The Spoils of War: Greed, Power, and the Conflicts that Made our Greatest Presidents is about one particular principal agent problem. Through a number of case studies, the authors,  Bruce Bueno de Mesquita (one of the coauthors of the Dictator’s Handbook) and Alastair Smith, make the case that presidents (and congresses) will start wars that are advantageous to that president, or to their political faction, but are not in the interests of the nation as a whole.

To give one incentive-misalignment, the authors note in the introduction that presidents tend to be better remembered and respected by posterity if they preside over wars. In fact, there is a positive correlation between the number of americans (as a percentage of the population) who die in war during an administration, and how highly rated (by a historians) that president will be.

Making war is one of the best ways to attain personal glory, somewhat independently of whether the war is just, or even good for the country. 

Similarly wartime presidents are reelected at higher rates than peacetime presidents, and therefore presidents are often incentivized to start wars as a way to secure a second term. 

It seems there are a number of different incentive misalignments, of this sort. The main body of the book is a set of historical case studies outlining particular US wars, and how these were more advantageous for the sitting executive than they were for the public as a whole. I found these case studies variously compelling, some seemed very solid, some seemed more speculative. 

But in the conclusion of the book, the authors offer a number of policy suggestions that might improve (on the margin) the alignment of the president with the populous on the question of going to war. This is both an important problem in and of itself, and an example of an important class of problems. so in this post, I want to outline those proposals.

Electoral Reform 

The first suggestion is a very general solution, inline with the theory Bueno de Mesquita proposed in his earlier books. The larger the number of people from which the executive needs support, the more that executive's incentives are aligned with the population as a whole. So in general, if you want to avoid Presidents making war for the benefit of the few, at the expense of the many, you should set things up so that they need the support of the many to stay in power. 

In particular, Beuno de Mesquita and Smith advocate abolishing the electoral college and correcting / preventing gerrymandering, both of which make it possible for politicians to stay in power with support from a only a minority of their constituents.

Eli’s commentary: 

It seems like there are already plenty of reasons for electoral reform, and this is just one more specific benefit. 

Independent rating agencies for the costs of wars

More specific to this particular misalignment problem, the authors recommend creating independent rating agencies to estimate the costs of war, and publicize those estimates.

Presidents, like most people in the position of advocating for a project, tend to low-ball how expensive a war will be. It is much easier to sell congress and constituents on a war, when the forecasted costs are low. 

If there was a non-partisan agency that independently estimated the costs of a proposed war, this would provide a reality check on presidential estimates. If a trusted non-partisan institution forecasts that a war will be 3 times more expensive than the administration predicts it will be, that administration will look pretty stupid, especially if the non-partisan estimate proves correct once the war is underway.

And furthermore, such estimates could provide a benchmark for the performance of the president. Ideally, instead of being lauded (and re-elected) for fighting a war at all (as seems to be what generally happens), the president would be evaluated on the basis of how effectively he fought that war, compared to the estimate. A president who prosecutes a war under budget or faster than expected, is rightly to be praised as having done better than most would have done in his shoes. Whereas a president who spends much more money, lives, or time accomplishing his or her military objectives would be viewed as incompetent. 

This has the added benefit of incentivizing an administration to give accurate cost-estimates for a war. If a president knows that the estimates of the cost of war given when first proposing that war are going to be used to bench-mark his performance, he/she'll be less inclined to low-ball those estimates, because doing so will inevitably make him/her look incompetent down the line. This would force him/her to give honest cost-estimates when proposing a war.

There are two existing government agencies that could fill this role: the Congressional Budget Office, and in the Executive Branch, the Office of Management and Budget. Or these institutions could be outside of government, in academia, or as private non-partisan think tanks.

In fact, the CBO, did make pre-war cost estimates for the Iraq war (based on comparison to similar invasions and regime changes), that came in around $362 Billion and 10 years of occupation. This still undershot the true figure of $784 Billion, but was a good deal higher than the Executive branch’s estimate  of 100 to 200 billion. [Note all of these numbers are from the book, and I imagine how much the war cost is a contentious issue, when I quickly googled “cost of the Iraq war”, I got $2.4 trillion as the “long term price tag” of the war.]

The cost of war is not only monetary. But we could also set up institutions to forecast the loss of human life. The authors suggest a law that would require any administration that seeks a war to confidentially submit their military plans to a similar non-partisan body, which would then prepare estimates of the length and lethality of the proposed war, using the standard political science methods for forecasting wars.

