Abstract: Optimizing the impact on the economy of control strategies aiming at containing the spread of COVID-19 is a critical challenge. We use daily new case counts of COVID-19 patients reported by local health administrations from different Metropolitan Statistical Areas (MSAs) within the US to parametrize a model that well describes the propagation of the disease in each area. We then introduce a time-varying control input that represents the level of social distancing imposed on the population of a given area and solve an optimal control problem with the goal of minimizing the impact of social distancing on the economy in the presence of relevant constraints, such as a desired level of suppression for the epidemics at a terminal time. We find that with the exception of the initial time and of the final time, the optimal control input is well approximated by a constant, specific to each area, which contrasts with the implemented system of reopening ‘in phases’. For all the areas considered, this optimal level corresponds to stricter social distancing than the level estimated from data. Proper selection of the time period for application of the control action optimally is important: depending on the particular MSA this period should be either short or long or intermediate. We also consider the case that the transmissibility increases in time (due e.g. to increasingly colder weather), for which we find that the optimal control solution yields progressively stricter measures of social distancing. We finally compute the optimal control solution for a model modified to incorporate the effects of vaccinations on the population and we see that depending on a number of factors, social distancing measures could be optimally reduced during the period over which vaccines are administered to the population.
Introduction
The central challenge of the COVID-19 pandemic was the dilemma between saving human lives and preserving economic stability. Policy makers struggled to balance the need to suppress viral spread to avoid overwhelming healthcare infrastructure with the massive economic costs associated with lockdowns. This research bridges the gap between data-driven epidemiological modeling and economic optimization, providing a quantitative argument for "steady-state" social distancing over the "phased" approaches adopted by many governments.
Methodology
The researchers employed a multi-step analytical pipeline to determine the most economically efficient way to suppress the virus.
Compartmental Modeling
They developed a modified SEIR model with eight compartments, specifically accounting for asymptomatic carriers ($A$), self-quarantined individuals ($I_{sq}$), and those identified through testing ($I_{tp}$). The model incorporates a time-varying control input $P(t)$ representing the level of social distancing, which modulates the transmission rate $\beta$.
Data Parametrization and Optimal Control
Using Markov Chain Monte Carlo (MCMC) methods and a negative-binomial noise model, the researchers fitted their model to historical case data from four US Metropolitan Statistical Areas (MSAs): New York City, Los Angeles, Houston, and Seattle.
They then defined an objective function ($J$) representing the economic cost of social distancing. The goal was to minimize this cost while satisfying two major constraints:
- Terminal Suppression: Reducing the infected population to a target level (e.g., 1 in 100,000) by a specific end date.
- Path Constraint: Keeping the number of infected individuals below the local hospital/ICU capacity ($I_{max}$) at all times.
Key Insights and Results
The Suboptimality of Phased Reopening
The most striking finding is that the optimal control input is nearly constant over time. This directly contradicts the "phased reopening" (Phase 1, 2, 3) strategies implemented globally. The authors argue that alternating between relaxing and tightening restrictions is economically inefficient because the benefits of momentary relaxation are outweighed by the high costs of subsequent "strangleholds" needed to fix the resulting surge.
Regional Variation and Actual vs. Optimal
In all four MSAs, the calculated optimal level of social distancing was stricter than the level estimated from actual data at the time. NYC’s existing control measures were found to be the closest to "optimal," while Houston and LA required much more significant initial tightening to reach a path toward suppression. Furthermore, for cities where the epidemic was on the rise, a longer planning horizon resulted in a lower total economic cost by allowing for a more gradual, steady approach.
Seasonality and Vaccination
When accounting for increasing transmissibility (e.g., due to colder weather), the optimal strategy shifts from a constant to a linearly increasing strictness in social distancing. Conversely, the introduction of vaccines impacts the optimal strategy almost immediately. Even when only 6–8% of the population is immunized, the optimal level of social distancing can begin to relax, serving as a "bridge" to eventual safety.
Discussion and Limitations
While the model provides a robust framework for optimization, it treats each MSA as an isolated system, ignoring viral "leakage" between cities due to travel. Additionally, the model assumes testing resources are not scarce and are 100% accurate. The economic cost function is a simplified proxy for GDP loss, and future work could incorporate more sophisticated socio-economic models that differentiate between types of essential and non-essential work. Finally, the vaccination analysis does not account for the emergence of new variants that might partially evade vaccine-induced immunity.
Summary updated