust is
placed and honored on the one hand and the outcome such that a
hostage has been posted and trust is placed and abused. Likewise,
consider two outcomes of the game where in one outcome the trus-
tor receives a hostage with value K1 6¼ 0 and in the other outcome
she does not receive a hostage. Then, generically, the trustor is
never indifferent between these two outcomes. Finally, we assume
that the trustor is never indifferent between withholding trust and
the expected payoff if no hostage has been posted and trust is
placed and honored. Without losing substantive insights, we thus
neglect very special instances of HTGUI such that a player’s payoffs
associated with two ‘substantively different’ outcomes are exactly
the same (see Weesie and Raub 1996: 216, 233, for further discus-
sion). We focus on equilibria where the associated equilibrium path
(the behavior of the players induced by the equilibrium) is interesting
from a theoretical and substantive perspective and derive necessary

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 339

Figure 2. Extensive form of the Hostage Trust Game with Uncertainty and
Incomplete Information (HTGUI) with payoffs and hostage value for player 1
ðR1 > P1 > S1 ; K1 Þ and player 2 ðT2 > R2 > P2 ; K2 Þ, and transaction costs
r and u for reliable and unreliable trustee (player 2);
0   < 1; maxðð S1 þ ð1  ÞR1  P1 Þ=ð S1 þ ð1  ÞR1  S1 Þ; 0Þ <   1; dotted
lines indicate information sets.

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


340 RATIONALITY AND SOCIETY 16(3)

and sufficient conditions for each of the equilibria. Therefore, we


derive theorems on equilibria in HTGUI that allow for conclusions
on binding, compensation, and signaling by hostages. We keep the
equilibrium selection problem in mind and thus also derive all
other equilibria, if any, that coexist or may coexist with some theo-
retically or substantively interesting equilibrium. Explicitly specify-
ing the equilibria for all (R1, P1, S1, T2, R2, P2, K1, K2, r ; u ; ; )
is feasible but tedious and does not yield much additional insight.
It is useful to first establish some properties of the game that pro-
vide intuition for the subsequent theorems and facilitate the proofs.
These properties are directly related to the binding and compensa-
tion function of hostages, and to possible gains from hostage posting
for the trustee. The properties refer to preferences of trustee and
trustor over endnodes of the games and to conditions for best reply
behavior (i.e. payoff maximizing behavior) of the players. We sum-
marize these properties as lemmas on HTGUI. Proofs of the
lemmas follow immediately from the extensive form of HTGUI
and our assumptions on the (ingredients of the) players’ payoffs.
Consider first the condition such that the hostage is binding for
a rational trustee. This is the case if the trustee honors trust after
hostage posting and placed trust.

Lemma 1: Generically, best reply behavior implies that the unreliable


trustee honors trust after hostage posting and placed trust if and only if
K2 ð1  Þ > T2  R2 : ð4Þ
The condition provides a lower bound on the value of the hostage
for the trustee. Note that T2  R2 are the trustee’s costs of honoring
trust in the trust game. Condition (4) thus says that the hostage is
binding if the value of the hostage for the trustee, weighted with
the probability that favorable contingencies obtain after honored
trust and that the hostage is not lost although trust has been
honored, exceeds the costs of honoring trust in the trust game. Con-
dition (4) can be seen as the incentive compatibility constraint for
the trustee.
Next, we specify conditions such that the trustee can benefit from
hostage posting. This can happen because hostage posting induces
placement of trust by the trustor that is subsequently honored,
while no hostage posting induces the trustor to withhold trust.

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 341

Lemma 2: Trustee j ð j ¼ u; rÞ is better off after hostage posting,


placed trust, and honored trust than after no hostage posting and with-
held trust if and only if
K2 þ j < R2  P2 : ð5Þ

This condition provides upper bounds on the value of the hostage


for the trustee and on his transaction costs. The critical value in
(5) is the gain R2  P2 from honored trust compared to the situation
of withheld trust. The upper bound on K2 is a result of the trustee’s
risk of losing his hostage under unfavorable contingencies even
though he does not abuse trust. Condition (5) can be seen as the
participation constraint for the trustee.
We will see in the theorems that it is also important to consider the
case in which the unreliable trustee is worse off if he abuses trust
after hostage posting than when he does not post a hostage and
the trustor withholds trust.

Lemma 3: The unreliable trustee is worse off after hostage posting,


placed trust, and abused trust than after no hostage posting and with-
held trust if and only if
K2 þ u > T2  P2 : ð6Þ

Finally, we turn to compensation of the trustor through hostage


posting by the trustee. Our next lemma provides the compensation
condition that will be crucial for our subsequent theorems.

Lemma 4: Assume that best reply behavior implies that trust will not
be abused after hostage posting by the trustee and placement of trust
by the trustor. Generically, best reply behavior implies that the trustor
places trust after hostage posting if and only if

K1 > R1  ðR1  P1 Þ=  S1 : ð7Þ

This condition provides a lower bound on the value of the hostage


for the trustor. Note that R1  ðR1  P1 Þ=  S1 < 0 if and only if
(3) is fulfilled. Thus, if the probability  of unfavorable contin-
gencies is small and TGUI is a dilemma, the lower bound on the
value of the hostage for the trustor in (7) is weak in the sense that
the condition only requires that the value does not represent too
much disutility for the trustor. Conversely, K1 > 0 and positive com-
pensation of the trustor is required to induce placement of trust after

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


342 RATIONALITY AND SOCIETY 16(3)

hostage posting if the probability of unfavorable contingencies is


too large. The intuition for this result is as follows. If the probability
of unfavorable contingencies is small enough (in the sense that (3) is
fulfilled), the trustor’s risk that has to be mitigated by hostage post-
ing is the risk that the trustee will abuse trust. This risk can be miti-
gated if the hostage is binding for the trustee or if hostage posting
signals that the trustee’s type is reliable. Positive compensation of
the trustor is not required since the trustor’s expected payoff R1
from honored trust exceeds her payoff P1 from withholding trust.
On the other hand, if the probability of unfavorable contingencies
is large (in the sense that (3) is not fulfilled), hostage posting induces
placement of trust by the trustor only if the hostage not only miti-
gates the risk of opportunistic behavior of the trustee but also the
risk of unfavorable contingencies.
We now derive conditions such that placing and honoring trust
based on hostage posting becomes individually rational. First, we
consider a pooling equilibrium. In this equilibrium, the hostage
posting decision of both types of trustees is the same so that hostage
posting does not signal the type of the trustee.

Theorem 1 Pooling equilibrium with trust based on hostage posting.


Generically, HTGUI has a pooling equilibrium such that
(i) the reliable and the unreliable trustee post a hostage,
(ii) the trustor places trust after hostage posting and withholds
trust otherwise,
(iii) the unreliable trustee honors trust after hostage posting and
trust placed but abuses trust if no hostage has been posted
and trust is placed
if and only if
K2 ð1  Þ > T2  R2 and K2 þ j < R2  P2
for j ¼ r; u and K1 > R1  ðR1  P1 Þ=  S1 : ð8Þ
The pooling equilibrium is unique if the probability  of unfavorable
contingencies is high so that condition (3) is not fulfilled (TGUI is
no dilemma) or if u  r . If and only if the probability  of unfavor-
able contingencies is low so that condition (3) is fulfilled (TGUI is a
dilemma) and r is sufficiently higher than u , there exist other equili-
bria such that the unreliable trustee randomizes with respect to his
hostage posting decision, the reliable trustee posts no hostage or posts

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 343

no hostage at least with positive probability, the trustor places trust


after hostage posting and randomizes between placing and withholding
trust if no hostage has been posted, and the unreliable trustee honors
trust after hostage posting and trust placed but abuses trust if no
hostage has been posted and trust is placed. These equilibria, if they
exist, are weakly Pareto-dominated by the pooling equilibrium.

