I love Julia's rampaging crusade against APC overclaiming but as an actuarial analyst, I had the impression that there were some genuine cohort effects that you could just read off of plots, e.g. in UK mortality improvement. But when I go look at it, it's not necessarily supremely persuasive.