Audience? Costs?

Imagine political games on the international political stage, momentum builds up between the two lead players...who will win?
A bit of theory of this, potentailly dangerous, game to understand tensions might help to understand what may, and does, happen in reality, not just when emotions seem to prevail but some 'moves' actually make sense rationally. A useful tool is the concept of audience costs (one of the plethora of game theoretical stuff used in political science). The term was first introduced by Fearon (1994) and describes a situation on the international political stage when a leader of one country backs down in an international crisis with another country. The longer the duration of the crisis, the more 'cost' the leader builds up, where payoff time of these costs depends on behaviour and decisions of the leader. Payment should not be interpreted in monetary terms (although, sure, s/he may lose money), but a measurable extension of loss of reputation in the form of not being re-elected by the public as the most serious incurred cost (that is, assuming the leader wants to, and at least theoretically, can be re-elected). This definition begs the question how audience costs can be measured in 'non-democracies', and for non-state terrorists, to which I will return after addressing general audience cost related factors.

The Game

The game is modelled as a War of Attrition, having a continuum of Nash equilibria, though with three options during the international crisis in the set of strategies (instead of two): attack the other country, back down or continue the crisis. The cost of continuing is mapped onto the discounted value for each round, hence imagine this as the increase of the build-up of the audience cost. Paying the cost counts for both challenging states backing down and for a challenged state that first resists and subsequently backs down. This model predicts that democracies, being able to generate more audience costs than a non-democracy, and therefore more capable of signalling their intention more accurately, are less likely to back down in a crisis situation.

Generating audience costs

Smith (1998) argues that, when in equilibrium, only the least competent leaders will back down during a crisis and will pay audience costs and that "The possibility of war is necessary to keep leaders honest", after all, when signals do not threaten they are worthless. Some others (Brito and Intriligator, 1985) claim it has its basis in asymmetric information in order to prevent bluffing by the informed state (never mind a "dodgy dossier" being "sexed-up"). However, the possibility that a leader may back down because new information emerged that would make a war unjustified, hence not legal and/or foolish to continue the crisis or attack is conveniently ignored (a leader can't be wrong, right?...)
It goes even further: "non-intervention signals lower competence" (Smith 1998:633) of the leader (which leaves one wondering about the status of diplomacy in these affairs). Inherent in the War of Attrition is the tendency towards "belligerent equilibria" (Myerson, 1991:330) and exacerbated by the two-tier bargaining (the delicate balance between domestic and international politics), it results in a bias towards hawkish strategies. A striking example of the limitations on audience cost build-up and the strong bellicose leader bias is the Iraqi crisis in 2002/2003: 'dove' US Secretary of State Colin Powell made a U-turn towards supporting an invasion of Iraq; though strictly according to the audience cost definition, he would not have suffered credibility, but he did.
An extension of the domestic politics factor of audience cost generation is the influence of an opposition party on the stance of the leader (/government), in addition to the voting public. An opposition party can lend additional credibility to threats signalled by the government when 'even the opposition supports the government's stance', but also makes the government more selective in signalling, in the form of the opposition as 'watchdog' because the opposition has no incentive to support a bluffing government. The credibility an opposition party lends to the government makes the leader more selective, but stronger, in the international crises. Because non-democracies do not have one or more opposition parties, nor voting citizens, they cannot build up audience costs like their democratic counterparts, in turn affecting the credibility of a threat, or any signalling for that matter .
Prins (2003) adds that institutional stability in general allows for more precise signalling (conversely, instability hampers successful signalling of true intentions) and regimes "with non-institutionalized political participation engage in more escalatory behavior".
However, interestingly, Roddy (2003) observed George W. Bush's build up to the Iraqi invasion that "the steering wheel long ago exited the driver's side window", suggesting that the 'strong and powerful leader' who does not back down according to the model, has actually lost control over his own power.

What about terrorists?

