The point is that this metric is dominated by the most numerous workers, which probably isn't IT outside of big tech (or even in big tech). The real metric is impact that a particular hire has on the bottom line, which is hard to measure. Really, I think it is more about their separate job markets: Exxon posts positions in Dallas and pays what they have to to fill their positions, Amazon does the same in Seattle and gets a very different set of candidates.
I wouldn't use Amazon since they have so many warehouse workers, unless you can restrict it to IT.
I was imagining some data scientist (call her "Diane") at Exxon, getting so-so wages in Dallas, and yearning to triple her salary at Facebook. Exxon just can't do that for Diane, because they have too many middle-managers who make only 30% more than Diane does now.
Then do the same for some big tech companies. Call that 'Y'. The employee count is way less than for Exxon.
Most likely, Y >> X. Vastly greater.
So to a corporate bean counter, the "value" of an Exxon employee, X, is about what they're paying them now. Hiring one more, even a brilliant tech guy, will have a negligible effect on any numbers they care about. Especially since he'll be frustrated by the dinosaur culture and his inability to have any impact.
You can say "you're wrong, what about <new energy thing>?" Well, this finance guy understands Exxon's numbers pretty well.