A contributing factor might be that the cost of training doesn't necessarily have a direct impact on the starting salary. In your example a petroleum engineering grad earns almost 60 percent more than an MIS grad, but presumably the cost of education is similar. The difference is that engineering school is highly competitive and people with engineering skills are widely sought.

Becoming a successful engineer takes a certain combination of intellectual capacity, willingness to work through difficult material, and interest in the field. Within the population of people who can become engineers (ie college freshmen), the percentage that will become engineers is fairly small - because most are weeded out by one of these hurdles. Hence, the "supply" of engineers is relatively small in a world where engineering expertise is increasingly valuable.

In short, the cost of training has a weaker relationship to salary than the relative demand for a certain skillset.

Of course if the cost of training gets too high (relative to the expected salary), fewer (rational) skill seekers will choose to participate in a given field. As a result of this and industry attrition and/or growth, the labor supply can become relatively more scarce. This is when elementary supply and demand concepts predict that wages will begin to increase, so as to satisfy equilibrium of supply and demand. And this is what so many regional pilots are betting on in the current environment.

I hope it works out. But the real world isn't fairly represented by the vacuum in which undergrad econ is taught - there's a lot of friction here, perhaps most notably government regulation and the CPAs that regionals operate under. In others words, I don't think meaningful wage increases are likely anytime soon.

I think you answered your own question... isn't this "the reason?"