Editor’s note: this is a guest post by William Spaniel, a doctoral candidate at the University of Rochester. See this previous Duck post describing some of his work, and this post at his own blog providing more information about the research discussed here.
Spurred by a new International Organization article by Alexandre Debs and Nuno Monteiro (DM), the Duck of Minerva has recently hosted a debate on the cause of the Iraq War (see here, here, and here). To sum, DM argue that the United States’ imperfect knowledge of Iraq’s weapons programs led to a rational war. In contrast, speaking strictly from a unitary actor standpoint, this post argues that imperfect information cannot explain the conflict. Just prior to the start of the fighting, Saddam took credible steps to reopen negotiations with the United States. The Bush administration outright ignored these efforts. War began soon thereafter.
I divide this post into three parts, based on work from my dissertation. First, I review DM’s main theoretical contribution. Second, I briefly run down the logic of nuclear negotiations. And third, I argue that weapons inspections increase the cost burden on a potential proliferator, which in turn makes nonproliferation commitments credible. I then trace this logic in the lead up to the Iraq War.
Imperfect Information as a Rationalist Explanation for War
To begin, I want to summarize the logic of DM’s model. Their contribution is quite clever, and they undersell themselves in their previous posts.
Consider an interaction between a rising state and a declining state. Each actor has two strategies. The declining state can either launch preventive war or not, while the rising state can either build weapons or not. Suppose, due to poor intelligence, the declining state cannot observe the rising state’s decision at the time of its move. Then the interaction is functionally the following strategic game:
To rule out uninteresting cases, suppose power shifts are worth the cost and the declining state would prevent if it knew the rising state built. Under such conditions, both states must randomize: the rising state sometimes builds and sometimes does not, while the declining state sometimes prevents and sometimes does not. As a result, all outcomes occur with positive probability, including the outcome in which the declining state prevents even though the rising state is not building. DM term this “accidental” preventive war and argue that the Iraq War fits this bill.
Before moving on, I want to make a brief note about the welfare of this mixing outcome. The declining state suffers here: it receives its war payoff in expectation. The rising state does not receive the entire surplus, however, since there is positive probability of wasted war costs and investment in weapons.
Bargaining over Proliferation
I now break from DM by discussing two theoretical points that affect the framing of their results. First, any discussion of bargaining in the shadow of preventive war must also discuss bargaining over the weapons program itself. This topic remains underexplored generally in the international relations literature.
Perhaps surprisingly, declining states can easily buy off rising states if they wish to. To understand why, note that even if the rising state proliferates, it still ultimately receives some division. Thus, if the declining state wants to stop the rising state from proliferating, it could just offer the rising state what it would expect to receive post-proliferation. The rising state cannot profitably build at this point, since it will not receive any extra benefits. Consequently, nonproliferation is credible given sufficiently large offers even if the declining state cannot observe the rising state’s building decisions.
Moreover, because the rising state saves on its investment cost, the declining state can offer slightly less than the rising state would ultimately receive and still induce acceptance. Thus, the declining state can steal the surplus created by the rising state not investing. If the declining state is sufficiently patient, this is preferable to taking as much as it can upfront and suffering the consequences of proliferation thereafter.
Problems only flare up if the remainder of the necessary bribe is less than the amount the declining state receives from war. (DM discuss this briefly in their infinite horizon extension but do not detail the full implications.) At this point, the declining state would rather fight than bribe. With perfect information, the declining state can use the threat of preventive war to deter the rising state from building. With imperfect information, however, the declining state remains stuck in the simultaneous move game from before. The declining state receives its war payoff, while the rising state receives an inefficient portion of the remainder.
Note that the cost of proliferation determines whether the declining state prefers bribing the rising state. If the cost of proliferation is large, the declining state only needs to offer the rising state a minimal bribe. But if the cost of proliferation is cheap, the declining state prefers its war payoff and is unwilling to bribe.
In between these extremes, a particular cost of proliferation makes the declining state indifferent between bribing the rising state and fighting. Thus, at that cost, the declining state is willing to take the efficient route. Here, the declining state still receives its war payoff. The rising state, meanwhile, receives the entire surplus, which is more than what it would receive if costs were lower. Thus, the rising state benefits from greater proliferation costs.
The figure above plots each state’s payoff by the cost of proliferation. When costs are low, the states find themselves in the inefficient mixing pattern, and both suffer. In the middle range, bribes succeed, and both states benefit. (The rising state fares increasingly worse as costs go increase, though.) When costs are great, the rising state cannot credibly threaten to proliferate and receives no concessions.
The Purpose of Weapons Inspections
Because the middle range Pareto dominates the lower range, states would unsurprisingly seek methods to increase the cost of proliferation if costs run too low. Weapons inspections serve this purpose. DM correctly argue that inspectors cannot provide perfect information about the rising state. However, they do not appreciate what weapons inspections can do—slow the proliferation process to a crawl. The most likely places to proliferate are also the cheapest places to do it. Weapons inspectors functionally take these locations off the table. Thus, if a rising state still wants to proliferate, it must pay a higher cost to do so. In turn, it becomes easier to bribe the rising state, as its commitment not to build becomes credible.
