We have found that some exposed to the virus do not develop any antibodies. A publication is on the way. These persons are asymptomatic or weakly symptomatic. Many children in this group. Probably the innate immune system that provides a natural barrier against the virus. We don't know how many though. And don't know yet how to estimate this number, this is complex. Any idea? What is sure is that this group has a non-negligible size and that in turn the IFR is lower than most believe.
I would survey and antibody test households where at least one member was previously confirmed with the virus. This would be an in-home survey for measuring 1) overall closeness of household quarters, 2) closeness (in time and space) of each individual to the infected individual and 3) collecting finger prick blood samples from each household member.
Correcting for 2), the null hypothesis might be that all family members would have the same antibody result (whether positive or negative). In reality however, what would probably be found is that the antibody production is correlated with increasing age and that children under 10 will generally not have antibodies.
(With influenza, things are somewhat different. Children produce antibodies to the infection, but of limited scope, restricted to the infecting organism. Adults produce a wide spectrum of antibody response. See https://mbio.asm.org/content/11/1/e03243-19 )
So ... given that the result is that children are generally immune from covid19 and produce little or no antibody, grade schools should reopen. (In Sweden they never closed.)
Kids should be allowed to play together. It's the post pubescent folks who should keep practicing social distance.