EFF: The evidence that short-sellers actually know something about market prices is weak. And even if they do, they only push prices more quickly to lower equilibrium levels, so they have little effect on the returns of long-term investors.

KRF: I think there are cases in which short-selling can have a real effect. Assume, for example that half the time short-sellers actually know something and half the time they don't. If other investors cannot tell which case is which, prices don't fall enough when short-sellers know something and they fall too much when short-sellers don't know anything. The market will eventually figure it out and prices will move to the correct level, but there can be real effects in the meantime. If the shorted firm is a financial institution, for example, counterparties may view the depressed price as evidence that the firm is in trouble and take their business elsewhere. Similarly, the firm may find it difficult or impossible to rollover its maturing debt. And note that these real effects occur even if the shorts had no real information. Their bet that the price will fall can be self-fulfilling.

Of course, the argument is symmetric in a couple of ways. First, the fact that shorts don't know anything in some cases implies that prices initially don't fall enough when they do know something. Second, and more important, overly-optimistic investors can push prices up in the same way that overly-pessimistic investors push them down. In this case, informed short-sellers make markets more efficient. Let me emphasize that I am not calling for general rule against short-selling or for us to stop lending securities. In general, I think the benefits of short-selling are far bigger than the costs, but I do think there can be costs.