Proof Refer to Appendix B for proofs of the theorems.

The theorem shows that hostage posting can indeed be a mechan-


ism for solving trust problems. The equilibrium strategies are
‘reactive’ in the sense that trustor and trustee condition their own
behavior on the behavior of the partner. The strategy of the trustor
makes her placement of trust dependent on prior hostage posting by
the trustee. Her strategy carries the tacit promise to place trust after
hostage posting and the tacit threat to withhold trust if the trustee
does not post a hostage. Conversely, the trustee’s strategy implies
a tacit promise not to abuse the trustor’s trust. The equilibrium
ensures that these threats and promises are credible. Note that the
pooling equilibrium shows that a rational unreliable trustee may
post a hostage and honor trust.
According to the theorem, the existence of a pooling equilibrium
is equivalent to the hostage being binding for the trustee (see the first
condition in (8) and Lemma 1), sufficiently small transaction costs of
hostage posting for both unreliable and reliable trustee and a not-
too-valuable hostage for the trustee (see the second condition in
(8) and Lemma 2), and sufficient compensation for the trustor (see
the third condition in (8) and Lemma 4). The theorem thus implies
that the binding and compensating functions of hostages depend
on the hostage being valuable enough for trustor and trustee as
well as being not too valuable for the trustee. The compensation
condition for the trustor requires K1 > 0 if and only if the probabil-
ity  of unfavorable contingencies is high (so that (3) is not fulfilled).
This is so because on the equilibrium path, compensation is never
necessary due to abused trust but only due to honored trust and
unfavorable contingencies.
Efficiency properties of the pooling equilibrium are easily verified.
The pooling equilibrium is a Pareto-improvement compared to the
‘no trust’ equilibrium in the underlying trust game: the players are
better off if a hostage is posted and trust is placed and honored
than if trust is withheld. The pooling equilibrium is not necessarily

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


344 RATIONALITY AND SOCIETY 16(3)

second best compared to trust that is placed and honored without


hostage posting. Of course, placing and honoring trust without
hostage posting is more attractive than the pooling equilibrium for
the trustee because hostage posting comes with transaction costs.
However, ‘things can go wrong’ due to unfavorable contingencies
after honored trust in the underlying trust game because uncertainty
is involved in the game. In the pooling equilibrium, the trustor
receives compensation in this case. Thus, if K1 > 0, the trustor pre-
fers the pooling equilibrium to placing and honoring trust without
hostage posting.
Finally, the theorem shows that the pooling equilibrium is either
unique or is at least a weak Pareto-improvement compared to all
other – if any – equilibria (more precisely, the trustor is better off
in the pooling equilibrium, while both types of trustees are indiffer-
ent between the equilibria). Thus, if the pooling equilibrium exists,
equilibrium selection problems are not severe. The prediction that
rational actors will post hostages and will place and honor trust
follows from the uniqueness of the pooling equilibrium or from
payoff dominance arguments.
We now consider the separating equilibria of HTGUI. These are
equilibria in which the two types of trustees differ with respect to
their hostage posting decision so that hostage posting involves a
signal for the trustor about the type of the trustee (see Dasgupta
1988 and Bacharach and Gambetta 2001 for related discussions of
using signals to mitigate trust problems). In a separating equili-
brium, hostage posting affects the trustor’s beliefs about the charac-
teristics of the trustee. Thus, hostages that serve signaling purposes
contribute to the ‘definition of the situation’ and to ‘framing’. These
are classical topics of (micro-)sociology. Scholars have tried to
integrate these phenomena into a rational actor perspective (e.g.
Lindenberg 1992; Esser 1993) but these approaches do not provide
a rigorous model for the analysis of ‘relational signals’ (Lindenberg
1994: 106–8). It has even been suggested that rational signaling is
unfeasible in trust problems: given rational action in trust problems,
‘economic life would be poisoned by incessant attempts to conceal
the true incentive situation for one’s own advantage’ (Granovetter
2002: 40). It is true – and has been widely acknowledged in the litera-
ture (e.g. Frank 1988) – that rational signaling can be problematic
due to mimicry and similar phenomena. However, as we will see,
signaling can be feasible among rational actors who are involved
in a trust problem, the conditions for rational signaling can be

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 345

explicitly specified, and framing phenomena can be analyzed in a


parsimonious model using standard game-theoretic rationality
assumptions.
We consider the case that is most interesting from a substantive
perspective and with respect to our examples. This is the case
when a reliable trustee posts a hostage while an unreliable trustee
chooses not to post a hostage and the trustor places trust after
hostage posting but withholds trust if no hostage has been posted.
Here, posting a hostage signals that the trustee is reliable. In our
examples, this would be the case in which providing a guarantee
indicates that the dealer in used cars does not have lemons or
never has an incentive to sell them. The client’s willingness to pro-
vide a deposit would signal that he is not a reckless driver. The
employee’s willingness to move close to his job would indicate that
he does not have and does not expect to receive an outside offer
from another employer. The next theorem provides necessary and
sufficient conditions for the corresponding separating equilibrium.

Theorem 2 Separating equilibrium with hostage posting by the


reliable trustee and placement of trust. Generically, HTGUI has a
separating equilibrium such that
(i) the reliable trustee posts a hostage,
(ii) the unreliable trustee does not post a hostage,
(iii) the trustor places trust after hostage posting and withholds
trust after no hostage posting,
(iv) the unreliable trustee abuses trust if no hostage has been
posted and trust is placed
if and only if
K2 ð1  Þ > T2  R2 and K2 þ u > R2  P2 and

K2 þ r < R2  P2 and K1 > R1  ðR1  P1 Þ=  S1 ð9Þ


or
K2 ð1  Þ < T2  R2 and K2 þ u > R2  P2 and

K2 þ r < R2  P2 and K1 > R1  ðR1  P1 Þ=  S1 : ð10Þ


If the separating equilibrium exists, the only other equilibrium is such
that both the reliable and the unreliable trustee do not post a hostage,
the trustor withholds trust after hostage posting and if no hostage has

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


346 RATIONALITY AND SOCIETY 16(3)

been posted, and the unreliable trustee abuses trust if no hostage has
been posted and trust is placed. This equilibrium is weakly Pareto-
dominated by the separating equilibrium.

