Prolonged conflict exacts a devastating toll on families, communities, and entire populations—a toll that, at least theoretically, we “thinking animals” should shun. That’s because, in strictly rational terms, war is supremely inefficient.
If both sides of a conflict were on precisely the same page about each side’s likelihood of winning, as well as how much waging a war would cost them—economically or politically, for instance—they logically ought to be able to skip straight to a successful resolution. It would be better for everyone to divide contested resources in a way that is commensurate to those relative strengths and costs than to fight.
“With perfect information, we could reach a deal,” explains Sandeep Baliga, a professor of managerial economics and decision sciences at Kellogg. “Let me just give you exactly the same payoff you would get after a war anyway, or maybe a little bit more than that, and just buy you off.”
So then, using this logical framework, why do wars—and especially long wars—happen at all?
One popular theory among game theorists is that long wars are caused by asymmetric information, with one side knowing its own strength, for instance, but the other side remaining in the dark. In this situation, war breaks out because the side with better information has an incentive to pretend to be stronger than it is (aka, to bluff) to win a better deal, while the side with poorer information is reluctant to take anything the other side says at face value (knowing, correctly, that they might be bluffing). This can cause negotiations to break down.
Still, not all game theorists are convinced that asymmetric information can fully explain why long wars happen—for reasons we’ll describe below.
But a new paper by Baliga and his colleague Tomas Sjostrom of Rutgers University confirms that asymmetric information can indeed cause wars to endure for long periods of time. It also looks at the effectiveness of various actions that third parties can take to bolster one side of a conflict—with implications that could be relevant for NATO’s support of Ukraine.
In the end, wars endure when the incentives to feign strength are strong. “Anything that increases the bluffing value makes war longer,” says Baliga.
The Coase conjecture
To understand the disagreement among economists about asymmetric information, we first need to understand what’s known as the “Coase conjecture,” named for the late economist Ronald Coase.
Consider a civil conflict involving a government and a rebel group. In this scenario, the government’s strength is well-known to all parties, while the rebel group’s strength is known only to itself—a classic example of asymmetric information.
The government’s optimal move would be to offer the rebel group a single deal, after which there will be no further deals for a long time; if the rebel group rejects the deal, the two sides begin to fight.
If the government suspects the rebel group is strong, it will offer it a generous deal to prevent a war, which would be very costly (and if the rebel group is strong enough, could possibly lead to the government’s collapse).
But if the government suspects the rebel group is weak, the above strategy doesn’t make a lot of sense. So instead, the government will offer a stingy deal. If the government is right, the rebels will be inclined to accept the deal anyway to avoid a long battle that could cause their group’s demise.
Either way, the government stands a good chance of getting what it wants.
Yet according to the Coase conjecture, there’s a problem with this line of reasoning. It’s unlikely that the government’s “take it or leave it” offer will be perceived as credible. That’s because, if the rebel group turns down the government’s stingy offer and the battle begins, the government now has new information suggesting the rebel group may be stronger than it had assumed.
This will make it in the government’s interest to quickly come back to the negotiating table and offer a stronger deal. But the rebel group, regardless of its actual strength, will anticipate this possibility, which means it will be incentivized to hold out for an even stronger initial deal.
For the rebel group, the incentives to bluff and project strength “become really, really high because you can get a great deal almost instantaneously later,” says Baliga.
If the government cannot credibly convince the rebel group that it will stick by its “take it or leave it” offer, then it follows that regardless of the rebel group’s strength, it will get an acceptable offer from the government almost immediately.
As the government, “basically I would just give up and give you a big slice of the pie pretty quickly,” says Baliga.
Thus, the Coase conjecture holds that, should negotiation opportunities be plentiful, these conflicts spurred by asymmetric information should be speedily resolved.
Yet it is clear this does not always happen in the real world, where wars regularly endure for months or years on end.
