One of Tyler Cowen’s readers writes,
Traditional productivity forecast research tends to assume the wage premium is entirely human capital.
[but] If sheepskin effects are purely relative status effects, then the impact on total output and income should be zero, right?
In a cross section, workers with more years of schooling will have higher wages. If you take this as an indicator of productivity differences, then in a time series in which years of schooling increase, you will predict higher productivity as these more-schooled workers enter the labor force. However, if education is only a signal of productivity and not a causal factor in productivity, then what?
Suppose that education produces zero useful work skills, and all useful skills are learned on the job. However, the workers with the best ability to learn on the job also are good at completing school. What does it mean when over time the number of workers with more education goes up? If it means that the pool of workers is getting better in terms of ability to learn on the job, then productivity should go up. If it means that more low-ability workers are somehow completing more years of schooling, then productivity should not go up at all.
Continuing with this scenario, my intuition is that the salary premium for highly-educated workers should fall, other things equal. However, in a time series, other things are not equal. For example, the technology may be changing so as to increase the value of high-ability workers. In that case, the wage premium for the high-ability educated workers could rise while that of the low-ability educated workers could fall.
Even though this scenario is extreme (education produces useful work skills in some cases), I think it may be approximately correct. In that case, the average wage premium for highly-educated workers overstates the marginal productivity premium of additional highly-educated workers over time.
In a Caplanian world, workers who do not complete a lot of schooling send an adverse signal. However, completing a lot of schooling is only a necessary condition for convincing employers that you are trainable. It is not sufficient, and firms use additional screening devices to distinguish among workers with equal numbers of years of schooling. The econometrician does not use those screening devices, and the econometrician ends up lumping together workers with different levels of ability in a way that firms do not. The firm, unlike the econometrician, sees through the worthless college degree in ____ studies. The econometrician is fooled into thinking that putting more kids through college will raise average productivity. The firm knows otherwise.