The following is a guest post by Miguel Casares (Universidad Publica de Navarra, Spain), Paul Gomme (Concordia University), and Hashmat Khan (Carleton University)
Of utmost importance is improving our understanding of the complex interactions between: (a) the epidemiology that describes the evolution of the coronavirus/COVID-19; and (b) the social and economic choices of individuals. Towards this end, we have developed a model that captures some salient elements of this interaction. The parameters of the model are calibrated to match epidemiological and economic developments in Ontario in the first half of 2020. We then evaluate the contributions of health and socioeconomic policies and the lockdowns since last March.
Interactions between epidemiology and socioeconomics. Epidemiology describes how the coronavirus spreads through the Ontario population. Key departure from the standard SIR (susceptible-infectious-recovered) epidemiological model: (a) the number of daily contacts (from the socioeconomic side of the model) affects the number of new COVID cases; and (b) asymptomatic transmission of the coronavirus. Socioeconomics characterizes how self-interested actors respond to their environment, including the prevalence of the coronavirus. These actors like consuming and socializing; they dislike working and the risk of catching COVID.
A key implication of our model is that individuals respond to the coronavirus even in the absence of government interventions. In fact, they did. For example, restaurant reservations in Ontario made through the Open Table website fell off more than a week before the declaration of a state of emergency on March 17, 2020. This is depicted in Figure 1 (percentage change relative to previous year). These individual-level actions can collectively affect the spread of the disease.
Figure 1: Restaurant reservations in Ontario made through the Open Table website
Fear of death in preferences provides a simple, elegant way of describing how COVID affects individual decision making. The fear of death introduces two wedges. First, it affects the work-consumption margin: the fear of death reduces the value of consuming since buying more goods implies more daily contacts; and the fear of death increases the disutility of working, again due to increased daily contacts. Second, the fear of death reduces the value of social activity, again because of the effect of such activity on daily contacts. As the number of COVID cases increases, individuals reduce their hours of work, their consumption, and their social activity.
First lockdown. We model government policy to fit with events from March 24 and June 11. Health-related interventions include measures that reduce contagion, like mask mandates, social distancing, increased hygiene; and testing and contact tracing. Socioeconomic policies operate through: busines shutdowns, and restrictions on socializing. While the socioeconomic interventions are very effective at reducing case numbers, they also lead to higher unemployment. The importance of the health protocols is in reducing the growth in COVID cases in the longer term. In Figure 2, the yellow dots are daily deaths, the blue line is the model’s fit, and the grey shaded areas are the lockdown periods.
Figure 2: Projected mortality with and without COVID lockdowns
Second lockdown. The model predicts rising COVID cases and mortality starting in the autumn, as is seen in the data. We then imposed a second lockdown: only essential businesses open and minimal social activity. The model predicts a rapid drop in COVID cases and mortality, saving about 4,000 accumulated deaths compared to the forecast from the scenario without the second lockdown (the red line in Figure 2).
Third wave. Absent vaccinations, the model predicts a third wave peaking in the summer. These results not only point to the importance of COVID vaccines, but also of getting as many people vaccinates as quickly as possible.
Miguel Casares (Universidad Publica de Navarra, Spain), Paul Gomme (Concordia University), and Hashmat Khan (Carleton University)
Interesting analysis. I was wondering - how is weather built into your model? I'm thinking that even if people take relatively few precautions in Ontario in the summer, most of the things that people tend to do in the summer are very low risk activities - it's just really hard to catch COVID on the beach because the UV zaps the virus and the wind blows it away. Plus school's out. So I'm just puzzled by the blue line that shows a big rise in cases in July - that seems implausible regardless of the vaccine roll-out or lockdown regime.
Posted by: Frances Woolley | February 15, 2021 at 01:20 PM
The second uptick could be the result of restaurants who never previously used reservations having to pivot to systems like OpenTable to adapt to sudden policy requirements. Furthermore, customers who never previously bothered to make reservations suddenly learned how to, and quickly got in the habit of doing so.
Disclaimer, I'm not an economist, but I do help manage a restaurant whose revenues dropped by 80%. Nothing spurs innovation like the need to survive, and using OpenTable was one of the many (often technology driven) adaptations we made.
I just took a quick look at the paper. I would like to run the simulation - I assume was done Stata or R?
Posted by: Peter Jakobsen | February 25, 2021 at 04:47 PM