VoxEU Column COVID-19

COVID-19 infection externalities: Herd immunity versus containment strategies

At the centre of the debate on how to deal with the novel coronavirus is whether to aim for containment or herd immunity. A crucial factor in this decision is whether we are guided by individually optimising behaviour or by overall societal welfare, since COVID-19 gives rise to substantial externalities. This column calculates that while individuals perceive the cost of becoming infected to be $80,000 the true social cost is more than three times higher, and argues that public health authorities should use mandatory measures to account for these externalities. To ameliorate the costs of the trade-off, it is crucial to develop sufficient testing and tracing capacity so that untargeted lockdowns and the economic cost involved can be ended. 

Should we let individuals decide how much social distancing to engage in, or are there good reasons why governments should infringe upon civil liberties and order citizens to stay at home? This is one of the crucial questions that underlies the debate about what public health measures governments around the world should prescribe. Economists have long recognised that infectious diseases give rise to significant externalities (e.g. Gersovitz and Hammer 2003, 2004) that call for public health interventions, but the magnitude of such externalities that the ongoing COVID-19 pandemic involves is unprecedented in modern times.

COVID-19 infection externalities

In a new paper (Bethune and Korinek 2020), we show that infectious diseases such as COVID-19 lead to significant externalities, and these externalities have stark implications for society ultimately overcomes the disease. Externalities are adverse effects on others that rational self-interested individuals do not internalize when they engage in their personal cost-benefit analysis. When choosing the extent of their social and economic activity, self-interested individuals weigh their own risk of becoming infected and the associated costs, from having mild symptoms to potentially dying, but do not internalise that when they become infected, they impose costs on others by passing on the disease. These externalities therefore call for mandatory public health interventions, such as lockdowns and quarantines, that limit the spread of the virus. 

In order to quantify how severe the infection externalities are from COVID-19, we develop an epidemiological model with rationally optimising individual agents that captures the main features of the coronavirus and the US economy. The model builds on the epidemiological literature by assuming that agents can be in one of three ‘compartments’: susceptible (S), infected (I), and recovered/resistant (R). When an agent becomes infected, the virus imposes a cost that reflects the statistical value of mortality risk, valued at approximately $50,000 given the risk profile of COVID-19. We also allow for the possibility that this cost rises as the number of infected grows to capture the potential that limited hospital capacity may increase the fatality rate of the virus. The rate of disease transmission from infected to susceptible agents depends on the extent of individually chosen economic and social activity. Importantly when choosing activity, agents only weigh the private gains from activity with the expected costs from increasing their risk of infection. 

Figure 1 

We show that individuals perceive the cost of themselves becoming infected to be around $80,000, whereas the social cost that includes infection externalities is more than three times higher, at around $286,000. The reason for the discrepancy is twofold: first, the planner recognises that infected individuals risk passing on the infection to others, who may in turn infect more individuals and so on; second, the planner also finds it optimal to impose more stringent isolation measures on infected individuals. 

Pursuing herd immunity or containment

The private undervaluation of the cost of an infection created by these externalities has stark implications for how society ultimately overcomes the disease.

Figure 2

The ‘herd immunity’ strategy

In the absence of public health interventions, susceptible individuals act cautiously, which ‘flattens the curve’ of infections. However, the disease is not overcome until herd immunity is acquired, which requires that approximately 60% of the population become resistant – either by becoming infected and recovering or by obtaining a vaccine. Since the development and roll-out of a vaccine is likely to take at least 18 to 24 months, a substantial fraction of the population will be infected. Since the risk of infection remains high for a prolonged period, the reduction in activity leads to a sharp recession and a slow recovery that takes several years. The fraction of the population infected and recovered as well as the associated output decline under the herd immunity strategy. 

The ‘optimal containment’ strategy 

By contrast, the optimal public health intervention attempts to quickly contain the outbreak of the disease. If public health measures can be targeted at infected individuals, then it is optimal to isolate and quarantine the infected, while only mild restrictions are imposed on the susceptible. This requires the widespread availability of testing, effective contact-tracing, and the isolation of infected individuals so they cannot spread the disease further, including to their families. Since the pandemic is short-lived in this scenario, economic activity could return close to normal within months.

The novel coronavirus poses stark challenges to pursuing this strategy. First, a significant fraction of individuals, estimated to be around 50%, do not present any symptoms, but some of them may still spread the virus. Second, the virus has a long incubation period of two up to 14 days, complicating efforts to engage in contact tracing. These challenges are exacerbated by the fact that many governments around the world have found it difficult to quickly ramp up their testing capacity.

Recent developments in several Far Eastern and Northern European countries warrant a certain degree of optimism that this strategy can be implemented in practice, although periodic setbacks must be expected, as experienced recently by Singapore for example.

The ‘blind containment’ strategy

If public health measures cannot be targeted because of these challenges, then containment comes at a far greater economic cost. Since public health authorities cannot single out the infected, it becomes necessary to lock down everyone’s activity, leading to a severe and long-lasting economic contraction involving a double-digit percentage decline in output. In this setting, the social cost of an additional infection is $576,000 – much higher than under the optimal containment strategy. 

For our baseline parameters, it is still optimal to pursue containment of the disease to avoid the large loss of lives, but only just. For policymakers who place a lesser statistical value on life – for example, if they only value the output losses arising from lost lives – it may be more desirable to let the disease spread through the population, pursuing the herd immunity strategy.

By implication, blind containment works best if it is used temporarily to buy time and put in place the testing, tracing and isolation strategies that enable the optimal containment strategy. If this approach is followed, then the severe cost of blind containment may be worth it under a wide range of parameter values – since it is only temporary. However, even if society follows the optimal containment strategy, there may be periodic outbreaks, especially if some countries do not get the epidemic under control, and it spreads to other countries that had previously contained it successfully. 

Life probably cannot go back to normal until an effective vaccine or treatment is developed. And here again, there are enormous externalities that individual decision makers do not internalise. When 1% of the population are infected and society follows the herd immunity strategy, we find that the social benefit of vaccinating one more person would be $430,000, whereas individuals perceive their personal benefit to be only $26,000. This implies that there are substantial public health reasons to subsidize and accelerate the development of an effective vaccine.


Bethune, Z A and A Korinek (2020), “COVID-19 Infection Externalities: Pursuing Herd Immunity or Containment?”, Covid Economics, Vetted and Real-Time Papers 11, 29 April.

Gersovitz, M and J S Hammer (2003), “Infectious Diseases, Public Policy, and the Marriage of Economics and Epidemiology”, World Bank Economic Review 18(2): 129-157.

Gersovitz, M and J S Hammer (2004), “The Economical Control of Infectious Diseases” Economic Journal 114: 1-27.

15,013 Reads