According to the theorem, rational behavior implies for suitable


parameters of the game that only the reliable trustee posts a hostage
so that hostage posting reveals the trustee’s type and hostage posting
induces the placement of trust. The trustor’s equilibrium strategy is
again conditional on prior hostage posting of the trustee so that the
trustor uses tacit threats and promises that are credible since they are
based on equilibrium behavior. The theorem shows that the separat-
ing equilibrium exists under two different sets of conditions. Under
(9), the hostage is binding for the trustee (see the first condition in (9)
and Lemma 1), but the transaction costs of hostage posting and the
value K2 of the hostage are so high for the unreliable trustee that
honoring trust after having posted a hostage is less attractive for
him than not posting a hostage and having the trustor withhold
trust (see the second condition in (9) and Lemma 2). Conversely,
the costs of hostage posting are sufficiently small and the hostage
is not too valuable for the reliable trustee (see the third condition
in (9) and Lemma 2). Therefore, posting a hostage can become a
signal of the trustee’s type. Finally, the hostage provides sufficient
compensation for the trustor (see the fourth condition in (9) and
Lemma 4) so that hostage posting induces the trustor to place
trust. Once again, trust is never abused on the equilibrium path
and therefore compensation is an issue in the equilibrium only if
unfavorable contingencies emerge after honored trust. Therefore,
K1 > 0 is required if and only if the probability of unfavorable
contingencies is large (so that (3) is not fulfilled). The second and
third conditions in (9) jointly imply that u > r : the transaction
costs of hostage posting are higher for the unreliable than for the
reliable trustee. Hence, we see that sufficient differences in the costs
of hostage posting between the two types of trustees can indeed
imply a signaling function of hostage posting.
From the perspective of hostages as signals, the conditions in (10)
are even more instructive. Due to the first condition in (10), the hos-
tage is not even binding (see Lemma 1) and an unreliable trustee will
abuse trust if trust has been placed after hostage posting. In this
situation, the trustor will place trust after hostage posting only if
she can conclude from hostage posting that the trustee is reliable.
This conclusion, however, can indeed be derived from the second

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 347

and the third condition in (10). According to Lemma 3 and the


second condition in (10), it is not attractive for the unreliable trustee
to imitate the reliable trustee by posting a hostage and subsequently
to abuse trust should the trustor decide to place trust (of course, if
such imitation were attractive for the unreliable trustee, the strategy
of the trustor would not be optimal). On the other hand, the third
condition in (10) implies that the costs of hostage posting and the
value of the hostage are sufficiently small for the reliable agent so
that his expected payoff after hostage posting and placement of
trust is higher than his payoff if he posts no hostage and the trustor
withholds trust (see Lemma 2). Therefore, posting a hostage is again
a signal of the trustee’s type. Since the fourth condition in (10)
ensures sufficient compensation for the trustor (see Lemma 4) and
thus induces placement of trust, signaling his type through hostage
posting becomes optimal for the reliable trustee. The first three
conditions in (10) jointly imply once again that u > r so that the
transaction costs of hostage posting are higher for the unreliable
than for the reliable trustee. Hence, the theorem shows that suffi-
cient differences in the costs of hostage posting between the two
types of trustees are indeed necessary for a signaling function of
hostage posting such that hostage posting shows that the trustee is
reliable and hostage posting induces placement of trust.
In summary, the theorem shows not only that the signaling func-
tion of hostages depends on differences in the transaction costs of
hostage posting but also that trust based on hostage posting pre-
supposes that transaction costs of hostage posting (in the present
case: the transaction costs of the reliable trustee) and the value of
the hostage for the trustee (in the present case, again the reliable
trustee) are not too high, and that the hostage is sufficiently valuable
for the trustor. From an efficiency perspective, the separating equi-
librium is a weak Pareto-improvement compared to the ‘no trust’-
equilibrium in the underlying trust game (the Pareto-improvement
is weak since only the trustor and the reliable trustee are better off
in the separating equilibrium, while the unreliable trustee is indiffer-
ent between the two situations). The separating equilibrium is not
necessarily second best to placed and honored trust without hostage
posting (see our comments on Theorem 1). If the separating equili-
brium exists, there is only one other equilibrium of HTGUI. Since
the separating equilibrium is also a weak Pareto-improvement
compared to the alternative equilibrium (trustor and reliable trustee
are better off in the separating equilibrium, the unreliable trustee is

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


348 RATIONALITY AND SOCIETY 16(3)

indifferent between the equilibria), a payoff dominance argument is


sufficient for the prediction that the separating equilibrium, if it
exists, is the solution of the game and that rational actors will
behave according to this equilibrium.
This model focuses on signaling via hostage posting that results
from differences between an unreliable and a reliable trustee with
respect to their transaction costs associated with hostage posting.
More precisely, we are interested in hostage posting as a signal
that the trustee is of the reliable type. An alternative way of allowing
for signaling via hostage posting would be to assume differences in
the value of the hostage between different types of trustees. Signaling
effects require that, in equilibrium, the trustor can infer the type of
the trustee from the hostage posting decision. Without any differ-
ences between the two types of trustees in the consequences of post-
ing a hostage, signaling effects are not possible since the unreliable
trustee could always mimic the reliable trustee and thereby invali-
date the trustor’s belief. Thus, without any differences between
the two types of trustees in the consequences of posting a hostage,
Granovetter’s claim that rational behavior would make signaling
unfeasible is correct. However, it seems reasonable to assume that
there are differences between the two types of trustees and Theorem 2
shows that this makes room for rational signaling.
From a substantive point of view and from the perspective of
these examples, the pooling equilibrium and the signaling equili-
brium are the important cases of trust based on hostage posting.
Raub (2003) provides equilibria for regions of the parameter space
of the game such that the conditions for an equilibrium according
to Theorem 1 and Theorem 2 are not fulfilled. In particular, it is
possible that, generically, all equilibria of the game are such that
both types of trustees do not post a hostage and the trustor with-
holds trust. A sufficient condition for this case is that the value of
the hostage is too small for the trustor in the sense that condition
(7) is not fulfilled. Also, no hostages are posted and the trustor with-
holds trust in equilibrium if the transaction costs of hostage posting
are too high for both types of trustee (see Raub 2003: Theorem 9 for
a precise specification of the relevant conditions).
We conclude the analysis of HTGUI by considering two special
cases that highlight the relevance of uncertainty for the compensa-
tion function of hostage posting and the relevance of incomplete
information for signaling via hostages. Consider first the hostage
game without uncertainty. Technically, assume that  ¼ 0 so that

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 349

unfavorable contingencies never obtain after trust has been placed


and abused. A generalization of an earlier result (Raub and Weesie
2000: 35) is then:

Corollary 1 Trust based on hostage posting without uncertainty.


Generically and for  ¼ 0 and K1  0, HTGUI has a pooling equili-
brium such that both types of trustees post a hostage, the trustor
places trust after hostage posting and the unreliable trustee honors
trust if and only if K2 > T2  R2 and r ; u < R2  P2 . Also, generic-
ally and for  ¼ 0 and K1  0, HTGUI has a separating equilibrium
such that the reliable trustee posts a hostage, the unreliable trustee
does not post a hostage, the trustor places trust after hostage post-
ing and withholds trust after no hostage posting if and only if (a)
K2 > T2  R2 and r < R2  P2 < u or (b) K2 < T2  R2 and
r < R2  P2 and K2 þ u > T2  P2 .

