Risk, Boundary Solutions, Suicide Attacks, Political Bias Flashcards
Risk and Conflict: Fearon explicitly ruled out risk-loving behaviour in his typology of conflict; he places it in the category of “irrational”. The assumption of risk neutrality is arguably unduly restrictive, at least from the perspective of rationality. Risk taking or risk loving behaviour can lead to preference structures that are well defined and well behaved and seem to be clearly admissible according to the basic criteria of rationality.
Calculate the expected utility: Consider a lottery with a good outcome and a bad outcome being present in some proportion (for example each outcome occurs with a 50 percent chance).In the diagram below note that the straight line is the linear average of the two levels of utility Ub and Ug. The point “e” represents the expected value of the lottery. For example, if the bad outcome is worth $100 and the good outcome is worth $200, and the probability of each outcome is 0.5.
“e’ would be at (0.5)x($100) + (0.5)x($200) = $150.
There’s a difference between expected utility which is on the vertical axis and expected outcome which is on the horizontal axis. Depending on the expected utility each player places on an expected outcome that will define their risk preferences.
Explain the different curves on the graph and how it represents risk averse, risk neutral, and risk seeking
Risk neutrality is constant linear because: expected utility always equals expected value – the bad outcome equals 100 to them the expected outcome equals 150 and the good outcome equals 200.
Risk averse are curved above (concave) the risk neutral linear line: the utility of getting a the expected value for certain is greater than the utility of a gamble. The space between going from the bad to expected outcome is equal to their utility but once you cross the expected outcome their utility diminishes as that space decreases showing that they prefer the guaranteed outcome. So past the expected outcome their expected utility is less than the expected outcome (going from 150 -🡪 200)
Risk seeking: Convex utility functions, the expected utility of the gamble is greater than the utility of getting the expected value of the gamble for certain. At the expected value of 150, the horizontal axis only moves by 50 (to good outcome) but the rate of expected utility increases much more drastically on the horizontal axis once the expected outcome is passed.
How much is each person willing to pay for the lottery ticket?
For example, a risk-averse person would not be willing to pay $150 to play the lottery, they would only be willing to pay some lesser amount, such as $120. The range could be quite large though. Extremely risk-averse people might be only willing to pay $101 (since the worst they can do with the lottery is $100), while slightly risk-averse people might be willing to pay as much as $149. The lower the value of the lottery to the person, the more risk averse they are. Hence there are different measures of risk aversion.
Under the assumption of risk-neutrality the utility of the expected outcome is equal to the expected utility of the outcome. In the above case the risk-neutral person would be willing to pay exactly $150 to play the lottery with an expected outcome of $150.
Risk-taking or risk-loving people would value the lottery more than the expected outcome of $150. A risk loving person might be happy to pay $175 to play the lottery simply because they want the chance to get the $200. Of course, a person would be considered foolish if he or she agreed to pay more than $200 for the chance to play the lottery, since $200 is the best possible outcome of the lottery. The more a person would be willing to pay to play a lottery, the more risk loving that person is.
Risk and conflict:
Let two sides, A and B, be symmetric so that they both have equal chances of winning a conflict over something worth 120 units. The benefit of winning is 100 and the benefit of losing is 0. The ex ante expected value of conflict is 50 to each player. The cost of war is assumed to be 20, so that if the two sides can agree on a division total resources would amount to 120. Under full information and risk neutrality there is clearly a contact space that would give player A an amount a where 50 < a < 70 and where B would get 120-a, i.e. it would also range between a low of 50 and a high of 70.
Now let player A be risk loving, specifically A values the lottery of 0.5(0) + 0.5(100) at 75. In other words, the chance of getting a fifty percent chance of getting 100 (and a fifty percent chance at nothing) is worth a lot to player A
Would war occur why or why not?
In this case risk neutral B would be unable to offer A a division of the 120 that could prevent war. The most B would be prepared to offer would be 70, keeping 50 for itself. But the 70 is insufficient to compensate A for the chance to win all 100.
War would now be a rational choice, and there is no settlement range. Why? By assuming risk loving behaviour the model has effectively “created” utility out of war by making it probabilistic and by having one (or both) players like games of chance.
Risk and Conflict notes:
- Having one or both players being risk loving does not guarantee conflict, why?
- Even if both sides are risk-seeking, war could in theory be avoided with a lottery, but why is this not credible?
- In conclusion, what can we say about predicting war through this model?
- For example, let two players fight over 120, with 20 being destroyed by war and each having a 50 percent chance of getting nothing or getting the remaining 100. The expected value of the “lottery” of war is 50. If both are somewhat risk loving and value the lottery at 55, which exceeds the expected value of 50, then there will not be war, because a peaceful arrangement of 60 each would be preferred to war. Two players that are risk loving may prefer peace to war if the cost of war is relatively high and their preference for risk is not that strong.