This has led some international-relations scholars to argue that perhaps the failure to reach a deal is caused not by asymmetric information but by something else, such as one side becoming more powerful over time and the other side deciding to just fight them now when they are weak rather than conceding to them later when they are strong.
But Baliga and Sjostrom argue—using a novel and unexpectedly simple proof technique—that asymmetric information alone can explain the existence of long wars.
Why the Coase conjecture falls apart
In their proof, Baliga and Sjostrom demonstrate that the possibility that either party could collapse, as well as the fact that the rebel group’s probability of collapse is not known to the government, have a big impact on the negotiation process. This situation incentivizes a weak rebel group to want to fight rather than accept a poor deal, and it also incentivizes the government to fight (if it suspects the rebel group is weak) rather than make a generous deal even a strong rebel group would accept. These incentives upend the logic upon which the Coase conjecture rests and lead to a theory of long wars based on asymmetric information.
The key insight here is that negotiations do not work if one party or the other fears getting such a bad deal that they will regret making the deal in the first place. The classic example in an economic context is a used-car market where a buyer does not know if the seller’s car is a peach or a lemon and would regret paying a peach-worthy price for a lemon.
The same situation can arise in a conflict context: the government might prefer to fight the weak rebel group rather than give it the territory that would satisfy a strong rebel group. If the government believes the rebel group is likely to be weak (and thus highly likely to collapse), then even as a “thinking animal,” the government prefers to fight.
So war begins—and Baliga and Sjostrom show it must continue until all of the benefits that the weak rebel group would accrue by bluffing dissipate. (These benefits decrease over time as the chance of a weak rebel group’s collapse increases and reduces their payoff to bluffing.)
A third-party intervention
In their analysis, Baliga and Sjostrom also looked at what would happen if a third party intervened on behalf of the rebel group.
The third party has some options here. It can intervene by weakening the government or by aiding the rebel group. It can also act decisively or do its work along the margins. The challenge is that, given all of the incentives and countervailing incentives, it can be difficult to predict whether an action will do more harm than good.
For instance, providing moderate levels of support to prop up a weak rebel group could backfire if it made bluffing more attractive. Say a third party provides the group with advanced weapons that a strong rebel group (composed of trained mercenaries) could make good use of, but a weak rebel group (composed of volunteers) would struggle to use effectively. In this case, the incentives for a weak group to bluff and pretend to be strong (and thus capable of wielding the weapons) would skyrocket, prolonging the war.
Similarly, taking modest actions to weaken the government could help the rebel group—but as the rebel group’s relative position seemingly improves, the cost of offering the rebel group a “good deal” also increases, making it more advantageous to bluff, and making the government less inclined to pay off the rebel group. This could draw out the conflict even further, harming a weak rebel group.
In other words, Baliga and Sjostrom find, interventions by a well-meaning third party aren’t always helpful—and some only serve to lengthen wars.
Ukraine and Russia
Baliga and Sjostrom began work on this paper early in the pandemic, before Russia invaded Ukraine. But the similarities between that conflict and the one described mathematically in their paper are not lost on them. After all, asymmetric information about Ukraine’s strength played a major factor in Russia’s decision to invade, with Putin falsely anticipating a quick takeover.
So what lessons does their proof offer, particularly to groups like NATO or the U.S. as they decide how to intervene on Ukraine’s behalf?
Given the difficulty of gaming out the various support scenarios, the optimal thing that a third party can do is take very decisive action, their proof suggests, offering Ukraine what they describe as “overwhelming” support.
“They should intervene in such a way that the bluffing factor is incredibly small because of your intervention,” says Baliga. That is, by quickly and dramatically bolstering Ukraine’s strength—and making sure Russia knows about it—NATO and the U.S. could theoretically reduce the country’s need to bluff and hopefully shorten the conflict.
“What you’re doing is making it less relevant whether Ukraine was initially weak or strong, which is then more likely to end the conflict quickly.”