It's not yet to be determined. Cases misses most infections because they're usually mild. Science knows how many people were infected as a result of serological studies that constantly reproduce the same rates of surveillance (cases) and infections (reality).
What I did was take outcomes to arrive at death rates as an outcome.
(Dead people/Recovered people) is roughly 4 to 5 percent.
IFR mixes yet to be determined outcomes with already determined outcomes.
This understates the risk, could be considered best case.
That is:
(Dead People/(people currently sick+recovered people))
If everyone currently infected will live, the IFR is a very best case assumption.
On the other hand, if all those currently sick people were to die, that is a worst case scenario.
In my view, anyone gauging their risk on IFR is doing so on a very wide error margin and said error understates real risk considerably.
None of this takes post infection effects into account. Living, but with organ damage is a real thing and we have poor data on it.
The chance of that happening is currently higher than outcome death rates are.