With outlined variations and extensions of the audience costs model, it indicates that strong leaders never back down (weak leaders do) and when the leader has a stable democratic apparatus behind him, the threats he's signalling are more credible and better reveal his true intentions than his non-democratic counterpart. Despite the fact that there is still plenty of further research possible on the audience costs in international politics, I endeavour to apply the concept of audience costs to the terrorist theatre, and assess factors that need to be addressed in order to make it a possible useful tool.
First, would it be possible for terrorist groups to generate audience costs according to aforementioned definition within their own supporters group, and in the 'electorate'? (The latter envisaged as citizens of the affected area, including supporters and non-supporters.) Aside from a few exceptions, aggrieved group leaders do not get voted into government nor can be voted out of office or Politburo every four or five years, which would make the situation analogous to a non-democracy, and less capable of generating audience costs in the first place. Restricting the possibility of generating audience cost to the inner working of an organised aggrieved group, for example the leader promising a new and better world or more equal pay to his followers which does not materialize, is an interesting avenue for investigation. Crenshaw (1991) asserts that one of the reasons terrorism declines is through organisational disintegration, which thus could be an effect of a bluffing leader, though data is hard to find and inconclusive (like pretty much any quantitative data on terrorism).
Second, even though I cannot assess the internal audience cost build-up, and the organisation of an aggrieved group is not as democratic as a democratic state, common sense points towards a likelihood of audience cost generation with the wider public, as it is exactly the threats made by these organisations that contribute for a large part to their importance. However, from the game theoretical framework outlined by Fearon, Smith and Schultz among others, this cannot be possible. It is easy to assume that either there is something lacking in the model, or the people rationally should not believe the threats because they originate from unreliable sources (according to the definition). There is another option: an aggrieved group exploiting the audience cost model, as opposed to being 'trapped' in it like a state leader. The reasoning is as follows: the aggrieved group commits a terrorist act, succeeded by several threats that are not carried out, leading the people to believe the aggrieved group is not trustworthy in its threats. The people are lulled into a sense of security, relax imposed restrictions and foster the idea that the terrorist act was an isolated event, and then the aggrieved group actually implements a threatened action. Thereby the aggrieved group is taking advantage of the less credible signalling, messing up the neat Bayesian updating of the public's belief system about the terrorist organisation in that the probabilities cannot be realistically updated. Worded differently: one can update the probability of the type of player after each threat, but this does not provide more information on the aggrieved group and/or 'terrorist' leader, defeating the main point that Bayesian updating is supposed to deliver in a game; alike a War of Nerves instead of a War of Attrition. This is formulated in the following proposition:

The effectiveness of threats signalled by an aggrieved group has a basis in the unreliability of the signalling compared to international politics, whereby identification of the type of player based on Bayesian updating is corrupted and cannot provide the same increase in the level of information as in the standard audience cost model, thereby exploiting the model.

