There is a statement in Gregory Zuckerman's terrific book that really struck me: he notes that people hate negative carry, and far prefer positive carry (I don't had the book in front of me right now, so I need to paraphrase). In Paulson's context, he was able to buy credit insurance very cheaply--this limited his downside risk in a way shorting would not, while allowing him to invest consistent his bearish views on the housing market. But it also meant he was paying out cash flow and not gettting anything in return until subprime mortgages and other instruments began failing.
To some extent, there is a discounting issue here: if investors take losses on the negative for several periods, the gains they receive in the future will be discounted. But still, it is an interesting question whether investors discount negative carry trades too much--whether the typical Wall Street investor sold Paulson insurance that was, under reasonably discounting, an ex ante positive NPV bet for Paulson. I am not sure how one would go about testing this, though....
