[mc-lovin]

>Microeconomic models that assume the marginal productivity of a worker can be calculated in advance are drastically at odds with the reality I see

Micro theory is very general, and almost tautological. For example, you can re-interpret "marginal value" to mean "marginal expected value conditional on what the employer knows about you" and then everything you said fits neatly into the usual framework again.

Hence my comment below about the market deciding what a worker's marginal value is. Micro theory makes no claims that the "true economic" (and unobservable) marginal value of a worker is equal to their marginal value measured by some accounting system. Hence why micro theory is to a large extent irrefutable and makes few predictions about the world: if people eat big macs, we assume that increases their utility more than eating kale, and if the market doesn't pay a worker well, we assume that person contributes little marginal value.

Corporations do certainly help spread risk, that just happens to fall slightly outside of the usual micro framework.