I think it's worth seeing how the breakdown turns out between the survivors and failures. Are the losses concentrated in a small number of major train wrecks or dispersed evenly? etc.

That makes sense, I would be interested to seeing that too.

For the purposes of assessing the industry's profitability though, we cannot exclude them - if losses are because they were badly run and lost market share to other players, then their losses will be offset by the survivals (who then buy them out).

But what I'm interested in is to measure the general health of the living. Counting in the dead (from the entire history) makes no sense.

If you are interested in the general health of the current companies over the last few years, that would make sense.

But it does not properly assess the profitability of the industry - unless you think you can pick in advance who the survivors are going to be.

If you systematically ignore the companies that go bankrupt, you will have a picture of the industry that is far healthier than the reality.

Further - especially in the airlines industry - you can't even go back too many years with the examination of the current companies - because they are the results of many mergers. So you should also be including the companies that used to be, but got merged in the existing ones.