A couple of the caveats here are:
a) The market generally has a discrete floor on contiguous blocks of time: as a reductio ad absurdum, I generally can't sell alternating 30 seconds of work out of every minute and then add that up to 1/2 time. A related concept is that there are fixed costs associated with setting up this exchange.
b) For exempt employees, more hours of work don't translate directly into more pay, though depending on the incentive structure of the firm, they may translate stochastically (e.g. working harder to get a raise). The main point is that this relationship isn't continuous and it isn't frictionless.
That being said, it's really not a bad heuristic for trying to compare the value of time vs money. If I have a preference against spending my time on some task roughly on par with my preference for not working further, I can at the very least set a bound on what my time is worth: e.g. if I know I can reliably get software contracting work for an aftertax $120/hr (incl the time taken to secure the contracts), then it's somewhere between a heuristic and a tautology to say that I'm valuing a marginal hour of free time at >$120, which suggests that I should be willing to spend $120 on something that saves me from an hour of work that's comparably-unpleasant to doing an hour of contract work.
> It doesn't cost me $400 every time I go to bed for the night.
This is a particularly poor example, since ~8 hours is a contiguous enough block that you could get an entirely different job with a night shift, so yea, the opportunity cost is literally $400/sleep if that choice was feasible.
[1] which is actually a pretty common problem for low-wage workers much more so than high-wage workers