The corollary follows immediately from Theorem 1 and Theorem 2


for  ¼ 0 and K1  0 and shows that trust based on hostage post-
ing is feasible without positive compensation for the trustor if no
uncertainty is involved in the trust game. In this sense, the necessity
of compensation for trust based on hostage posting depends on
uncertainty. Note that there is no upper bound on the value of the
hostage for the trustee who posts the hostage. Uncertainty – the
possibility that the hostage is lost due to unfavorable contingencies
and even without opportunistic behavior – is necessary for such an
upper bound.
Finally, remove incomplete information from HTGUI (uncer-
tainty may or may not be involved in the underlying trust problem).
Technically, assume that  ¼ 1 so that the trustee is always
unreliable and HTGUI becomes a game of complete information.
A generalization of an earlier result (Weesie and Raub 1996: 218)
on hostage posting as a mechanism of trust in the standard trust
game without uncertainty is then:

Corollary 2 Trust based on hostage posting under complete infor-


mation. Generically and for  ¼ 1, HTGUI has a subgame perfect
equilibrium such that the trustee posts a hostage, the trustor places
trust after hostage posting and the trustee honors trust if and only if
K2 ð1  Þ > T2  R2 and K2 þ u < R2  P2 and K1 > R1 
ðR1  P1 Þ=  S1 .

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


350 RATIONALITY AND SOCIETY 16(3)

This corollary also follows immediately from Theorem 1 and


Theorem 2 and shows – not surprisingly – that signaling via hostages
presupposes a context with incomplete information.

5. Discussion

We have shown how hostage posting can promote trust through


binding the trustee, compensating the trustor, and signaling charac-
teristics of the trustee that are unobservable for the trustor. The
results follow from a game-theoretic analysis of trust based on hos-
tage posting in a game of uncertainty and incomplete information.

Conditions for Trust Based on Hostage Posting


We derived necessary and sufficient conditions for equilibria that
induce trust based on hostage posting. Our theorems first show
that the hostage has to be valuable enough for the trustee if the
hostage promotes trust by binding the trustee. Binding is an issue
if the trustee who posts the hostage may be unreliable – the reliable
trustee cannot abuse trust. Our theorem on the pooling equilibrium
(a hostage is posted by the unreliable trustee) specifies a lower bound
on the value K2 of the hostage for the unreliable trustee and this
lower bound depends on the unreliable trustee’s costs T2  R2 of
honoring trust in the trust game.
Second, we have seen that the hostage has to be valuable enough
for the trustor to induce trust through compensation. Our theorems
specify a lower bound on the value K1 of the hostage for the trustor.
The lower bound on the value of compensation depends crucially on
the amount of uncertainty that is associated with placing trust.
Uncertainty has been modeled by the probability  of unfavorable
contingencies so that things go wrong after trust has been placed
and honored. If unfavorable contingencies are sufficiently unlikely
(condition (3) is fulfilled), the lower bound on the value of com-
pensation is mild and requires only that the hostage does not pro-
vide too much disutility for the trustor. If the risk of unfavorable
contingencies is high (condition (3) is not fulfilled), the hostage
must provide positive compensation for the trustor. The threshold
for the probability  of unfavorable contingencies is also the thresh-
old that determines if the ‘no trust’-equilibrium of the underlying

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 351

trust game TGUI is Pareto-suboptimal so that TGUI is a dilemma.


If the probability  of unfavorable contingencies is below the thresh-
old, the ‘no trust’ equilibrium is Pareto-suboptimal. The underlying
trust problem is a dilemma if the trustor’s expected payoff from trust
that is placed and honored exceeds her payoff from withholding
trust. This is the case if and only if the probability of unfavorable
contingencies is below the threshold. In this case, a rational trustor
is willing to run the risks of uncertainty, given that the hostage
mitigates the risk of opportunistic behavior. Thus, if the probability
of unfavorable contingencies is small enough, it suffices that the
hostage binds the trustee or signals that the trustee is reliable and
it is not required that the hostage also includes positive compen-
sation of the trustor in case things go wrong.
The third result is that signaling via hostages is related to differ-
ences between different types of trustee in their transaction costs
associated with hostage posting. In this model, differences in trans-
action costs are the only possible differences between the two types
of trustees. Our theorem on the separating equilibrium shows that
signaling requires that the transaction costs of hostage posting are
lower for the type of trustee who posts the hostage than for the
other type of trustee who does not post the hostage. Moreover,
Corollary 2 formally establishes that signaling via hostages requires
a context with at least some incomplete information of the trustor on
characteristics of the trustee.
The fourth condition for trust based on hostage posting that
follows from this model is that transaction costs associated with
hostage posting have to be small enough. The theorems show that
an equilibrium in which a hostage is posted requires upper bounds
on the transaction costs of hostage posting for the (type of ) trustee
who posts the hostage. More precisely, the upper bound depends on
the gains R2  P2 from honored trust compared to the situation of
withheld trust. Also, the upper bound on the transaction costs of
hostage posting depends on the probability  of unfavorable con-
tingencies after trust has been placed and honored.
Finally, we have seen that the hostage should not be too valuable
for the trustee in order to allow for trust via hostage posting. Our
theorems provide upper bounds on the value K2 of the hostage for
a trustee who posts the hostage. The upper bound depends on the
gains R2  P2 from honored trust compared to the situation of with-
held trust. Corollary 1 shows that an upper bound on the value of

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


352 RATIONALITY AND SOCIETY 16(3)

the hostage for the trustee is only required if the underlying trust
problem involves uncertainty ( > 0) so that the trustee may lose
his hostage even without opportunistic behavior.
It is helpful to relate these results to our examples of trust prob-
lems and of hostages as a mechanism to induce trust. We conjecture
that the guarantee as a hostage in the case of buying and selling used
cars is a relatively valuable hostage for the car dealer (K2 is high) so
that the hostage serves binding purposes. Likewise, the guarantee is
valuable for the buyer (K1 is high) and provides positive compen-
sation. This would be consistent with uncertainty involved in such
transactions and a relatively high likelihood of unfavorable con-
tingencies. We guess that the guarantee does not serve signaling
purposes. It seems that guarantees are typically provided for second-
hand cars that are not too old and therefore not likely to need
frequent and expensive repairs. Restricting guarantees to relatively
new cars would be consistent with an upper bound on the trans-
action costs of hostage posting as well as an upper bound on the
value of the hostage for the dealer. The deposit as a hostage in the
car rental example – typically a relatively minor amount – may
bind the client to abstain from reckless driving, assuming that the
incentive for reckless driving is mild (it would be interesting to see
if the deposit for sports cars is typically higher than the deposit
for family cars with a similar price tag). The hostage seems to pro-
vide rather low compensation for the trustor (it would be interesting
to see if the deposit varies between regions with regional differences
in, say, overall accident rates). The transaction costs associated with
posting the hostage are minor and presumably hardly differ between
different types of clients so that the hostage hardly serves signaling
purposes. Finally, consider hostages provided by an employee as
safeguards for investments of the employer in the training and
schooling of the employee. A contract stipulating that the employee
has to reimburse the employer if he quits prematurely presumably
serves binding purposes and also provides some compensation for
the employer. The hostage ‘moving and acquiring real estate close
to the job’ does not provide compensation for the trustor. This
hostage seems to serve not only binding but also signaling purposes.
After all, the transaction costs of posting the hostage differ severely
between employees who have or expect an outside offer and those
who do not have and do not expect such an offer.

