Wednesday 2 September 2020

Covid and the counterfactual, and the longer term

Counterfactuals are always tricky: what would have happened but for the policy change you're trying to evaluate?

With Covid it's especially tricky because, obviously, when things look riskier out there people will adjust their behaviour even in the absence of policy. They'll avoid places that look particularly risky, they'll be more likely to work from home, they'll avoid public transport if they can. Or, at least, the risk-averse will. The risk-preferring won't, along with the deluded, and the uncoordinated efforts of everyone else then might get you to a R-naught of 1 rather than an R-naught of less than one. So you get a lot of costs of activities not undertaken, but without it really being enough to knock the thing out. 

How then to evaluate the costs of policies that make some of those risk-avoiding behaviours mandatory? You can't use pre-Covid as counterfactual because that doesn't exist any more. You'll conflate the costs of the outbreak with the costs of the policy, and the two will largely coincide.

Austin Goolsbee and Chad Syverson had a crack at it in a June NBER working paper.

Here's the abstract:

Fear, Lockdown, and Diversion: Comparing Drivers of Pandemic Economic Decline 2020

NBER Working Paper No. 27432. Issued in June 2020.

The collapse of economic activity in 2020 from COVID-19 has been immense. An important question is how much of that resulted from government restrictions on activity versus people voluntarily choosing to stay home to avoid infection. This paper examines the drivers of the collapse using cellular phone records data on customer visits to more than 2.25 million individual businesses across 110 different industries. Comparing consumer behavior within the same commuting zones but across boundaries with different policy regimes suggests that legal shutdown orders account for only a modest share of the decline of economic activity (and that having county-level policy data is significantly more accurate than state-level data). While overall consumer traffic fell by 60 percentage points, legal restrictions explain only 7 of that. Individual choices were far more important and seem tied to fears of infection. Traffic started dropping before the legal orders were in place; was highly tied to the number of COVID deaths in the county; and showed a clear shift by consumers away from larger/busier stores toward smaller/less busy ones in the same industry. States repealing their shutdown orders saw identically modest recoveries--symmetric going down and coming back. The shutdown orders did, however, have significantly reallocate consumer activity away from “nonessential” to “essential” businesses and from restaurants and bars toward groceries and other food sellers.

Policies coordinating anti-covid activities in the US haven't been particularly successful - if one county managed to stamp it out, it would quickly come back through travel. 

Absent policy measures here, we'd very likely have had the same outbreaks seen abroad, very likely with the collapse of the health system which absolutely was not placed to deal with it. International travel would have been dead regardless of policy. Most, but not all, of the economic consequences of lockdowns have been inframarginal: they would have obtained even in the absence of policy. Some have been marginal. Among those that have been marginal, some have been warranted as a way of buying us elimination. Others were stupid, but potentially unavoidable given the capacity of the public sector to manage things. In the first lockdown, there was probably no way of getting around very coarse and blunt rules about who could open and who could not.

The failure to develop more nuanced rules for future lockdowns after the first one is a substantial failure. The government pursued a pile of other policy objectives, diverting effort that should have been going into Covid preparedness. Auckland's Level 3 very likely could have been avoided by better practices at the border - the entirety of the costs of that outbreak, both the costs of lockdown and the costs of the virus, could reasonably be tallied as a cost of policy failure. The extra costs imposed by a L3 that had blunt rules about who could open and who could not, rather than risk-sensitive ones, are also a cost of policy failure - the failure to devote appropriate attention to the single most important policy area facing the country when the government seemed to think it had beaten the virus and wanted to muck around in a pile of irrelevancies in the leadup to an election campaign. 

There is still much work to be done in setting policies at the border to be able to deal with the longer term, and little evidence that that work is being undertaken. Some of the costs of a closed border are fast becoming not costs of Covid, but costs of a failed policy response. Not all - even with best-practice at the border, there is no way of returning to the status quo ex ante. 

That isn't the relevant counterfactual. 

The relevant counterfactual is a border system that increases effective capacity not only by allowing more facilities to enter the system through the kind of voucher scheme I'd suggested rather some time ago, but also by shortening stays in isolation for those coming from lower-risk places who would be required to provide location tracking facility to contact tracing teams and to present for testing post-isolation. Halving a stay in MIQ doubles the effective capacity of that room. Layering on additional testing requirements and taking advantage of the rapid cheap saliva tests coming on-stream would allow shorter stays without increasing risk. You still wouldn't get swarms of short-stay bus tourists, but you would enable piles of other things to happen. Remote workers could shift here and continue to be paid by their overseas employers. Companies finding time zones and short spells in MIQ less disruptive than dealing with Covid in their home countries could shift here along with staff willing to make the move. All kinds of options start opening up. And if this is going to be around for a while, the costs of not enabling this really start mounting - along with all of the humanitarian consequences of borders that cannot accommodate travel, and all of the consequences for domestic firms stymied in bringing in overseas experts. 

My Newsroom column this week went through some of the issues canvassed in my post on the worst case. We need to be thinking more about what the longer term looks like. This thing has at least a year to run, and potentially rather longer. 

More worrying would be that, in the longer run, in the worst case where there is no vaccine and only rolling waves of illness, Goolsbee's counterfactual won't hold either. Folks will instead, I expect, largely internalise the risk in the same way that people were happy to drive cars in the 50s that had drum brakes, no seat belts, and steering columns that would kill you. And everything then gets even trickier in running the assessments of policy. Let's hope for a successful vaccine that prevents that counterfactual from obtaining. 

No comments:

Post a Comment