How does this relate to the Iraq War? Prior to February 2003, it appears that the DM logic was at work. Saddam took every opportunity to befuddle weapons inspectors and was not alleviating the Bush administration’s proliferation concerns. However, Saddam had reason to not be forthcoming. Iraq had repeatedly violated the terms of the Persian Gulf War treaty in the decade earlier. The U.S. had only responded meekly with occasional air strikes. As David Lake noted in his International Security article, Saddam also had to balance his relationship with Iran. Suffering the occasional bombing might have been worth maintaining the bluff toward Tehran.
By the middle of February, though, the U.S. had clarified its intentions. 100,000 American troops flooded Kuwait. This was the costly signal Saddam needed to separate the credible U.S. from the bluffer. At that point, the relationship with Iran became irrelevant—Saddam needed to appease the U.S. immediately or risk jail (or death).
So Saddam opened up. On the eve of the war, Hans Blix, head of the U.N. task force, felt that his “Iraqi counterparts had toward the end become almost frantic in submitting material, seeking and finding persons [they] could interview….Had there been denials of access? Any cat-and-mouse play? No. Had the inspections been going well? Yes” (pages 10-11 of Disarming Iraq).
Had the Iraq War been about imperfect information, this is where the United States would have looked to negotiate. However, the Bush administration ignored Saddam’s newfound attitude and pressed forward. War began the next month.
Conclusion
As a final note, this post does not claim that the Iraq War cannot be explained through rationalist approach. It is neutral in this regard. Rather, the goal here was to argue a negative, namely that imperfect information did not cause of the Iraq War. Consequently, more research needs to be done on the conflict—whether the focus is on rationalist unitary actor explanations, rational domestic politics explanations, or behavioralist explanations as Lake suggests.
We’d like to thank William Spaniel for engaging with our work. A few remarks:
First, we agree that exploring bargaining over a weapons program is a potentially fruitful pursuit, and indeed would encourage Spaniel and others to take it up. Our purpose in this paper was partly to open up the space for such work by conceptualizing military power shifts as endogenous. Viewed through this lens, a crucial feature of inspections is that they provide information about proliferation attempts, and states can respond to such information with a series of tools, from peaceful offers (and perhaps even concessions, if it appears that the potential proliferator is not proliferating) to preventive wars. In the paper, we took a first stab in this direction and encourage others to pursue this work further, perhaps capturing other important features of inspections, such as their effect on the cost of an investment in nuclear weapons, as Spaniel argues.
Second, we disagree with the main point of Spaniel’s post, which is that, contrary to our argument, imperfect information cannot account for the Iraq War. To begin with, let us clarify that Spaniel actually agrees with our general theoretical claim that imperfect information is a necessary condition for war between a potential proliferator and its enemy. In Spaniel’s words: “With perfect information, the declining state can use the threat of preventive war to deter the rising state from building.” Problems arise only when information is imperfect. The rising state may be building nuclear weapons and in anticipation the declining state may strike preventively. Per Spaniel’s theoretical argument (as laid out in this post), the Iraq war could not have occurred without imperfect information.
Then Spaniel goes on to assert that there was no imperfect information prior to the Iraq War. However, he provides no evidence that the United States possessed perfect information about Iraq’s WMD programs. The sole piece of evidence Spaniel presents in support of his argument is a comment that Hans Blix made in his (retrospective, apologetic) memoirs, saying that Saddam had been quite forthcoming in the last few weeks leading up to the war.
This is problematic. As we document in our paper, in his U.N. reports about the status of Iraq’s WMD programs, Blix was quite ambiguous right up to the beginning of the war. More generally, as Spaniel himself admits, “inspectors cannot provide perfect information about the rising state.” Washington did not have incontrovertible knowledge that Iraq was not investing in WMD at the time, and it could not be certain that it would catch future militarization attempts. Even if it knew that there were no WMD programs in Iraq in March 2003, U.S. concerns about detecting future militarization attempts (in the wake of 9/11) might be sufficient to cause war. The one-shot game Spaniel lays out in his post does not do justice to the richness of the strategic interaction between the United States and Iraq, nor to our treatment of it in our infinite-horizon game.
Spaniel is also incorrect in saying that we “do not appreciate what weapons inspections can do -– slow the proliferation process to a crawl.” In our framework, if inspections are sufficiently informative, then proliferation will be prevented because the threat of preventive war is credible. The rising state knows that it would be caught if it tries to acquire nuclear weapons, and does not attempt to proliferate. The declining state is confident that it can deter militarization attempts, and does not feel the need to strike based on ambiguous evidence. Peace prevails and there is no proliferation.
One key problem in the post-9/11 world was that U.S. decisionmakers doubted their ability to detect militarization attempts in a timely manner. With such severe information problems, and the inability of enemy states to commit to refrain from investing in nuclear weapons, war occurred against a state like Iraq, whose proliferation was perceived to have strong adverse effects, relative to the cost of preventive war.
Finally, a minor point: we do not call preventive wars against states that are in fact not investing in additional military capabilities “accidental” wars, as Spaniel writes in the post. Rather, we call them “mistaken;” the difference being that they do not happen by accident but are nonetheless mistaken in targeting that particular state.
Thanks again.
Alexandre Debs and Nuno P. Monteiro