There's a drastic difference between earnings for unfinished and finished college degree and in general this effect is dubbed as sheepskin effect.
Here's a Wikipedia starter on the effect: https://en.wikipedia.org/wiki/Signalling_(economics)#Spence_...
Is that really signalling, or is it that the reasons for not finishing a degree (not willing to put in the hard work, for instance) bleed into the workplace? The degree may play no part at all.
Interestingly, a GED does not statistically elevate ones earning position like a traditionally acquired high school diploma. It has been suggested by researchers that the reasons for not completing high school the traditional way are what lead to lower earnings, and receiving certification does not remove those existing traits. Why would college be different?
Additionally, the biggest statistical gains come from post-graduate studies. 49% of the top 1% of Americans have a post-graduate degrees. Further, there are more people with high school or less in the top 1% than there are with bachelor degree-only. This suggests to me that the gains from having a degree are skewed by supply managed professions (for example: doctors), where incomes are artificially high because the law prevents unconstrained competition. There, again, may be no advantage outside of those protected industries, and certainly not if you plan to stop at the bachelor level.
As such, degree holders may simply earn more because they are overly represented in supply managed industries with artificial incomes, like the aforementioned doctor.
The second sentence flips the causality you describe around. Employees who are already productive buy education to certify their productivity. The education does not necessarily lead to higher productivity. The employer wants productive employees, not certified employees. Certified employees are just a proxy for productive employees. Another proxy, such as aptitude tests or GitHub repositories, could function just as well. These "cost/benefit structures" may very well be more appropriate than formal education because they are less costly in time and money.
Part of the key is that degrees pay off over long periods of time in promotions and other career opportunities, so the advantage is not always apparent at age 17.
For example, someone who has innate determination to receive a degree might also be determined to finish projects in the workplace, leading to better job opportunities over someone who is less innately determined. While someone with high determination and no degree may do just as well as any college graduate, since they are lumped into the group of people with low determination, we pass them off as outliers instead of looking at the most important features.
The correlation is clear, but that does not mean there is causation.