The COVID-19 pandemic exposes policymakers to several sources of uncertainty:
Under this uncertainty, policymakers must determine which policies to introduce and when to introduce them. On the one hand, acting early may prevent the virus from spreading uncontrollably; on the other, delaying action allows policymakers to collect more information about the virus and the policies best suited to combat it.
This post compares two approaches to policymaking under uncertainty: “commit early,” and “wait and learn.” These approaches highlight the trade-off between acting decisively and waiting for more information. I discuss this trade-off in the context of the New Zealand (NZ) government’s response to COVID-19. However, my discussion also applies in other (non-NZ and non-pandemic) contexts.
To frame my discussion, consider a policymaker (PM) in a world with three time periods. Each period, the world moves into one of many “states” that partition the set of possible futures (e.g., “recession” and “no recession”). Moving between the first and second periods provides more, but not complete, information about the probability distribution of period three states. The PM can influence this distribution by implementing policies in the first or second period.
Under the “commit early” approach, the PM implements policies in the first period based solely on information available in that period. To the extent that these policies are expensive and irreversible, their implementation signals the PM’s confidence in the policies’ necessity and efficacy. This signal can help prevent public dissent. For example, NZ’s relatively early transition to COVID-19 Alert Level 4 signalled our government’s strong belief that the cost of letting the virus spread outweighed the cost of entering a nationwide lockdown. This signal probably made New Zealanders more willing to tolerate the consequences of staying home—they were told, in no uncertain terms, that doing so would “save lives.”
Committing early also provides information about future conditions to households and firms, who may be less informed than the PM about the distribution of future states. For example, one of the NZ government’s earliest actions during the COVID-19 pandemic was to implement a wage subsidy scheme that provided 12 weeks of financial support to firms and their employees. The scheme signalled that our government expected the economic downturn caused by the virus to last at least 12 weeks. This signal allowed households and firms to calibrate their expectations about, and adjust their behaviour to prepare for, the future.
One problem with the “commit early” approach is that the PM may commit to policies that appear optimal ex ante but turn out to be sub-optimal ex post. As time passes, the PM gains more information about the distribution of future states and about which policies are most likely to deliver favourable states. Consequently, the PM may end up regretting committing to, and paying for, policies they wouldn’t have chosen if they had waited for more information.
This regret can be avoided by adopting a “wait and learn” approach. Under this approach, the PM delays implementing policies until more information about their relative merits arrives in the second period. This delay allows the PM to preserve the real options that would otherwise be destroyed by implementing irreversible policies in the first period. For example, delaying wage subsidy payouts until the economic impacts of COVID-19 were clearer may have allowed the NZ government to avoid giving money to businesses that didn’t need it.
However, delaying policy decisions may also delay decisions made by households and firms, who rely on policies as coordination devices. For example, delaying the decision to allow inter-regional transport may delay freight companies’ decisions to schedule inter-regional shipments, which, in turn, delays production decisions by firms with inter-regional supply chains. These supply-side delays may induce undesirable demand-side responses, such as “panic buying” food and homeware. The more peoples’ decisions depend on others’ decisions, the more the PM’s initial delay spreads throughout the economy and the more disruptive that delay becomes.
Delaying policy decisions also allows the costs of indecision (e.g., deaths from uncontrolled exposure to COVID-19) to accumulate. The PM must trade these costs off with the benefits of waiting for more information. These benefits may appear large ex ante but turn out to be small ex post. Moving into the second period may not change the PM’s preferences over policy options if the new information merely confirms what was already known in the first period. This confirmation may make the PM believe that they accrued the costs of delaying action unnecessarily. Thus, whereas the “commit early” approach induces regret when initial estimates turn out to be wrong, the “wait and learn” approach induces regret when initial estimates turn out to be right.
Although committing early destroys some real options, it may create other real options in their place. For example, NZ’s early lockdown likely prevented the virus from overburdening our healthcare providers, giving them time to plan and prepare for a range of future COVID-19 scenarios. Similarly, our government’s agreement to let the Reserve Bank buy government bonds gave the Bank more flexibility to provide monetary stimulus if the need arises. In both cases, taking early, decisive action opened paths that may have been unavailable if our government had chosen to wait and learn.
Ultimately, the PM should adopt the “commit early” approach whenever the net benefits of acting early exceed the net benefits of waiting. However, valuing these net benefits requires
These three requirements raise the complexity of the PM’s choice problem and the cognitive cost of finding its solution. Such bounds on the PM’s rationality may lead to sub-optimal policy decisions, both ex ante and ex post.
Thanks to Arthur Grimes for commenting on a draft version of this post.