There's a debate amongst economists about whether the value of college comes in the form of human capital versus signaling. What you're describing is sort of the signaling theory, i.e. colleges mostly function as a sieve to help employers identify students who are driven and have passed a battery of tests that have no relation to job-relevant skills except that they're hard.

The human capital theory is that education gives students new skills and knowledge that makes them more valuable. In other words, all other things equal a student that learns, say, calculus is more valuable than a student that does not.

I'm not sure what the current economic consensus is, but an interesting thought experiment to test these theories is: the human capital theory implies that you accrue value as you accrue skills/knowledge in school, so if you drop out halfway through college you should still have earned half of your human capital and economically your wages should be higher than someone who didn't go to school at all but lower than someone who graduated. The signaling theory is more discrete, if you drop out halfway you don't get any of the value of signaling because you can't say you're a graduate. The wages of college dropouts should be the same as the wages of someone who never went to college at all if the signaling theory holds true.

I'm not sure what the data says but I suspect it would be muddied by different industries and degrees. My guess would be that software engineering education follows the human capital theory -- engineering hiring managers don't seem to care where you went to school or if you went to school at all. Other industries, typically those with less hard skills like management consulting, might rely more on signaling. That's just my theory though, haven't tried to look up any data to support it.