A prime example of policy uncertainty
Many of the economic and societal impacts of SARS-CoV-2 are connected to a single policy measure: the “shutdown” that enforces social distancing, including major steps like school closures, curfews and travel restrictions. The shutdown reduces the speed of economic decision making significantly. In particular, since most economic decisions are forward-looking, many of them are currently delayed by the fact that we lack an answer to a simple question: when will the governments lift the restrictions? Like the shutdown itself, the answer to this question lies in the hands of the governments. The current uncertainty about the impact of the virus is, therefore, an example of policy uncertainty. If households and firms were able to predict the shutdown’s ending date – independent of when it is – then everyone would be better off. All economic agents could return to scheduling meetings, booking tickets, training for tests and competitions, building inventories, etc.; no more future planning would go to waste.
Answering the question about the timing is, however, difficult and politically costly – one may therefore naturally ask how beneficial it is, quantitatively, to reduce the policy uncertainty. This relates to a strand of the academic literature, on measuring the negative effects of volatility and uncertainty. Papers in this tradition usually make a set of structural assumptions (e.g. expected utility with constant relative risk aversion) and ask how the welfare of economic agents would increase in a counterfactual world where all uncertainty is gone. A remarkable paper asks this question for social security policy, with minimal assumptions: Luttmer and Samwick (2018) conduct a survey among a representative sample of the U.S. population and elicit the respondents’ beliefs about their personal social security benefits, including questions about the subjective probability distribution of receiving the benefits. They also ask the respondents about their preference for a (hypothetical) guaranteed benefit. These data allow calculating a risk premium for each respondent: how much of the social security entitlement would a person be willing to give up if the entitlement came with certainty? It is very intuitive that in such a hypothetical case of certainty, long-term financial planning would be easier. The answer, depending on the precise calculations in the paper, is that the risk premium is about 5-12% of the overall benefit entitlement. A substantial amount!
What can we transfer from this very different policy context to the policy uncertainty in the present crisis? Clearly, one has to be very careful about extrapolating quantitative insights. But it is no small insight that the Luttmer/Samwick paper shows that policy uncertainty can be substantial, measured as a fraction of what is at stake. Also, one can adapt the method of Luttmer/Samwick for an analogous thought experiment: how much would economic agents be willing to pay for a perfect prediction of the shutdown’s ending date? Given how important the policy is, it is likely that the answer to this thought experiment about the policy’s end would also be substantial. The governments, of course, cannot erase the uncertainty about the shutdown’s ending date completely. But if they could make the decision better predictable, e.g. by announcing a set of objective criteria for their decisions, then prediction markets and other sophisticated tools would likely allow many economic actors to fare better.
Luttmer, Erzo F. P., and Andrew A. Samwick. 2018. "The Welfare Cost of Perceived Policy Uncertainty: Evidence from Social Security." American Economic Review, 108 (2): 275-307. DOI: 10.1257/aer.20151703