In April, President Joe Biden announced that the US and its allies would withdraw from Afghanistan, after nearly 20 years of trying to impose a fractious peace. Since then, the number of NATO troops there has been steadily withdrawing: more than 90% of US forces had left by early July.
As American and allied troops have left, the Taliban has rapidly gained control over much of the country, especially in rural areas, and the Afghan government has struggled to respond. Many major cities are now encircled by Taliban forces; Taliban leaders say they control more than 85% of the country by area.
We wanted to try to forecast what would happen next. The biggest question is when – or whether – Kabul, where the Afghan government is based, will fall to the Taliban, as well as some subsidiary questions, such as whether the Pakistani air force will support Taliban forces, or whether the US will use air strikes to defend Kabul once it has withdrawn its ground forces.
The Fall Of Kabul
On the central question, the group was asked: When will the Afghan government lose control of at least half of the area of Kabul? They were given five options with the following pooled median estimates:
A (before September 2021): 6.5%
B (September - October 2021): 21.5%
C (November - December): 18.0%
D (January - May 2022): 28.5%
E (in June 2022 or later): 24.5%
The most likely outcome, the forecasters thought, was that the Afghan government forces would hang on until the end of the year; they also thought it fairly likely that they would survive beyond that into the second half of 2022.
Nonetheless there was considerable uncertainty: none of the options was a clear winner. A fall in autumn or winter this year was only slightly less likely than survival for longer.
That’s partly because good predictions require suitable base rates. That is, they need to be compared to a class of similar events in the past: if you are forecasting the likelihood of a home win in a football match, you can compare to that team’s existing record of winning games at home. But there are not many recent examples of occupying powers retreating from overseas territories.
That said, the forecasters found examples for comparison. One drew analogies with the Soviet withdrawal from Afghanistan in 1989; with the US withdrawal from Iraq in 2011; and with the US withdrawal from Vietnam in 1973.
“These historical examples strengthen the case for the Afghan government hanging on,” wrote one forecaster. “After the Soviets withdrew from Afghanistan in 1989, the Soviet supported government maintained control until the dissolution of the Soviet Union in 1991.” It collapsed rapidly then, in just four months, because once the USSR had dissolved, no more financial or military hardware support was available for the government. But while it had backing, it held on.
In Iraq, “the last US troops were withdrawn in December 2011.” The rise of the Islamic State in 2014 challenged the Iraqi government, but US air strikes helped defeat it, and by 2017 the Iraqi government declared victory.
And in Vietnam, “the fall of Saigon happened around two years after the US's withdrawal”. So these three recent examples suggest that foreign-backed governments can survive for several years after the withdrawal of supporting troops.
That said, there were also good reasons to think that the Afghan government’s defence will be weak. “The Afghan armed forces slightly outnumber the Taliban,” writes one. “But the Taliban have much greater morale, and as of now, momentum.” Another says that “the Afghan army is shockingly led, and will struggle with logistics and supply”, especially now that the borders are controlled by the Taliban. Another quotes an assessment of Afghan military strength which says that the government forces’ shortcomings include “poor leadership, high attrition and inability to effectively manage personnel, rampant corruption, and poor sustainment, maintenance, and logistics practices”.
And the collapse so far has been rapid. “The Taliban has expanded its territory extraordinarily quickly since the NATO forces started their withdrawal,” writes one forecaster. “Since April 13th, the number of Taliban controlled districts has increased from 77 to 222, or 19% of districts to 55%.” At that rate, within four months the Taliban would control 90% of districts – for comparison, it controlled just 75% in 2001, before the US invasion. On the other side of the equation, that forecaster points out, the districts that have fallen are mainly rural and not well defended: Kabul is “denser, urban, and harder to take, if the Afghan government is determined to defend it.”
US Air Strikes
One thing that will affect the likelihood of a rapid collapse is whether the US, having withdrawn ground troops, still offers support to the Afghan government in the form of air strikes, either to defend Kabul or elsewhere in the country.
One thought it was pretty likely: “In general, the US forces' threshold for conducting any airstrikes at all in various conflicts is fairly low, so I'm starting from a fairly high probability.” Another suggested that if Kabul falls while the US Embassy remains staffed, attacks are very probable, but it’s not clear how long staff will stay there. The median estimate for at least one US air strike attacking anywhere in Afghanistan, including Kabul, after the withdrawal was 62%; for it to happen in any case other than attacking Kabul, they put the likelihood at 50%.
Again, one problem was establishing a suitable base rate: one suggested using the air strikes in Iraq after the withdrawal as a reference, but added that in Iraq the US “never fully withdrew. They also had access to bases closer to the conflict zone, and could use aircraft carriers, both of which aren't available in the case.”
A related question was whether Pakistan would send its air force to aid the Taliban. Again there was significant uncertainty: one forecaster put it at greater than 60% likely, and one at less than 5%. The median was 36%. The forecaster with the highest estimate pointed out that “The Taliban has close historic ties to Pakistan, so I would expect Pakistan to join the current conflict, especially at the margins.”
Another thought that it was deeply unlikely: “The Taliban are a dangerous terrorist group, and although there are some in the Pakistani government who would prefer them to the current Afghani regime, this would cause big problems with the US. Would Pakistan want to risk losing the $680m they get in aid from the US, even temporarily, and cause a diplomatic incident with the rest of the world?”
One thought that while air strikes are fairly likely to actually happen, they will happen at the lawless borders of Pakistan and Afghanistan and clear, verifiable reports might not reach us, so we would never be able to resolve the question. That pushed their forecast below 50%.
On a related question, will Pakistan use ground troops to support the Taliban, the forecasters were more united: it’s fairly unlikely. “Why do when the Taliban are winning anyway, and when by doing so you'd antagonise the US for no real purpose?”, wrote one. “The Pakistanis are smart enough to realise this is a very difficult war to win, so won't get involved,” said another. Others thought it was not implausible – two forecasters put the likelihood at 40% – but thought more subtle support, such as ambiguous “military advisers” or intelligence agents, would be more likely. The median estimate was 21.5%.
UN Peacekeeping Forces
The group was also asked whether the UN will send any peacekeeping forces to Afghanistan before 2023. For that, unlike the other questions, there was significant consensus: it seems very unlikely. The median estimate was 9%, and the highest was 20%; one said just 2%, and the only reason they didn’t say 0% was “because of the stochastic nature of Afghanistan”.
“There have been six new peacekeeping missions since 2010, and four of them were in Africa, so the base rate for new peacekeeping missions is low,” wrote one. “I don't see the veto powers agreeing on this,” said another, “and no one wants to send UN troops into an actual hot shooting war.”
One wrote: “I really struggle to imagine a world where the UN sends troops to get killed by the victorious Taliban.”
A Peace Deal
Another way the conflict might end, rather than Kabul falling, would be a peace deal between the Taliban and the Afghan government. But since the Taliban seem to be winning, it’s not obvious why they would agree to one. “Not impossible, but not too likely,” wrote one. “Can anyone outline what a peace deal acceptable to both sides could really look like?” They put the likelihood at 30%, but that was mainly “for a scenario where a deal is publicly agreed to that everybody knows is not going to hold and/or is a de facto surrender.”
“Why the Taliban would bother to do this isn't clear to me,” wrote another, “but I could see it as a hypothetical thing they might bother with if they were perhaps slightly anxious about the US rejoining the fight.” The median estimate was 26%.