I dislike our national obsession with anniversaries and tendency to convert solemn occasions into maudlin ones; to fetishize perceived collective victimization rather than simply recognizing real victims. That kept me from joining in the outpouring of September 11 reflection, now mercifully receding. But I have reflections on the reflections.

The anniversary commentary has, happily, included widespread consideration of the notion that we overreacted to the attacks and did al-Qaeda a favor by overestimating their power and making it easier for them to terrorize. Even the Wall Street Journal allowed some of the bigwigs they invited to answer their question of whether we overreacted to the attacks to say, “yes, sort of.”

Unsurprisingly, however, the Journal’s contributors, like almost every other commentator out there, did not define overreaction. It’s easy and correct to say we’ve wasted dollars and lives in response to September 11 but harder to answer the question of how much counterterrorism is too much. So this post explains how to do that, and then considers common objections to the answer.

That answer has to start with cost-benefit analysis. As I put it in my essay in Terrorizing Ourselves, a government overreaction to danger is a policy that fails cost-benefit analysis and thus does more harm than good. But when we speak of harm and good, we have to leave room for goods, like our sense of justice, that are harder to quantify.

Cost-benefit analysis of counterterrorism policies requires first knowing what a policy costs, then estimating how many people terrorists would kill absent that policy, which can involve historical and cross-national comparisons, and finally converting those costs and benefits into a common metric, usually money. Having done that analysis, you have a cost-per-life-saved-per-policy, which can be thought of as the value a policy assigns to a statistical life—the price we have decided to pay to save a life from the harm the policy aims to prevent.

Then you need to know if that price is too high. One way to do so, preferred by economists, is to compare the policy’s life value to the value that the target population uses in their life choices (insurance purchases, salary for hazardous work and so on). These days, in the United States, a standard range for the value of a statistical life is four to eleven million dollars. If a policy costs more per life saved than that, the market value of a statistical life, then the government could probably produce more longevity by changing or ending the policy. A related concept is risk-risk or health-health analysis, which says that at some cost, a policy will cost more lives than it saves by destroying wealth used for health care and other welfare-enhancing activities. One calculation of that cost, from 2000, is $15 million.

In a new book, Terror, Security, and Money: Balancing the Risks, Benefits, and Costs of Homeland Security, John Mueller and Mark Stewart use this approach to analyze U.S. counterterrorism’s cost effectiveness, generating a range of estimates for lives saved for various counterterrorism activities. I haven’t yet read the published book, but in articles that form its basis, they found that most counterterrorism policies, and overall homeland security spending, spend exponentially more per-life saved than what regulatory scholars consider cost effective.

That is a strong indication that we are overreacting to terrorism. It is not the end of the necessary analysis however, since it leaves open the possibility that counterterrorism has benefits beyond safety that justify its costs. More on that below.

Objections to this mode of analysis have four varieties. First, people have a visceral objection to valuing human life in dollars. But as I just tried to explain, policies themselves make such valuations, trading lives lost in one way for lives lost in another. So this objection amounts to an unconvincing plea to keep such tradeoffs secret and make policy in the dark.

Second, people challenge the benefit side of the ledger by arguing that terrorists are actually far more dangerous than the data says. Analysts say that weapons of mass destruction mean that future terrorists will kill far more than past ones. One response is that you should be suspicious anytime someone tells you that history is no guide to the present. It tends to be the best guide we have, for terrorism and everything else. Our analysis of terrorists’ danger should acknowledge that the last ten years included no mass terrorism, contrary to so many predictions. Another response is that one can, as Mueller and Stewart have, include high-end guesses of possible lives saved to show the upwards bounds of what counterterrorism must accomplish to make it worthwhile. The results tend to be so far-fetched that they demonstrate how excessive these policies are.

A third objection is to claim that some counterterrorism costs are actually terrorism’s costs. Government should spend heavily to avoid terrorism, this logic says, because our reaction to the attacks we would otherwise fail to prevent will cost far more. In other words, if an expensive overreaction is inevitable, it helps justify the seemingly excessive up-front cost of defenses.

One problem with this objection is that it approaches tautology by treating a policy’s cost as its own justification. See, for example, Atlantic writer Jeffrey Goldberg’s recent response to Mueller’s observation in the Los Angeles Times that more people die annually worldwide from bathtub drowning than terrorism and the article’s suggestion that we might therefore be overreacting to the latter. Goldberg argues, essentially, that we have to overreact to terrorism lest we overreact to terrorism. Then, after his colleague James Fallows points out the logical trouble, Goldberg, without admitting error, switches to argument two above, while failing to acknowledge, let alone respond to, Mueller’s several books and small library of articles shooting that argument down.

Another problem with the inevitable overreaction argument is that overreaction might happen only following rare, shocking occasions like September 11. Future attacks might be accepted without strong demand for more expensive defenses. Moreover, the defenses might not significantly contribute to preventing attacks and overreaction.

The best objection to Mueller and Stewart’s brand of analysis is to point out counterterrorism’s non-safety benefits. The claim here is that terrorism is not just a source of mortality or economic harm, like carcinogens or storms, but political coercion that offends our values and implicates government’s most traditional function. Defenses against human, political dangers provide deterrence and a sense of justice. These benefits may be impossible to quantify. These considerations may justify otherwise excessive counterterrorism costs.

I suspect that Mueller and Stewart would agree that this argument is right except for the last sentence. Its logic serves any policy said to combat terrorism, no matter how expansive and misguided. We may want to pay a premium for our senses of justice and security, but we need cost-benefit analysis to tell us how large that premium now is. Nor should we assume that policies justified by moral or psychological ends actually deliver the goods. Were it the case that our counterterrorism policies greatly reduced public fear and blunted terrorists’ political strategy, they might indeed be worthwhile. But something closer to the opposite appears to be true. Al-Qaeda wants overreaction—bragging of bankrupting the United States—and our counterterrorism policies seem as likely to cause alarm as to prevent it.