The main economic argument that the proliferation of financial derivatives has been beneficial is that it has moved us closer to the theoretical world of "complete markets" where agents can insure themselves fully against idiosyncratic risk. That is, we can make arrangements to make payments in the circumstances where we are well-off in exchange for receiving them when we are badly-off. For example, commodities futures markets allow farmers to insure against swings in the prices of their crops by locking in a price in advance (so the farmer loses some of the upside of a price increase, but avoids the downside of a price decrease). By doing so, welfare is improved, because we are reducing our consumption when it is high (and marginal utility is low) and increasing it when it is low (and marginal utility is high). That is, the "consumption smoothing" logic of the life-cycle/permanent income hypothesis that holds over time also applies across different states of the world. If financial assets are traded that pay off for each possible state, then markets are said to be "complete."
However, at his Maverecon blog, Willem Buiter argues that derivatives markets have strayed far from their insurance purpose, and that they do involve real resource costs and, in some cases, create, rather than reduce economic inefficiency. Following his suggestion to curb these markets would be consistent with Paul Krugman's recent argument that the economy would benefit if we return to a world where finance is boring (and less remunerative). Ivy League seniors appear to see that change coming, as the Times reports their preferences are shifting away from finance as a career.