- Even if both players are risk loving to the point of stupidity (for example they both value the above war lottery. The game replicates the lottery, but there is 20 left over to divide between the two or to add to the lottery as a prize. In other words, the fact that war imposes a cost does not mean that war is inevitable when there are risk loving players. So, it is not really risk-loving behaviour at all that can lead to war if full contracting is available. This solution to war with risk-loving behaviour could be impeded, however, because of a commitment problem. For example, a country realizing that it had a 75% chance of losing a certain portion of its territory if it had a war with its neighbour could offer a non-destructive lottery that would leave the opponent happier than having a war. However, once the lottery outcome was realized, neither side would likely voluntarily carry out the transfer or give up on the idea of seizing the territory though a later war.
The real problem is either the fact that a lottery is not as much “fun” as war to some people (which would put us back into the realm of irrational sociopathic behaviour) or because it is too difficult to create and enforce the terms of a compensatory lottery. These would be two distinct, but interesting, approaches to understanding risky behaviour and conflict. That means that it is not risk-seeking behaviour alone, but risk- seeking with an inability to credibly commit to the peaceful lottery outcome, so a commitment problem
3: The more risk averse a player is, or the less risk loving a player is, or the higher the cost of war, the less likely conflict will occur. The interaction between these factors determines whether conflict is “rational” or not.
Boundary solutions and conflict: It is possible to think of
some very simple examples at the frontier between extinction and survival. If a person (or community) is slightly above the level of consumption needed for survival, they may act in extremely risk- averse ways to avoid the chance of falling below the survival level.
Solve the following: which will they choose, and what would it seem like in terms of risk?
1) If the minimum necessary ‘income’ for continued survival is 100 and the choice for the community is a guarantee of 105, or a lottery that provides a 50% chance of getting 98 and a 50% chance at getting 200
2) Similarly, if the community needing 100 to survive has a choice between a sure outcome of 95, and a lottery that yields 20 with probability 50 percent and 110 with probability of 50 percent,
1) the community may well choose the 105 even though the expected value of the risky option is 149. Indeed, the 105 could shrink to100 and the highest outcome of the lottery rise to 1000 and the community may still choose the sure thing because it guarantees survival. Essentially it is because the outcome of extinction is so bad that its value is effectively negative infinity.
This would seem like risk-averse behaviour
2) the community might be prepared to gamble to get over the minimum threshold for survival. In this lottery the expected value is only 65, but it at least affords some chance of survival. In this case being on the wrong side of the ‘extinction’ level might lead to apparently very risk loving behaviour.
Boundary solutions: Comparasion to indivisibility, thresholds, and indication of risk
At a community level it is identical to the indivisibility problem: the community survives or it does not. That seems arbitrary, since if a community fell 10% below its required subsistence level then presumably 90% of the community would be able to survive. The hard threshold for survival may apply better at the individual level, or in the extreme where an entire community is indeed threatened.
I can have actions that look risk-loving or risk-averse but it doesn’t have much to do with risk but more so on the threshold of survival.
Suicide attacks and rationality: Consequently, Ferrero develops a model where the individual is self-interested, assigning no value to others, a group, or to rewards received after death. The model is simple in concept. Martyrdom is a full information contract offering clear benefits in a first period and probabilistic benefits in the second. The individual is utility maximizing and has a constant degree of risk aversion. The model starts by offering an individual benefits B1 in period 1 to be a recruit. In period 2 the recruit must then face a lottery where she might get picked for a suicide mission with probability P, in which case her received benefit is 0, or she might not, in which case with probability 1-P she gets benefits in period 2 of B2. Of course, if picked for a suicide mission she can then refuse to accept to become a martyr, in which case she faces a sanction of S. Finally, if she does not join the group and sign the martyrdom contract she gets benefits U1 and U2 in periods 1 and 2 respectively.
Looking at the extensive form game,
1) name the policy options available to the terrorist group
2) name the policy options available to the government
1) In what case would the recruit become a terrorist in period 2: As long as the sanction cost S is worse than 0, the person will fulfill the martyrdom contract. In effect the sanction for reneging has to be so high as to be worse than death. The requirement for the sanction is crucial, and essentially amounts to an effective enforcement policy. This is like punishing the family of someone who reneges making S worse than death.
In what case can the organization recruit more people: It is possible to present conditions under which everyone agrees to become a martyr (B1 big, S really bad)
Making joining better than the status quo (U1, U2): Terrorist groups can provide benefits to the families of martyrs (though state authorities can impose punishments on remaining family members), which increases B1 and B2.
