The relative power of individual distancing efforts and public policies to curb the COVID-19 epidemics

Abstract

Lockdown curbs the COVID-19 epidemics but at huge costs. Public debates question its impact compared to reliance on individual responsibility. We study how rationally chosen self-protective behavior impacts the spread of the epidemics and interacts with policies. We first assess the value of lockdown in terms of mortality compared to a counterfactual scenario that incorporates self-protection efforts; and second, assess how individual behavior modify the epidemic dynamics when public regulations change. We couple an SLIAR model, that includes asymptomatic transmission, with utility maximization: Individuals trade off economic and wellbeing costs from physical distancing with a lower infection risk. Physical distancing effort depends on risk aversion, perceptions of the epidemics and average distancing effort in the population. Rational distancing effort is computed as a Nash Equilibrium. Equilibrium effort differs markedly from constant, stochastic or proportional contacts reduction. It adjusts to daily incidence of hospitalization in a way that creates a slightly decreasing plateau in epidemic prevalence. Calibration on French data shows that a business-as-usual benchmark yields an overestimation of the number of deaths by a factor of 10 compared to benchmarks with equilibrium efforts. However, lockdown saves nearly twice as many lives as individual efforts alone. Public policies post-lockdown have a limited impact as they partly crowd out individual efforts. Communication that increases risk salience is more effective.

Fig 1. Periods and time intervals.

Fig 2. Epidemic diagram.

Fig 3. Computed value of weight k during the lockdown period [t₁, t₂].

Fig 4. Contact intensity under actual lockdown, and counterfactual rational equilibrium contact intensity for different prevalence weights k.

Fig 5. The impact of lockdown.

(A) Cumulative number of deaths under various scenarios. (B) Avoided deaths compared to business-as-usual.

Fig 6. Contact intensity modeling up to Oct. 30.

Fig 7. Calibration of daily hospitalizations.

(A) Incidence. (B) Cumulative.

Fig 8. Weight k up to Oct. 30.

Fig 9. Cumulative number of deaths during intermediate lockdown (data) vs in the equilibrium counterfactual.

Fig 10. Contact intensity under various perception weights.

Fig 11. The impact of perceptions.

Fig 12. Impact of 30% and 60% partial lockdown on contact intensity.

Fig 13. The impact of 30% and 60% partial lockdown.

(A) Daily hospitalizations. (B) Cumulative number of deaths.