An agent with an infectious disease confers a negative externality on the rest of the community. If the cost of infection is sufficiently high, they are encouraged and in some cases required to quarantine themselves. Is this the efficient outcome? One might wonder if a Coasian approach would generate it instead. Define a right to walk around when infected which can be bought and sold. Alas, infection has the nature of public bad which is non-rivalrous and non-excludable. There is no efficient, incentive compatible individually rational (IR) mechanism for the allocation of such public bads (or goods). So, something has to give. The mandatory quarantine of those who might be infected can be interpreted as relaxing the IR constraint for some.
If one is going to relax the IR constraint it is far from obvious that it should be the IR constraint of the infected. What if the costs of being infected vary dramatically? Imagine a well defined subset of the population bears a huge cost for infection while the cost for everyone else is minuscule. If that subset is small, then, the mandatory quarantine (and other mitigation strategies) could be far from efficient. It might be more efficient for the subset that bears the larger cost of infection to quarantine themselves from the rest of the community.
The Leisure of the Theory Class 
2 comments
March 11, 2020 at 8:10 am
Ron Peretz
Nice point, but one should also consider the fact that those who might be infected need only quarantine themselves for 14 days, whereas those who bear high risk of death from the disease will have to quarantine themselves indefinitely. Also, the individual costs increase as the congestion on the health services increases.
One more remark: another strategy might be, forcing people (possibly the young and healthy) to get the disease and then quarantine themselves until they overcome it (with high probability) and become immune.
March 11, 2020 at 6:12 pm
rvohra
Good point about quarantine duration. Re immunity, not clear that for all viral infections that surviving them confers immunity. Further, the immunity may not be long-lived as the relevant antibodies may not survive.