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 353

Testable Predictions
Consider testable predictions for laboratory experiments. Obviously,
such experiments require explicit assumptions on how material
incentives provided by the experiment relate to utility functions of
subjects. In an experimental design, HTGUI could be implemented
by using a random device that determines the type of the trustee at
the beginning of the experiment in such a way that the trustor knows
the relevant probabilities but cannot observe the outcome produced
by the random mechanism. An approach for deriving predictions for
experiments is to vary some parameter of HTGUI that is interesting
from a substantive perspective so that different theorems on trust via
hostage posting apply. To illustrate, consider a first scenario with
r ¼ u and parameters of HTGUI so that the theorem on the pool-
ing equilibrium applies. Now compare the first scenario with a
second that differs only in that u is increased so that condition (9)
for the separating equilibrium is fulfilled. One would then predict
less hostage posting by unreliable trustees in the second scenario,
no difference with respect to placement of trust by the trustor
after hostage posting, and also no difference with respect to honor-
ing trust by unreliable trustees. Another promising approach for
experimental tests is to vary the probability  of unfavorable con-
tingencies. For example compare the theorems on the pooling and
the separating equilibrium with Corollary 1 and consider that all
equilibria of HTGUI are such that both types of trustee do not
post a hostage and the trustor withholds trust if compensation is
too low (more precisely, compensation is too low if (7) is not ful-
filled). It follows that positive compensation should have a stronger
effect on hostage posting as well as on inducing trust by hostage
posting if the underlying trust problem involves a high probability
of unfavorable contingencies ( is large) than if one removes uncer-
tainty from the underlying trust problem ( ¼ 0). Research in this
direction seems to be particularly promising because previous experi-
mental evidence on hostages as a mechanism of cooperation in
dilemmas such as the trust game and the prisoner’s dilemma (Raub
and Keren 1993; Snijders 1996: chap. 6; Snijders and Buskens 2001)
seems to indicate that – empirically – positive compensation through
hostage posting has quite a strong effect on trust and cooperation.
It is easily seen that additional predictions for experimental tests
of the model can be generated by systematically varying one of the
parameters of the model so that HTGUI ‘moves’ between the sets

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


354 RATIONALITY AND SOCIETY 16(3)

of conditions for different types of equilibria specified in our


theorems.
A focus on the effects of uncertainty and the probability of un-
favorable contingencies also seems useful for empirical applications
outside the laboratory. The prediction would be that hostages
tend to provide more compensation for the trustor if the underlying
trust problem is one with a higher probability of unforeseen contin-
gencies. Conversely, keeping the probability of unforeseen con-
tingencies constant, one would expect that binding and signaling
properties of hostages are stronger (K2 increases and different
types of trustees differ strongly with respect to the transaction
costs of hostage posting) in the case of increasing risks from oppor-
tunistic behavior. Other testable predictions can be generated by
considering the effects of increasing differences in transaction costs
of hostage posting for different types of trustees. Based on our
theorem on the separating equilibrium, we would expect that
increasing differences in transaction costs come with more variation
in hostage posting decisions and also with an increase in hostages
that are not binding. Finally, predictions for social situations outside
the laboratory could focus on the characteristics of trustees who are
willing to post certain types of hostages. For example, an employee
will tend to move and acquire real estate close to his new job if his
transaction costs associated with posting this hostage are low; the
employee will be more likely to move if he has a rented flat rather
than privately owned real estate, if he is single or has a household
with a partner who is not active in the labor market. While predic-
tions of such characteristics may seem obvious, predictions of the
effects of hostage posting on subsequent behavior of trustor and
trustee may be more interesting. We would expect that the actual
period of employment is longer if the employee moves, that the
investments of the employer in the employee will be higher, and
that the employee will produce more output and will be more influ-
ential within the employer’s firm.

Research Agenda
Various questions on trust via hostage posting could and should be
tackled in future research. Some of these questions are inspired by
features of the examples that are not or not fully covered in our
model. First, consider the ‘expropriation hazard’ (Williamson

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 355

1985: 177) and the loss of flexibility (Becker 1991: 12-13) associated
with hostage posting. We have used the assumption that only the
trustee and not the trustor faces opportunities and incentives to
abuse trust. Our examples show, however, that this need not be
true. In the car rental example, a conflict between the rent-a-car
firm and the client may arise and the firm may keep the deposit of
the client strategically. Also, consider the employee’s hostage post-
ing by moving close to his job. This hostage is not subject to the
expropriation risk since the hostage as such is not valuable for the
employer but posting the hostage reduces the flexibility of the trustee
if contingencies change. Such changing contingencies can depend on
‘moves of Nature’ rather than on incentive-guided behavior of the
employer. An example is the risk that the firm goes bankrupt due
to market fluctuations rather than bad management of the employer
as owner/manager. However, contingencies may also change due to
strategic behavior of the employer after the employee’s hostage post-
ing. For example, when hiring the employee, the employer may offer
not only investments in future training and schooling but may also
promise future general policies that are attractive for the employee.
Such promises are often non-contractual and not legally binding or
enforceable. If the employee commits himself by moving, he loses
flexibility to react to the employer’s future deviations from the
promises. In the model developed here, we have included the expro-
priation risk and the expected costs of loss of flexibility in the trans-
action costs associated with hostage posting. This seems satisfactory
for risks that depend on moves of Nature rather than strategic
behavior of the partner. A more appropriate analysis that takes
the strategic behavior of the other player into account would require
to model hostage posting as a mechanism of cooperation in more
complex dilemma situations with possibilities for opportunistic
behavior from more than one player. Reciprocal hostage posting
by both players is an obvious mechanism for mitigating risks from
two-sided opportunistic behavior (see Williamson 1985: chap. 8 for
examples; Weesie and Raub 1996 for formal model building using
games with certainty and complete information; and Raub and
Keren 1993 for some experimental results).
While the expropriation risk and the loss of flexibility are
problems associated with hostage posting, one should not overlook
possible further benefits of hostage posting in addition to promoting
trust. One such benefit that merits closer analysis is that posting

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


356 RATIONALITY AND SOCIETY 16(3)

the hostage may make the relation between trustor and trustee more
productive. For example, the employee who moves close to his
job may not only promote investments of the employer in the
employee’s training and schooling. Moving close to his job, more-
over, may imply an additional contribution to the employee’s
productivity and job satisfaction since he benefits from reduced
commuting time, better (opportunities for) contacts and interaction
with colleagues, etc. It should be easy to show that ‘productivity’ of a
hostage in this sense implies that the conditions for trust based on
hostage posting become less restrictive since posting the hostage
also increases the payoffs Ri from trust that is placed and honored.
A variant on this theme is that hostage posting by the trustee is
associated with costs for the trustor. For example, the used car
with a guarantee comes at a higher price for the buyer. This would
reduce R1 after a hostage has been posted. In this scenario, the con-
ditions for trust based on hostage posting become more restrictive.
We have motivated an analysis of hostage posting as a mechanism
of trust with the observation that dyadic embeddedness and network
embeddedness can be insufficient for stabilizing trust. Our model,
however, is one without these forms of embeddedness rather than
one in which dyadic and network embeddedness are too ‘weak’ to
stabilize trust without using hostages. For example, the dealer in
used cars may (want to) sell used cars infrequently but nevertheless
repeatedly to the same buyer or there may be some information
exchange about the dealer among buyers. Accounting for the com-
bined effects of learning and control through dyadic and network
embeddedness and of opportunities for hostage posting through
institutional embeddedness remains a challenging task for further
model building (see Weesie et al. 1998 and Raub and Snijders 2001
for some work in this direction).
A final feature of our examples that is not yet modeled in HTGUI
is hostage selection. The employee can post a hostage by moving to
his job or by signing a contract stipulating that he pays back training
and schooling costs of the employer after a quit. When to choose one
or the other of these hostages, when both? It seems reasonable to
assume that the transaction costs associated with posting different
hostages as well as the value of the hostage will be important factors
that affect the trustee’s decision but a more thorough analysis seems
useful (see Snijders 2000 for first steps in the analysis of the hostage
selection problem).

