Is the Marginal Cost of a Drone Strike Too Low?

The New York Times hosted an interesting but somewhat unsatisfactory debate on Tuesday on one of the most difficult moral issues of the day: whether U.S. drone strikes in Pakistan and elsewhere do more harm than good. My main problem as a somewhat though not entirely uninformed consumer is that neither proponents nor critics accurately characterize the strategic trade-off.

The critics, James Jeffrey, Ibrahim Mothana, and James Cavallaro and Sarah Knuckey highlight the civilian cost of drone strikes but do not discuss what, if anything, drone strikes replace. By contrast, the advocates for drone strikes,  Bradley Strawser and political scientists Avery Plaw and Christine Fair, rely heavily on the argument that drone strikes are cheaper, more accurate, and kill fewer civilians than the alternatives: air strikes and intelligence operations. This misses the point: the availability of cheaper options makes the U.S. more likely to authorize assassinations than it otherwise would. If it is cheaper to assassinate, then lower value targets will more easily make the list than when assassinations are expensive. It is not a priori clear that having many more “precise” assassination attempts is welfare improving for the U.S. or the Pakistani than fewer attempts with greater collateral damage.

The key question is whether the marginal cost of a drone strike is too low? There are probably good examples where a drone strike took out a dangerous terrorist with minimal collateral damage. But is the U.S. now authorizing too many drone strikes because it has paid the fixed costs of an infrastructure that makes it possible? Much of the information that would guide a better judgment is confidential. Yet, as someone who believes that assassination attempts should be rare I worry about this a lot. Multiple parts of the military and government agencies have developed the ability to use drone strikes. Generally, when bureaucracies invest in such infrastructure they will find ways to use it; allowing the standards for individual justifications to slip.

Pretty much all authors in the forum call for more transparency and institutional safeguards. I agree and I believe the goal of such safeguards should be to raise the marginal cost of a drone strike. Yes, that reduces some of the agility one gains from this technology and perhaps some exceptions should be created for emergency situations. However, from what I understand, most drone strikes are carefully planned missions that follow targets for days if not weeks before striking.

ps: The full report by Cavallaro and Knuckey is here and contains the most detailed information about the impact in Pakistan that I have seen.

Update: Omar Bashir (a graduate student at Princeton) has an excellent piece on the Foreign Affairs web-site that explains just how a feasible oversight mechanism could work (drawing on the UK example).

4 Responses to Is the Marginal Cost of a Drone Strike Too Low?

  1. Ryan S September 27, 2012 at 5:22 pm #

    Let’s be careful throwing around economic terminology here. Given that there has been some fixed cost investment in a given technology, the question no longer even considers those sunk costs. While I agree having the technology makes it less costly to deploy those resources for a given goal, we can’t ignore the long-term costs in the form of externalities, associated with these strikes. The problem is not one of low marginal cost persay, but another failure to optimize over a long timespan in a world of externalities. Currently existing military actors are not incorporating the costs, in the form of blowback, increased resentment towards the US etc., which will be incurred by future military actors. Much the same as blowing today’s salary on that sweet new TV throws your future self under the bus, today’s military actors are creating a new host of trouble for future military leaders. If we were to incorporate both the short term costs of firing that rocket and the long term externalities, I don’t think we can say with such certainty that this is a low marginal cost vs. high marginal revenue situation.

    • Erik Voeten September 27, 2012 at 5:32 pm #

      Ryan: fixed costs are not the same as sunk costs. The fixed costs are the very large bureaucratic infrastructure investments (including 1000s of personnel) that keep drone operations going and that apply regardless of how many “units of assassination are produced,” to use some really ugly econspeak.

      • Ryan S. September 27, 2012 at 11:16 pm #

        This is true, not all fixed costs are sunk costs, but in this case the way I was interpreting the argument was that given a current stock of assassination methods what maximizes US interests relative to costs to the US. If the argument also takes into account the cost of building the aircrafts, and the thousands of personnel involved, then yes this is a case of fixed, and not, sunk costs. But the meat of my argument above still holds. We can’t say that the benefit of assassination is weighed against only the short term cost of missiles, aircraft and fuels, we have to optimize over a longer horizon, an economist would go for an infinite horizon, but I have a rule of excluding time horizons that can include our sun becoming a red giant.

  2. don December 27, 2012 at 12:20 pm #

    Perhaps one should just compare the unit costs. For example, 9/11 was estimated to cost half a million for almost four thousand dead and a trillion dollar hit to the US economy. A drone strike that takes out five combatants in Yemen goes for, what, a million dollars when all the inputs are calculated? A trillion dollars later and the US has little to show for its efforts in combating “man made natural disasters.” Seems like the disaster is spreading, quickly. Maybe a medical metaphor is more appropriate, as in runaway cell growth? I suspect this growth has more to do with a culture into distributing free condoms with an 18 percent failure rate rather than killing the enemy.