Via Peter Tchir of TF Market Advisors [10],
We know how AIG and MF ended, as of yet we don't know how LTRO will end.
AIG was the ultimate carry trade. They sold massive amounts of CDS for a very small spread. They had no funding cost, no collateral requirements(initially), and no mark to market risk for their own P&L. The year before AIG blew up, management was spending all sorts of money buying back their "undervalued" shares. Actually they probably bought back a lot of shares in early 2008. What could go wrong with the trade? The only thing that could go bad was a downgrade to AIG at the same time as the assets they wrote CDS on went down, because then and only then would they have to post collateral. Their CDS had massive negative mark to markets long before they were on the verge of collapse. It was the fact that a downgrade to their ratings could enable their CDS counterparties to call for collateral. It was the threat of collateral calls that ended it for AIG.
LTRO has variation margin.
What happened at MF? Massive positions in short dated bonds to earn the "carry". All was fine until the size concerned people. They had trouble financing the position. The position was so large the market pushed against them. With so much leverage they couldn't meet the margin calls and selling the bonds would wipe them out.
LTRO has variation margin and the banks have disproportionately large positions in the debt of their country.
Lots of "carry" trades have worked out well, but when they don't, the result is pretty ugly. Many argue that LTCM was also just a giant "carry" trade that unwound when vol and mark to market blew through their ability to post collateral.