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 357

NOTE

I am indebted to Jeroen Weesie for comments on this paper and earlier joint work on
hostages as a mechanism of trust and cooperation. Comments by Marcel van Assen,
Davide Barrera, Vincent Buskens, Thomas Gautschi, Chris Snijders, and seminar
participants at Utrecht University are gratefully acknowledged. Comments by
anonymous reviewers were helpful when revising the paper. The research on which
the paper is based is part of the PIONIER-program ‘The Management of Matches’
of the Netherlands Organization for Scientific Research (NWO; grant PGS 50-370).
A first draft was written at the Department of Sociology, Berne University. The hos-
pitality of Andreas Diekmann and other colleagues in Berne was much appreciated.

REFERENCES

Akerlof, G. A. 1970. ‘The Market for ‘Lemons’: Quality Uncertainty and the Market
Mechanism.’ Quarterly Journal of Economics 89: 488–500.
Arrow, K. J. 1974. The Limits of Organization. New York, NY: Norton.
Axelrod, R. 1984. The Evolution of Cooperation. New York, NY: Basic Books.
Bacharach, M. and D. Gambetta. 2001. ‘Trust in Signs.’ In Trust in Society, ed.
K. S. Cook, pp. 148–84. New York, NY: Russell Sage.
Barber, B. 1983. The Logic and Limits of Trust. New Brunswick, NJ: Rutgers Univer-
sity Press.
Barrera, D. and V. Buskens. 2003. ‘Imitation and Learning under Uncertainty:
A Vignette Experiment.’ Utrecht. Mimeographed.
Becker, G. S. 1964. Human Capital. New York, NY: NBER, Columbia University
Press.
Becker, G. S. 1991. A Treatise on the Family, enlarged edn. Cambridge, MA: Harvard
University Press.
Burt, R.S. and M. Knez. 1995. ‘Kinds of Third-Party Effects on Trust.’ Rationality
and Society 7: 255–92.
Buskens, V. 2002. Social Networks and Trust. Boston, MA: Kluwer.
Buskens, V. and W. Raub. 2002. ‘Embedded Trust: Control and Learning.’ Advances
in Group Processes 19: 167–202.
Calvert, R. L. 1995. ‘Rational Actors, Equilibrium, and Social Institutions.’ In
Explaining Social Institutions, eds J. Knight and I. Sened, pp. 57–94. Ann Arbor,
MI: University of Michigan Press.
Camerer, C. and K. Weigelt. 1988. ‘Experimental Tests of a Sequential Equilibrium
Reputation Model.’ Econometrica 56: 1–36.
Coleman, J. S. 1990. Foundations of Social Theory. Cambridge, MA: Belknap Press of
Harvard University Press.
Dasgupta, P. 1988. ‘Trust as a Commodity.’ In Trust: Making and Breaking Coopera-
tive Relations, ed. D. Gambetta , pp. 49–72. Oxford: Blackwell.
Esser, H. 1993. Soziologie. Allgemeine Grundlagen. Frankfurt: Campus.
Frank, R. 1988. Passions Within Reason. New York, NY: Norton.
Granovetter, M. 1985. ‘Economic Action and Social Structure: The Problem of
Embeddedness.’ American Journal of Sociology 91: 481–510.

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


358 RATIONALITY AND SOCIETY 16(3)

Granovetter, M. 1992. ‘Economic Institutions as Social Constructions: A Framework


for Analysis.’ Acta Sociologica 35: 3–11.
Granovetter, M. 2002. ‘A Theoretical Agenda for Economic Sociology.’ In The New
Economic Sociology: Developments in an Emerging Field, ed. M.F. Guillén et al.,
pp. 35–60. New York, NY: Russell Sage.
Güth, W. and A. Ockenfels. 2002. ‘The Coevolution of Trust and Institutions in
Anonymous and Non-anonymous Communities. Jena. Mimeographed.
Hardin, R. 2002. Trust and Trustworthiness. New York, NY: Russell Sage.
Kelley, H. H. and J. W. Thibaut. 1978. Interpersonal Relations: A Theory of Inter-
dependence. New York, NY: Wiley.
Kreps, D. M. 1990. ‘Corporate Culture and Economic Theory.’ In Perspectives on
Positive Political Economy, eds J. E. Alt and K. A. Shepsle, pp. 90–143. Cambridge:
Cambridge University Press.
Kreps, D. M., P. Milgrom, J. Roberts and R. Wilson. 1982. ‘Rational Cooperation
in the Finitely Repeated Prisoner’s Dilemma.’ Journal of Economic Theory 27:
245–52.
Lindenberg, S. 1992. ‘An Extended Theory of Institutions and Contractual
Discipline.’ Journal of Institutional and Theoretical Economics 148: 125–54.
Lindenberg, S. 1994. ‘Norms and the Power of Loss: Ellickson’s Theory and Beyond.’
Journal of Institutional and Theoretical Economics 150: 101–13.
Macaulay, S. 1986. ‘Private Government. In Law and the Social Sciences, eds
L. Lipson and S. Wheeler, pp. 445–518. New York, NY: Russell Sage.
Nash, J. 1951. ‘Non-Cooperative Games.’ Annals of Mathematics 54: 286–95.
Rapoport, A. 1974. ‘Introduction.’ In Game Theory as a Theory of Conflict Resolution,
ed. A. Rapoport, pp. 1–14. Dordrecht: Reidel.
Rasmusen, E. 1994. Games and Information: An Introduction to Game Theory, 2nd ed.
Oxford: Blackwell.
Raub, W. 2003. ‘Hostage Posting as a Mechanism of Trust: Further Results.’ Utrecht.
Mimeographed.
Raub, W. and G. Keren. 1993. ‘Hostages as a Commitment Device: A Game-theoretic
Model and an Empirical Test of Some Scenarios.’ Journal of Economic Behavior and
Organization 21: 43–67.
Raub, W. and C. Snijders. 2001. ‘A Reluctant Match: Models for the Analysis of
Trust in Durable Two Party Relations.’ In Trust and Deception in Virtual Societies,
eds C. Castelfranchi and Y.-H. Tan, pp. 195–220. Dordrecht: Kluwer.
Raub, W. and J. Weesie. 1990. ‘Reputation and Efficiency in Social Interactions:
An Example of Network Effects.’ American Journal of Sociology 96: 626–54.
Raub, W. and J. Weesie. 2000. Cooperation via Hostages.’ Analyse und Kritik 22:
9–43.
Schelling, T. C. 1960. The Strategy of Conflict. London: Oxford University Press.
Schotter, A. 1981. The Economic Theory of Social Institutions. Cambridge: Cambridge
University Press.
Snijders, C. 1996. Trust and Commitments. Amsterdam: Thesis.
Snijders, C. 2000. ‘Trust via Hostage Posting.’ In The Management of Durable Rela-
tions. Theoretical Models and Empirical Studies of Households and Organizations,
eds J. Weesie and W. Raub, pp. 114–16. Amsterdam: ThelaThesis.
Snijders, C. and V. Buskens. 2001. ‘How to Convince Someone That You Can Be
Trusted? The Role of ‘‘Hostages’’.’ Journal of Mathematical Sociology 25: 355–83.

