A nice thought is that learning something like Latin, Western Civilization, or chess, generates insights one can generalize to many things later in life. Many like to think that solving logical puzzles can help the elderly stave off dementia. Unfortunately, it's hard to find evidence that this works. Bryan Caplan notes that "learning is highly specific." That is, if you want to become a programmer, or derivatives expert, don't waste your time on Classics, rather, get a job in programming or derivatives. Formal education aspires to transfer. I think it's very important to identify those concepts that transfer most easily. that is, learning to read and write is clearly a good thing useful outside of poetry. Statistics, basic logic, physics, chemistry all clearly essential at some level; number theory, set theory, string theory, not so much. Economics tried to create a theory with their basic utility function that would dominate all social science, and it had all these nice qualities and was amenable to various mathematics in generating very progressive results. Unfortunately, if they were true, risk premiums would be omnipresent and important, and they are neither. In finance learning is highly domain specific. An expert on junk bonds is just that, and knows nothing of equities (in general, of course). Same for oil futures option traders, and those trading muni-bonds. Finance has failed because finance PhDs are not preferred to simply 'smart people' in these highly remunerative fields, highlighting that all this education is not concentrated on a transferable toolkit, in contradiction to the field's assumptions.