Post-9/11 Counterterrorism Failed a Basic Cost-Benefit Test

The counterterrorism response likely killed more people than it saved

Nineteen Islamist terrorists murdered 2,979 people on September 11, 2001. The government’s counterterrorism (CT) response to 9/11, which was the deadliest terrorist attack in world history, included domestic security like the TSA, the PATRIOT Act, surveillance, and numerous wars overseas that we lost. The costs of 9/11 were horrific, and the response was gargantuan. When studying public policy, the question we must ask is: Was the counterterrorism response to 9/11 worth it? Did it save enough lives to pay for itself or at least break even?

No, not even close. For the costs of CT spending to equal the benefits of the value of lives saved, it would have to have saved between 390,000 and 442,000 lives, or between 15,600 and 17,680 per year since 9/11. From September 12, 2001, through September 10, 2026, I estimate that 347 total people were murdered on US soil by all terrorists of all ideological motivations and locations of birth. That includes some victims in attacks since that analysis was published. Thus, all CT spending would have to have saved 1,124 to 1,274 times as many lives as were taken by terrorists in attacks on US soil for the benefits of CT spending to equal the costs in terms of lives saved. To put that in perspective, the number of lives that would have had to have been saved for CT just to break even, never mind producing a positive return, would have had to have been roughly equal to every murder committed on US soil since 9/11. That large number of lives was not saved by post-9/11 CT spending and foreign wars to stop terrorists.

This matters because scarce resources that funded CT could be used to save lives elsewhere, such as better road safety, improved policing, or the innumerable other ways that people make their lives safer when they’re richer. Much of the actual CT spending is worse than just the opportunity cost of diverted safety resources implies, because the actions that caused the spending directly or indirectly killed Americans. Around 7,000 Americans died in Iraq and Afghanistan, not including the hundreds of thousands of locals. The requirement to screen all checked baggage after 9/11 reduced flight passenger volume by about 6 percent, with some travelers opting for more dangerous road travel. Some of the marginal drivers were spooked by terrorism instead of the extra cost of security, and the authors cannot separate the two, so the extra cost of security explains some unmeasured share of the as many as 2,300 road deaths they attribute to the attacks.

Regardless of the high cost in lives of the US CT strategy after 9/11, which should make all Americans skeptical of the social utility of CT actions, let’s look at whether the spending and economic costs could plausibly have saved lives. The methods for estimating the costs herein include the fiscal costs and an economic estimate. They do not include the losses from diminished civil liberties, psychological costs, or state or local costs. The accounting of fiscal costs here follows the 2018 Stimson Center study’s definitions. It includes federal homeland security budgets, spending from the Department of Defense (DOD) emergency and overseas contingency operations (OCO), war-related Department of State and USAID, and spending on foreign CT operations. The OCO designation ended in FY2022 after the US withdrawal from Afghanistan. Thus, the FY2018–FY2026 flows are estimated from DOD enacted appropriations and the BBA 2019 targets and the State Department and USAID component comes from the $8 billion annually designated as OCO from FY2019. OMB’s government-wide homeland security report, the series Stimson used, was discontinued after FY2017, so the homeland security component for FY2018 onward scales the DHS gross discretionary budget authority to the last government-wide total. Nominal flows are deflated to constant July 2026 dollars using monthly CPI-U index levels from the BLS historical series and summed. Interest on war borrowing, war-attributable growth in the DOD base budget, and veterans’ medical and disability obligations are excluded merely because that would require more steps to avoid double-counting than a back-of-the-envelope calculation could justify.

Controversially, this analysis assumes that real GDP in 2026 is 0.1 percent below the counterfactual level of no increase in CT spending after 9/11, with the gap widening linearly from zero in FY2002. Because a permanent level shift generates a flow loss in every period, the annual output gaps are summed and deflated to the same base. Truncating the output-gap sum at FY2026 treats a permanent level effect as terminating, so the output term is a lower bound. It is unsound to think that wars, immigration restrictions, and the resulting regime uncertainty didn’t lower GDP by a conservative 0.1 percent.

The last and most important step is estimating the value of a statistical human life to compare with the cost of CT spending. This is an emotional and methodologically fraught way to measure human life. As a father and a husband, I understand this emotional reaction well. Still, the fact remains that if the government spends more to save a life through enhanced CT than it is worth, then that means that other people died because of neglected safety in other areas. As a necessary preamble, it’s necessary to state that the value of a human life is not infinite. You don’t even think that about your own life because, if you did, you wouldn’t do anything to increase the chance of dying, such as drive in an automobile when it’s not a life-or-death situation, because that increases your chance of dying. My method of avoiding this problem is to pick a high number, even though an error in that direction is just as dire as an underestimate.

As a hypothetical, suppose the value of a statistical life is $15 million. If the government spends $30 million to save one life, then at least one person died who otherwise would not have died if that money had been spent where it would save more lives. Thus, spending that amount of money to reduce risk results in more deaths than otherwise would have occurred. Although emotional and hard to calculate, estimating the statistical value of life can help policymakers save more lives. The death of human beings is the highest and most significant cost of terrorism, but not the only one, as other forms of destruction are also costly but relatively minor compared to death. For simplicity, I will focus on the cost in terms of human life and assume a $15 million value. You can substitute your own statistical value of life in the table below to recalculate.

Table 1 shows low- and high-end fiscal and economic cost estimates from CT policies after 9/11. The federal government’s CT costs are spread across many agencies, programs, and decades, so they must be estimated to some extent. Most economists who read this will think the $15 million value of statistical life is too high, and they would balk at the inefficiencies that would result from informing policy on that basis. However, non-economists will probably think that’s too low. Picking an outrageous number like $100 million per life doesn’t save the analysis, though, because that merely means CT spending would have to save 58,500 to 66,300 people to break even.

Table 1 doesn’t quite capture how disproportionate the costs of CT were compared to the possible benefits. Figure 1 shows how much higher the number of people killed in terrorist attacks after 9/11 would have had to be, relative to the number of people actually killed, to justify the level of CT spending. It’s calculated by multiplying the number of people murdered in such attacks by 1,124 each year, the minimum ratio of the deaths required to break even with CT costs to the actual number of deaths, spread proportionally over each year by the number of real victims. The number of actual murdered people in terrorist attacks is a flat line by comparison to this counterfactual that would justify the spending.

My colleague John Mueller and his coauthor, Mark G. Stewart, answered this question decisively over a decade ago, and I contributed modestly to answering it over the years. The cost numbers above are merely an update. Mueller and Stewart also looked for the attacks that CT spending averted and found almost none. My estimate of 347 people killed in terrorist attacks after 9/11 is likely high because it includes some victims of attacks that were likely not terrorists, out of an abundance of caution that is prejudiced in favor of security hawks, a bias that should be familiar to anyone who watched how CT money was spent after 9/11.

The 9/11 terrorist attacks were a crime of shocking proportions that deserved a proportionate response, but the costs of those attacks were multiplied by the government’s grossly disproportionate response. None of that matters to the people who were killed or their loved ones, nor should it. Some CT spending, increased law enforcement actions, and perhaps even military actions were justified after the attacks, but what the government actually did was wasteful and destructive to a degree that’s difficult to comprehend. There are several lessons from 9/11, but the biggest one is “don’t overreact.”