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 359

Spence, A. M. 1974. Market Signaling: Information Transfer in Hiring and Related


Processes. Cambridge, MA: Harvard University Press.
Taylor, M. 1987/1976. The Possibility of Cooperation. Cambridge: Cambridge Univer-
sity Press. (Rev. ed. of Anarchy and Cooperation, London: Wiley 1976.)
Weesie, J., V. Buskens and W. Raub. 1998. ‘The Management of Trust Relations via
Institutional and Structural Embeddedness.’ In The Problem of Solidarity: Theories
and Models, eds P. Doreian and T. Fararo, pp. 113–38. Amsterdam: Gordon and
Breach.
Weesie, J. and W. Raub. 1996. ‘Private Ordering: A Comparative Institutional
Analysis of Hostage Games.’ Journal of Mathematical Sociology 21: 201–40.
Williamson, O. E. 1985. The Economic Institutions of Capitalism. New York, NY:
Free Press.
Wippler, R. and S. Lindenberg. 1987. ‘Collective Phenomena and Rational Choice.’
In The Micro–Macro Link, eds J. C. Alexander et al., pp. 135–52. Berkeley, CA:
University of California Press.

Appendix A. Notation and Assumptions on Parameters

Symbol Description and Assumptions

Trust Game with Uncertainty and Incomplete Information (TGUI)


Ci Placing (for i ¼ 1) and honoring (for i ¼ 2) trust
Di Withholding (for i ¼ 1) and abusing (for i ¼ 2) trust
R1 ; P1 ; S1 ; T2 ; R2 ; P2 ; R1 Payoffs in the trust game; R1 > P1 > S1 and
T2 > R2 > P2 ; R1 ¼ S1 þ ð1  ÞR1
 Probability that trustor receives S1 after trust is placed and
honored in TGUI; 0   < 1
 Probability that trustee is unreliable in TGUI;
maxððR1  P1 Þ=ðR1  S1 Þ; 0Þ <   1

Hostage Game for the Trust Game with Uncertainty and Incomplete Information
(HTGUI)
H þ, H  Trustee’s decision to post (H þ ) or not to post (H  ) a
hostage
Ki Value of the hostage for player i ; K2 > 0
r ; u Transaction costs of hostage posting for reliable (r )
and unreliable (u ) trustee; r ; u > 0

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


360 RATIONALITY AND SOCIETY 16(3)

Appendix B. Proofs of the Theorems

Theorem 1: Pooling equilibrium with trust based on hostage posting.


We show first that (8) implies the existence of the pooling equili-
brium described in the theorem. To see this, note that Lemma 1
and Lemma 2 imply that the unreliable trustee uses a best reply
strategy under the first two conditions in (8), given the equilibrium
strategies of the reliable trustee and of the trustor. Lemma 2 implies
that the reliable trustee uses a best reply strategy under the second
condition in (8), given the equilibrium strategies of the unreliable
trustee and of the trustor. Lemma 4 then implies that the trustor
uses a best reply strategy under the fourth condition in (8), given
the equilibrium strategies of both types of trustees. This establishes
that (8) implies the existence of the pooling equilibrium described in
the theorem.
Next, we show that the existence of the pooling equilibrium
described in the theorem implies (8). To see this, we prove that a
pooling equilibrium described in the theorem does not exist if at
least one of the conditions from (8) is violated. If the first condition
in (8) is violated, it follows from Lemma 1 that best reply behavior of
the unreliable trustee is inconsistent with honoring trust after hos-
tage posting and placement of trust. If the second condition in (8)
is violated for the unreliable trustee, it follows from Lemma 2 that
best reply behavior of the unreliable trustee is inconsistent with
hostage posting, if the strategies of trustor and reliable trustee are
as in the pooling equilibrium. If the second condition in (8) is vio-
lated for the reliable trustee, it follows from Lemma 2 that best
reply behavior of the reliable trustee is inconsistent with hostage
posting, if the strategies of trustor and unreliable trustee are as in
the pooling equilibrium. Finally, if the third condition in (8) is
violated, it follows from Lemma 4 that best reply behavior of the
trustor is inconsistent with placement of trust after hostage posting,
if the strategies of the two types of trustees are as in the pooling equi-
librium. Thus, the existence of the pooling equilibrium described in
the theorem implies (8).
Finally, we consider conditions such that another equilibrium
exists together with the pooling equilibrium and derive the proper-
ties of such an alternative equilibrium. Note first that (8) implies
that the trustor will place trust after hostage posting in equilibrium.
Also, assume an equilibrium such that both types of trustees surely
post a hostage. This implies that placing trust by the trustor is on the

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 361

equilibrium path and then the equilibrium would be the pooling


equilibrium as described in the theorem. Next, note that (8) excludes
an equilibrium such that (a) the unreliable trustee always posts a
hostage while the reliable trustee never does so, or (b) the unreliable
trustee never posts a hostage while the reliable trustee always does
so, or (c) both types of trustees never post a hostage. If (a) is part
of equilibrium behavior, not posting a hostage would signal that
the trustee is reliable and the unreliable trustee would not use a
best reply. If (b) would be part of equilibrium behavior, posting a
hostage would signal that the trustee is reliable and the trustor
would place trust after hostage posting but would not place trust
after no hostage posting. Again, the unreliable trustee would not
use a best reply. Finally, (c) cannot be part of equilibrium behavior
for both reliable and unreliable trustee, because the trustor would
not place trust after no hostage posting but would place trust
under (8) after hostage posting.
Thus, if an equilibrium other than the pooling equilibrium exists,
at least one type of trustee randomizes in that equilibrium with
respect to his hostage posting decision. Assume now that in such
an equilibrium the unreliable trustee always posts a hostage while
the reliable trustee randomizes between posting and not posting a
hostage. Not posting a hostage would thus signal that the trustee
is reliable, the trustor would place trust after no hostage posting,
and again the hostage posting decision of the unreliable trustee
would not be optimal. Conversely, assume an equilibrium such
that the unreliable trustee never posts a hostage while the reliable
trustee randomizes between posting and not posting a hostage. Post-
ing a hostage would thus signal that the trustee is reliable, so the
trustor would place trust after hostage posting but would not
place trust after no hostage posting, and again the hostage posting
decision of the unreliable trustee would not be optimal. Thus, if
an equilibrium other than the pooling equilibrium exists, the unreli-
able trustee randomizes in that equilibrium with respect to his
hostage posting decision. This requires that the unreliable trustee
is indifferent in that equilibrium between posting and not posting
a hostage, given the equilibrium strategies of trustor and reliable
trustee. Note further that if an equilibrium other than the pooling
equilibrium exists and if the unreliable trustee randomizes in that
equilibrium with respect to his hostage posting decision, the reliable
trustee uses a strategy such that he posts no hostage at least with
positive probability. Namely, if the reliable trustee definitely posts