Some negotiations with terrorists

A more positive aspect on the potential for generating audience cost is when the aggrieved group is part of a peace solution, or at least taking part in negotiations to achieve a peace agreement. Kydd and Walter's (2002) extensive form game with Bayesian updating and separating or pooling equilibria, analyses terrorist violence as a problem of trust. The lack of trust was a problem with the credibility of signalling in the section above, but a peace negotiation is a distinct setting and signalling can be effective in determining the type of player, i.e. if the opponents are moderate or violent terrorists, weak or strong and trustworthy or not. Intriguingly, their observation goes against the audience cost model as well, in that weak moderates "may be forgiven for failing to prevent terrorist attacks, but strong moderates will not". The moderates, bargaining with the government, may promise peace, thus signalling their intentions in the same manner as a state leader may do, but if they're weak, they won't have to pay audience costs (in full), because it is not expected that they could curtail extremists anyway. Therefore, the incurred terrorist acts by extremists to avert a peace deal provide useful information not about the extremists themselves, but about the strength of the moderates on the moderates' capabilities to curtail the extremists (two 'game boards' that might come in mind is the Arab/Israeli conflict and Northern Ireland). This leads to a paradox that weak moderates are better off in peace negotiations when there is an active violent faction, yet a weak negotiator achieves less in a bargaining process. Alternatively, is it like before exploitation of the model, in that an aggrieved group has an incentive to be perceived as weak, yet strong at the negotiation table? For why is it, that the combination violent terrorist and affiliated political party lasts longer then either one of them separately? If they were to be more effective when working in tandem or complementing one another's strategy, should they not only be capable of persisting longer, but also come to a resolve faster as they are 'battling on two fronts'? I have no answer to this based on empirical data, but bargaining strength does shed some light on this. However, what it does imply according to Kydd and Walter (2002), is an indicator for due audience costs when the negotiations involve a strong moderate aggrieved group: if violence does occur, the other player infers that the moderates have been bluffing and not capable of keeping their commitments.
Third, when agreeing that aggrieved groups can build-up audience costs, albeit not in the same manner as in the standard international political scene, is this quantitatively measurable? This faces the same problems as Schultz (2001) discussed. An option to overcome this would be to rely on opinion polls; with all its imperfections not ideal either. Besides, establishing baseline credibility poses a problem, as well as (subjectively?) deciding if with every statement, bluff and lie the terrorist group should always be deducted equally as the government. Intuitively, catching a bluffing democratic government seems more serious than a lying terrorist organisation, but this lies in the eye of the beholder as well as the parameters of the situation/game. Take for example a hostage situation: one subjectively may assume that a dishonest government lying to release hostages might be deducted less, i.e. incur lower audience costs, than unreliable hostage takers. Although according to Lapan and Sandler (1988:16), governments will lose reputation "when governmental declarations are not completely credible and uncertainty characterizes the government's costs of not negotiating" (say, the Iran-contra scandal). Thus a policy stance to never negotiate with terrorists is likely to be time inconsistent and implausible (US diplomates actually negotiate with Hezbollah, according to the Lebanese newspaper The Daily Star), a factor affecting audience costs related to the terrorist theatre, but which could have less impact than unclear positions in the more regulated international political arena.

And finally...

In terrorist frameworks like peace negotiations, audience costs can be generated and identified, in the non-negotiation phase, aggrieved groups exploit the audience cost model to their own benefit, and audience cost modelling parameters, especially the rate of deduction in crisis prolongation, depends on the problem being modelled and the preference of the modeller.
I leave it up to the reader to ponder about further applicability of the above...

Consulted literature:
- Brito, Dagobert L and Intriligator, Michael D., (1985), 'Conflict, war, and redistribution'. American Political Science Review, 79, 943-957.
- Crenshaw, Martha, (1991), 'How terrorism declines'. Terrorism and political violence. 3(3), 69-87.
- Fearon, James D., (1994), 'Domestic Political audiences and the escalation of international disputes'. American Political Science Review, 88(3), 577-592.
- Kydd, Andrew and Walter, Barbara F., (2002), 'Sabotaging the peace: the politics of extremist violence'. International Organization, 56(2), 263-296.
- Lapan, Harvey E. and Sandler, Todd, (1988), 'To bargain or not to bargain: that is the question'. American Economic Review, 78(2), 16-27.
- Myerson, Roger B., (1991), Game theory - Analysis of conflict. Massachusetts: Harvard University Press. 567 p.
- Roddy, Dennis B., (2003), 'Bush is playing 'chicken' not only with Saddam, but with the U.N. and allies, as well'. Post-Gazette, 16-3-2003.
- Schultz, Kenneth A., (2001), 'Looking for audience costs'. Journal of Conflict Resolution, 45(1), 32-60.
- Smith, Alastair, (1998), 'International crises and domestic politics'. The American Political Science Review, 92(3), 623-638.

More info on applying game theory to "settings involving aggrieved groups generally branded as terrorist organisations" is available via: (the above is a slightly modified version of a paragraph of my dissertation 'Terrorism and Game Theory').