In situations where time is of the essence, and the decision to go to war is too urgent to await analysis by a panel of experts, we might instead plug in the relevant variable into a pre-established algorithm, to quickly produce estimates of monetary and human cost. But in practice, wars are typically debated for at least a few months before they are declared.

Eli’s Commentary:

While I think this is a promising kind of idea, and I’m excited about policies like this one, I’m not clear on how this is supposed to work. I suspect that the issue here is not that there is no institution producing public estimates of the cost of war, and more that constituents don’t take those estimates into account. In which case, this is less of a proposal about what organizations should exist, and more a proposal about what the average voter should care about.

And if forecasts like the ones discussed here did become commonplace in politics, I imagine that there would be a proliferation of institutions, with partisan-alignment, all producing their own estimates, and that individuals (politically informed voters and politicians) would cite the estimates that supported their side best (just like we see with other difficult to asses questions like health-care and economic policy).

In principle, I suppose that it is possible for an institution to acquire a reputation for reliable forecasts, and therefore bear the mantle of non-partisan legitimacy. Are there any institutions that are trusted in that way today?

I think the idea of using an administration's own estimates for the cost of a war as a benchmark for the performance in executing that war is pretty interesting. But again, I don’t yet see what specific policy proposal that entails. In general, it seems like voters rapidly forget the proclamations and promises of politicians. So this ends up sounding like a claim that voters should care about and vote on the basis of performance-cost estimates. Which I agree with, but I don’t see how to make actionable.

Any suggestions of how to actually implement something like this?

Also, I note that this is an example of a general class of solutions that seem like they might improve politics: better metrics for assessing the performance of politicians. If there were a suite of reliable, measures of well being at different governmental levels (city, state, nation, etc.), and voters could be made to care about those measures, than politicians would be incentivized to optimize those measures. 

(Not perfectly incentivized, because of a commons problem: any given congress person individually has a small amount of control over the wellbeing of the country, and it may well be that it is in his electoral interests to insist on political grandstanding and blaming the other side, over the hard and uncertain work of making things actually better, but this would probably help on the margin.)

But again, this is in part a proposal for better metrics, but mostly a wish for voters to care about different things.

Require citizens to pay the cost of war

Finally, the authors recommend that the financial cost of war be borne by all citizens. 

Generally, the US has funded wars by taking on a deficit and by gradually increasing taxes. Bueno de Mesquita and Smith suggest that if funding for war was required to be paid by citizens (a opposed to by cutting other services) and clearly demarcated as such, so that individual citizens could feel the personal cost, wars would be much less popular.

Between 2001 and 2008, wealthy, probably Republican voters saw their post-tax income increase by many tens of thousands of dollars...Had instead they been asked to contribute, say, $40,000 to pay for the occupation of Iraq, they might well have regarded it as being the same unmitigated failure as Democrats did. However, this was not what happened. Bush transferred the cost of the war to the poor, who saw programs that benefited them slashed, and to future generations of taxpayers on the hook for the enormous cost of the war, much of which was kept off the books for appearance’s sake. The $40,000 burden is far from unrealistic. If we take the cost of the Iraq War as $1 trillion, and others estimate the real cost to be as much as three times higher, then that equates to a little over $3,000 for every person in the United States. Of course, not everyone pays an equal share of the tax burden. The top two income quintiles pay close to 90 percent of the total federal income tax burden, which readily translates into a wealthy family’s paying about $40,000.

They advocate, not just giving accurate estimates of the cost of a war, but also outlining a schedule for the funding of that cost via taxation, so that before a war is embarked on, each citizen knows how much the war will cost them, personally.

If war proposals came with a clear price tag, citizens might be a lot more enthusiastically opposed to them. And when wars drag on for years, they’re more likely to vote presidents out.

If “We, the people” believe that the issue at stake is sufficiently important that they are willing to pay their share, then congressmen can safely endorse conflict secure in the knowledge that their reelection prospects will be unharmed. If a president fails to end the war in the timely manner predicted, then the new taxes will have to be increased and “We, the people” will naturally be unhappy. This is bad for a sitting president, but as a consequence that president will fight to win the war as quickly and cheaply as possible. Further, presidents won’t push for war unless they are confident of achieving a rapid victory.

Eli’s commentary:

Again, I note that this particular policy is an example of a general class, ie make the specific expenses that are paid for by taxation legible to the voter, so that they can see what their tax dollars are buying. On general principles, that sounds good to me, but I’m not knowledgeable enough to guess with confidence whether a shift in that direction would be positive or negative on net.