We are still in the relatively early stages of the COVID-19 outbreak in the United States and Europe. From an economic standpoint, the magnitude of the initial shock is still highly uncertain. Furthermore, the nature of the shock itself is also uncertain in the sense that we don’t know how this will affect long-term investment, long-term expectations of asset values, inflation expectations, and a myriad of other possible variables in our models. In that sense, it is still too early to have a strong understanding of the impact of this outbreak on the economy.
We are starting to get initial indicators on the state of the economy. Initial unemployment claims were the first real-time indicator we got for the US at the end of last week, and today the Purchasing Managers’ Index has been released for a series of countries. Both of these figures indicate a significant and sharp contraction in the economy in the last two weeks, with the New York Times reporting on state level figures indicating roughly 630,000 claims in just 15 states, and PMI figures show levels similar to 2009 in many countries. In addition, we are now seeing forecasts from major banks as they start to grapple with the implications of the current restrictions on economic activity, with some bank forecasts predicting quarterly GDP to decline at a 30% annualized rate. See here, here, and here for examples.
Each of these news items might seem scary, but I would caution against becoming too alarmed at the economic news of the moment. The forecasts by any particular bank will be unreliable, as all of them rely on previous economic relationships to either estimate or calibrate their models, and the current situation is arguably a break from the previous structural relationships. Carmen Reinhart has a good description of these limitations. Unemployment claims figures give an insightful snapshot of the current situation, but the most pressing question of unemployment remains unanswered—whether these newly unemployed people will be able to return to their jobs once the social distancing orders are lifted. PMI is probably the most insightful piece of information, as it provides direct information on consumer demand and business investment decisions. However, I would also caution that this is a particularly volatile indicator, and the numbers are likely to reverse themselves quickly once the situation changes.
Paul Krugman raises a strong point: economic indicators should be taken into perspective. What does it mean to consumer welfare for GDP to decline if both supply and demand decrease in conjunction? Likewise, if unemployment is decreasing because people are staying home due to a legitimate health concern and can return to the same job afterwards, is it as much of an economic policy concern as in normal times?
Relevant literature:
Some of the best insight we have into this type of crisis is from economic simulations of general equilibrium models under an influenza epidemic. See here and here for examples. These studies largely explore the supply-side shock of labour market shortages, with the effects of social distancing and school closures as exacerbating these shortages. In a scenario similar to the COVID-19 outbreak, with an expectation of widespread contamination, school closures, social distancing, and a mortality rate of around 2.5%, they predict a decline in GDP of roughly 6%, with tourism and entertainment sectors being especially hard hit. Arguably, the scenarios they describe are more mild than the current pandemic, with school closures lasting two weeks in one case. However, these studies also predict a relatively quick recovery as the amount of absenteeism would be limited by the length of the pandemic, with many of the effects reversing themselves quickly due to pent up demand. Some economic forecasters making the news are also predicting similar quick recoveries.
Arguably, one of the best corollaries to the current situation is in local responses to hurricanes, in which a large external event creates a need for people to stay indoors and cease economic activity. One paper by Aladangady et. al. is a particularly interesting example. These studies also point to a relatively quick recovery after the event, though the natural disasters they analyze last significantly shorter length of time than we might expect for the current social distancing order.
The real cause for concern:
In light of the potential for the economy to rebound, the real question should not be the size of the shock but of the potential transmission mechanisms. That is, how might a negative shock to the economy transform into a sustained and protracted downturn? In this context, corporate debt, distressed industries, and supply chain constraints deserve attention. However, the data on these mechanisms will not be available for some time, and current public policy initiatives on mitigating these effects are not yet determined. I will post on this topic soon.