Is the need for the terrorist group to make the status quo seem less desirable so that the martyrdom contract, by contrast, looks better. So U1 and U2 (not joining) can’t seem like a good option, the organization can bomb work places and reduce unemployment to achieve this.
Lowering the probability of being selecting by increasing the number of people in the group so that members being recruited will calculate the benefits of B1 and B2 with a decreased risk of having to carry out the attack
What you should do to not be selected: make yourself more valuable so the organization needs you for other things
What you want as a group is to reduce the amount of people who just want to eat up B1 and B2 and are not committed to the cause as those people have a higher chance of reneging. Rather you want people who are more committed and will obey. So there’s a trade-off of increasing B1 and B2 because you will attract those who just want to benefit.
In which case would less people carry out the attack: If S is easy and not worse than 0
Here are other policy implications, such as whether to make it easy to renege on such a martyrdom contact. For example, the state could offer rewards, witness protection, and assist recruits chosen for a suicide mission to escape.
Similarly, the and the state to the state can use policies to improve normal life for the population from which recruits are drawn (increase U1, U2).
Reduce the benefits B1 and B2 by attacking the group’s resources
Making obeying the attack worse than death by arresting their family members and so on
2)
Suicide attacks:
1) a critique
2) organizations as club goods versus hardening targets as Berman puts it
3) economic contributions to suicide bombings and terrorist mobilization
1) The first obvious big objection to the model is assigning a value of 0 to death by carrying out the 11 suicide mission. This amount is arbitrary. Maybe it is – infinity? The real issue is whether there is an S that is so bad that it will discourage reneging.
2) They have a similar framework to Ferrero (also found in papers by Iannaccone): a (religious) group provides benefits (in this case public goods or club goods) to a group of distinguishable members. In the model potential group members have outside opportunities to gain utility as well as utility from the group in the form of the club good.
First, they argue that the marginal benefit of making targets harder is likely to be low. Indeed, the harder targets are in some sense more desirable targets because they also likely reduce the chance that an operative from the group will be captured (posing a threat to the organization. Most suicide attacks are in the original borders of Israel where targets are identified as “harder” and riskier to attack. By contrast targets outside the 1949 borders are characterized as “softer”, being less well protected, often with other Palestinians as targets, and thus more likely to be conducted using other attack methods rather than suicide bombers.
Therefore, more resources should be used to provide club goods that compete with and undermine the provision of goods by the terrorist group. Similarly, the restriction of the amount of club goods the group can provide to its members (reduce the group’s resources) will also be more effective than target-hardening. Improving outside options (better wage opportunities, for example) will also increase defection rates and lower the level of difficulty of operations that groups can undertake.
3) Economic models suggest that suicide bombing is still affected by economic factors as a whole. •Education, age, experience all influence effectiveness of suicide bombers •When unemployment is high and groups can recruit more highly educated members, they are more effective as suicide bombers.
Political bias: If war shifts some benefit to a key player (income, resources, prestige, power, etc.), then there might be an internal preference for war. The standard assumption of a unitary actor as the state is relaxed. They present their self-interested leader as pursuing personal profit, though there are also models where the prize is votes.
Explain the PPF, what is lost for the population, and leader, what can be negotiated.
In peace, the two countries operate on PPFP at point P. Of B’s share, its leader takes BLP as her cut of the country’s income. With war, the PPF curve shifts in to PPFW due to war losses and damage, and the two sides operate at point W with lower levels of income.
First, this is exactly the case Fearon assumes away in the “self-interested” leader case. Second, essentially the problem becomes an internal one that mirrors the exchange problem we have presented for two states.
BTp – BTw represents the amount lost to the general population as a result of war. The share of income to the general population goes down even further than that because it includes the cost of war + the gains from war for the leader. BLw – BLp represents the amount gained to the leader as a result of war.
In this case the leader gets more income after going to war. But the general population is getting much less.
Fearon would say however, that the population and leader should be able to negotiate a settlement without the cost of war. So the population would have to offer some point above what the leader would get from going to war and would take the rest as seen above in the graph.
1) What’s wrong with the case where the population would gain more from war, and the leader would gain from war, so the population would urge the leader to go to war?
2) What makes a leader gain more from war than the population?
1) The problem is that it does not address why that side could have negotiated to that point with the other side without going to war – so it only focuses on that one side. Why can’t the two sides (in this case within the same country) agree on a division that is clearly feasible and Pareto superior?
2) It doesn’t have to be that they gain more money/resources, but it could just be that they are seen as a good leader, gain more votes, and stay in power.
