Concerning the 1998 resolution of troubled hedge fund Long Term Capital Management, Brad Setser writes,
The big banks called to the New York Fed were the creditors of LTCM and they were in some sense "bailed-in." To avoid taking losses on the credit that they had extended to LTCM, they had to pony up and recapitalize LTCM.*
It just so happened that the market recovered and it was possible for LTCM to exit many of its positions without taking large losses, or in some cases any losses. The banks that took control of LTCM when LTCM was on the ropes were able to unwind LTCM's portfolio in a way that didn't result in additional losses. But the result [Tyler] Cowen desired -- large losses for the banks and broker-dealers who provided credit to LTCM - was quite possible if LTCM's assets weren't sufficient to cover all its liabilities. No creditor of LTCM was able to get rid of its exposure as a result of the Fed's actions.
This sounds to me very much like the stern-sheriff model that I think should have been applied to the credit default swaps market. Force all of the creditors (firms who have obtained credit protection from counterparties such as AIG) to sit back and wait for the underlying bonds to default before taking action. Make them bear more risk, just as LTCM's creditors did, rather than reduce the choice to a government bailout or a bankruptcy. Note that Setser thinks that the stern-sheriff model would have been difficult to execute with Lehman, because there were so many counterparties involved.
Concerning LTCM, Tyler Cowen could say that even though the stern-sheriff model was followed back then, the lesson that investors learned was that folks could lend to over-leveraged institutions and get away with it. If so, then that was a bad lesson to have learned, because investors should have stopped taking on credit exposure to Bear, Lehman, and the other over-leveraged investment banks much earlier than in 2007.