Afghanistan Trade-offs

First, a bit of administration: I’ve been posting some blogs at outsideview.io, a new group project focused loosely on longtermist interests, although I’ve also been writing less due to demands on my time. If that project pans out, the number of posts here may be slightly lower.

The U.S. is withdrawing from Afghanistan this year. I’ve long been a critic of the war so this should be a positive development. There is, however, a consequentialist argument which seems somewhat convincing. It’s articulated by The Economist here, and it goes something like this: the Taliban is pretty horrible and as the U.S. withdraws, the Taliban is taking over the country again, placing 40 million people under a despotic extremist regime. Additionally, not many U.S. forces are required to keep the status quo going, and casualties are at a minimum in the last few years (fewer than 20 U.S. casualties per year), while political pressure to end the war is diminished. Thus, the U.S. could presumably keep the status quo going and protect Afghanistan from the Taliban indefinitely.

The ultimate cost-benefit may turn out in favor of staying, but I think The Economist misses some clear costs to the conflict. The most important is that Afghanistan’s status quo for the past five years has not been rosy; it’s been one of the bloodiest conflicts on the planet. Afghan security forces have suffered an average of almost 10,000 casualties a year for the last five years according to this Brookings Institute report (page 15). There have also been over 1300 civilian deaths every year for the past five years according to the UN Assistance Mission in Afghanistan. On a per capita basis, these combined figures would be equivalent to ~98,000 deaths a year in a country the size of the U.S. The implications of this discussion have some strange results.

It’s extremely concerning that millions of people would be left under the control of the Taliban, but I suspect that a total collapse of the Afghan government would result in fewer fatalities. The Economist notes that already security forces in Afghanistan are abandoning their posts. The government’s collapse by necessity would remove one of the main belligerents to the conflict. Of course, maybe these are bad assumptions! Maybe the Taliban won’t take over and the conflict will drag on. But that also means the argument that 40 million people will suddenly be under an extremist regime is no longer valid. Or perhaps there is a worst-case scenario where the Taliban controls most cities (meaning most of the population is under their control) and the conflict remains at its current intense levels.

If we take this belief that the U.S. withdrawal will lead to a collapse of the government, we can construct an extremely simple model for a policy trade-off. The U.S. can either stay in Afghanistan and allow the continued bloodshed at 11,000+ deaths a year, or the Biden administration can withdraw, allowing the Taliban to take over. The consequentialist calculation the administration faces is to determine at what level of fatalities would it be worth it to stay or leave. How many lives is it worth to keep millions out of reach of an authoritarian regime?

This is an extraordinary choice and it feels gross to think about, but my own misgivings about discussing it doesn’t change the fact that it’s largely up to the U.S. government to decide which to sacrifice.

The position that seems most high status is that any struggle against a totalitarian regime is a good fight. But we really need a more carefully calculated answer to make a good decision. I suppose it’s also worth asking if the status quo is sustainable; will Afghan security forces stick around indefinitely if they are taking this level of casualties year after year? It doesn’t seem realistic to think casualties will go down either. Empirically speaking, we’ve now tried 20 years on attempting to outlast the Taliban and the result is that “we keep the Taliban out of power, but 1000 civilians die each year and 10x as many security forces”.

Other Costs

It’s also worth mentioning (where The Economist didn’t) the high fiscal cost; in 2018 the Pentagon estimated that yearly costs were around $45 billion. That’s larger than the 2020 NIH budget of $42 billion. There are probably more egregious misuses of government money than spending ~$1200 per capita to keep millions of people out of reach of a totalitarian regime. And there are also likely more efficient uses of money as well, but budget items each have strong political support, and it doesn’t make much sense to talk about shifting budgets around arbitrarily as that isn’t a political reality. Thus, I don’t see a strong fiscal argument either way.

Finally, there is a hard to measure legal and constitutional question. Congress authorized the invasion of Afghanistan with the 2001 Authorization for Use of Military Force (AUMF). Since then, presidents have used it as a legal basis for a wide range of interventions that Congress in 2001 clearly had no way to foresee. Would Congress have supported these interventions? Certainly sometimes. But it seems dangerous to allow presidents to have sole discretion, especially given broad concerns over presidential authority in the last few years. I think a world with a more powerful Congress relative to the president is better in many ways, but Congress has shirked responsibility perhaps due to structural political issues. Given this situation, it could make a lot of sense for a president to try and push for Congressional action by arguing that further operations in Afghanistan require a new authorization of force. However, this would take away from a key argument The Economist put forward, which is that the small U.S. contingent remaining in Afghanistan has largely escaped political scrutiny; my proposal here would thrust it back into debate, where I suspect it would do poorly.

Moreover, political capital is presumably much more limited than federal funding, so spending energy campaigning for a new AUMF would have a massive opportunity cost, reducing the ability of the administration to pass other reforms. So we’re left with either remaining in Afghanistan without real Congressional backing, leaving open the risk of presidential overreach in the future (and the previously mentioned deteriorating conflict over the last few years), or trying to withdraw and solve the legal issue that way. There may not be good options here, but this is another benefit of withdrawal that I think most Afghanistan hawks ignore. Since it is hard to place a value on these constitutional benefits though, there’s a lot of room to wrangle over how much this factor should be weighed.

None of this is to say that advocates of staying in Afghanistan are wrong, but I think the arguments offered often avoid the hard trade-offs.

Predictions

One thing that might help clarify some of the trade-offs here would be predictions. I’ve compiled a short list of questions I would like answered:

  1. Will the current Afghani government remain in control of Kabul by the end of 2021?
  2. …what about July 1 2022?
  3. Will civilian deaths be worse than in any year since 2016?
  4. Will fatalities among Afghan Security Forces in 2021 be worse than any year since 2016?
  5. Will fatalities among Afghan Security Forces in 2021 be 25% worse than any year since 2016?
  6. If the government is toppled by the Taliban in the next 12 months, will there be widespread human rights violations by 12 months after that?
  7. If the government is toppled by the Taliban in the next 12 months, will there be a >5000 war or regime-caused deaths in the 12 months after that?

My current predictions, which honestly I have not spent a ton of time on are:

  1. Yes, 70%
  2. Yes, 60%
  3. Yes, 60%
  4. Yes, 60%
  5. No, 70%
  6. Yes, 70%
  7. No, 60%

I’ll attempt to track these on the blog here over time.