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


362 RATIONALITY AND SOCIETY 16(3)

a hostage in the equilibrium, the trustor would place trust after hos-
tage posting and would not place trust if no hostage has been posted.
Then, however, the unreliable trustee would be better off posting a
hostage rather than not posting a hostage, which would be incon-
sistent with the assumption that the unreliable trustee randomizes
with respect to his hostage posting decision. To summarize, the
equilibrium different from the pooling equilibrium described in the
theorem has to be such that the unreliable trustee randomizes with
respect to his hostage posting decision, the reliable trustee posts
no hostage at least with positive probability, and the trustor places
trust after a hostage has been posted.
Next, we observe that if an equilibrium different from the pooling
equilibrium exists, the trustor randomizes in that equilibrium after
no hostage has been posted: after no hostage posting, she places
trust with probability  and withholds trust with probability 1  
ð0 <  < 1Þ. This can be seen as follows. If the trustor always
places trust after no hostage posting, the supposed equilibrium strat-
egy of the unreliable trustee could not be a best reply since he would
be better off by posting no hostage. If the trustor never places trust
after no hostage posting, both types of trustee would not behave
optimally with respect to their hostage posting decisions. Note
that the trustor randomizes in the equilibrium after no hostage has
been posted if she is indifferent in the equilibrium between placing
and withholding trust after no hostage posting. This requires that
condition (3) is fulfilled so that an equilibrium different from the
pooling equilibrium cannot exist if (3) is violated.
Denote the expected payoff for the unreliable trustee in the equi-
librium after no hostage posting by Uu ðH  Þ and his expected payoff
after hostage posting by Uu ðH þ Þ. We must have

Uu ðH  Þ ¼ T2 þ ð1  ÞP2 ¼ Uu ðH þ Þ
¼ ð1  ÞðR2  u Þ þ ðR2  K2  u Þ
and thus  ¼ ðR2  K2  u  P2 Þ=ðT2  P2 Þ It is easily seen that
0 <  < 1. Also, denote the expected payoff for the reliable trustee
in the equilibrium after no hostage posting by Ur ðH  Þ and his
expected payoff after hostage posting by Ur ðH þ Þ. We must have
Ur ðH  Þ ¼ R2 þ ð1  ÞP2  Ur ðH þ Þ
¼ ð1  ÞðR2  r Þ þ ðR2  K2  r Þ

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 363

and thus   ðR2  K2  r  P2 Þ=ðR2  P2 Þ. Thus, if the equili-


brium different from the pooling equilibrium exists,
ðR2  K2  u  P2 Þ=ðT2  P2 Þ  ðR2  K2  r  P2 Þ=ðR2  P2 Þ
(11)

and this implies r > u . Note that (11) is fulfilled if u is sufficiently


small and r is sufficiently close to the upper bound R2  K2  P2
from the second condition in (8). Note finally that the trustor is
indifferent in the equilibrium between placing and withholding
trust after no hostage posting if in the equilibrium the probability
that the trustee is unreliable after no hostage posting is sufficiently
small. This completes the proof of the theorem (note that it has
already been indicated in section 4 that the pooling equilibrium is a
weak Pareto-improvement compared to the alternative equilibrium).

Theorem 2: Separating equilibrium with hostage posting by the reli-


able trustee and placement of trust. We start by showing that (9)
implies the existence of the separating equilibrium described in the
theorem. Using our Lemmas, we see that the first two conditions
in (9) imply that the strategy of the unreliable trustee in the separat-
ing equilibrium is indeed a best reply against the equilibrium strate-
gies of the trustor and the reliable trustee. Similarly, the third
condition in (9) implies that the strategy of the reliable trustee in
the separating equilibrium is indeed a best reply against the equili-
brium strategies of the trustor and the unreliable trustee, while the
fourth condition in (9) implies that the strategy of the trustor in
the separating equilibrium is indeed a best reply against the equili-
brium strategies of the reliable and the unreliable trustee. In the
same way, one shows that (10) implies the existence of the separating
equilibrium described in the theorem.
Next, we show that the existence of a separating equilibrium as
described in the theorem implies that (9) or (10) hold. First, since
the trustor uses a best reply strategy, it must be optimal for the
trustor to place trust after hostage posting and to withhold trust
after no hostage posting. Given the equilibrium strategies of both
types of trustees, this implies the fourth condition in (9) and (10).
Since the reliable trustee uses a best reply strategy, it must be optimal
for the reliable trustee to post a hostage. Given the equilibrium
strategies of the trustor and of the unreliable trustee, this implies
the third condition in (9) and (10). Now we consider the unreliable

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


364 RATIONALITY AND SOCIETY 16(3)

trustee. Since he uses a best reply strategy, it must be optimal for the
unreliable trustee not to post a hostage, given the equilibrium
strategies of the trustor and of the reliable trustee. By deviating
from his strategy and posting a hostage, the unreliable trustee
could obtain T2  K2  u if he abuses trust and (1  ÞðR2  u Þ þ
ðR2  K2  u ) if he honors trust after hostage posting and place-
ment of trust. The existence of the separating equilibrium implies
that P2 – the payoff of the unreliable trustee if he follows his equili-
brium strategy – is higher than both these alternative payoffs.
Assume now T2  K2  u > ð1  ÞðR2  u Þ þ ðR2  K2  u Þ.
This implies the first condition in (9) and the existence of the separ-
ating equilibrium then also implies the second condition in (9).
Assume T2  K2  u < ð1  ÞðR2  u Þ þ ðR2  K2  u Þ. This
implies the first condition in (10) and the existence of the separating
equilibrium then also implies the second condition in (10).
Finally, we have to show that (9) as well as (10) imply that the only
other equilibrium next to the separating equilibrium is the additional
equilibrium described in the theorem. Consider first properties of
an equilibrium if (9) holds. Using the Lemmas, the first condition
in (9) implies that the unreliable trustee abuses trust after no hostage
posting and placement of trust but honors trust after hostage post-
ing and placement of trust. However, the second condition in (9)
implies that the unreliable trustee prefers no hostage posting and
no placement of trust to hostage posting, placement of trust and
honoring trust. Thus, (9) implies that the unreliable trustee posts
no hostage. Assume now that the trustor places trust after hostage
posting. In this case, the best reply of the reliable trustee is to post
a hostage due to the third condition in (9). In that case, placing
trust after hostage posting would indeed be optimal for the trustor
due to the fourth condition in (9) and we get the separating equili-
brium. Assume, however, that the trustor withholds trust after
hostage posting. In that case, the best reply for the reliable trustee
would be not to post a hostage and we get the alternative equili-
brium described in the theorem. Note that hostage posting is not
an element of the equilibrium path of play induced by the alternative
equilibrium. This completes the proof of the theorem (note that it
has already been indicated in section 4 that the separating equili-
brium is a weak Pareto-improvement compared to the alternative
equilibrium).

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011


RAUB: HOSTAGE POSTING 365

WERNER RAUB is a Professor of Sociology at Utrecht Univer-


sity and the Interuniversity Center for Social Science Theory and
Methodology (ICS). His research interests include cooperation
theory, applications of game theory and related formal models in
social science theory and empirical research, and organization
studies.

ADDRESS: Department of Sociology/ICS, Utrecht University,


Heidelberglaan 2, 3584 CS Utrecht, The Netherlands
[email: w.raub@fss.uu.nl].

Downloaded from rss.sagepub.com at Umea University Library on April 2, 2011

Das könnte Ihnen auch gefallen