http://www.econbrowser.com/archives/2008/03/tslf.html
[....]
This is the logical next step in Bernanke's vision of monetary policy using the asset side of the Fed's balance sheet. This strategy shift began last September, when the Fed was simultaneously implementing two kinds of operations. On the one hand, the Fed was conducting repurchase agreements, in which it takes temporary possession of certain private sector assets (including possibly mortgage-backed securities) and gives cash to the recipient by creating new Federal Reserve deposits. Essentially a repo is a short-term collateralized loan from the Fed. On the other hand, the Fed was simultaneously conducting open market sales out of the Fed's holdings of Treasury securities (or allowing existing holdings to expire through redemption), either of which by itself would absorb Federal Reserve deposits . The combined effect of the dual operations was to leave the Fed's total assets (and therefore its total liabilities) unaffected. Since there was no change in total liabilities, it had no direct implications for the total volume of Federal Reserve deposits or the money supply, which is how we usually think of monetary policy affecting the economy. But by reducing the Fed's holdings of Treasuries and increasing the Fed's holdings of the procured collateral, the swap allowed banks temporarily to replace problem assets with good funds, at least for the term of the repo.
In December, the Fed introduced the term auction facility as a device for implementing such swaps on a bigger scale. In these operations, banks could offer a variety of assets as collateral, and receive loans of Federal Reserve deposits. Again these operations were offset by open market sales of Treasuries so as to keep the total volume of Fed assets and liabilities unchanged. By my calculations, the Fed is currently holding $80 billion in assets under this facility, almost identical to the amount by which it has reduced its holdings of Treasury securities, and amounting to about 10% of its previous total Treasury holdings.
And the new TSLF will do the same thing on an even bigger scale-- $200 billion has been announced as the first step. Under the TSLF, the Fed will temporarily swap more of its Treasury holdings for private sector troubled assets, and thereby eliminate the middle man required with repos or the TAF. No need to add reserves through the repo or TAF and then drain them out with open market sales. Instead we just swap the Treasuries directly for the target assets.
[....]
It appears to me that the Fed's unconventional new measures surely involve at least some absorption of risk by the Federal Reserve itself. It therefore seems appropriate to try to think through the implications of what will happen if indeed the Fed is not repaid on some of these loans and gets stuck holding the inferior collateral.
It strikes me that the immediate accounting implications of such a default would be nil-- the Fed now holds the collateral as an asset, can claim its value equals that of the original loan, and can carry it on that basis at least for several years, with no changes in its other assets or liabilities necessitated by the fact that the collateral is, in fact, not worth what it's being carried at. The main cash flow implication that I can see is the following. When those Treasury securities were held by the Fed, the interest that the U.S. Treasury owed on the securities was a line entry for the U.S. Treasury (gross interest expense) that was exactly canceled by another entry (receipts returned from the Fed) to determine the "net interest" that the Treasury had to pay. With those Treasury securities now owned by the private sector instead of by the Fed, the Treasury's going to have to make those payments with actual cash, and if the collateral is nonperforming, the Fed doesn't have any receipts to return to the Treasury. The net cash flow consequences for the Treasury of a default would therefore be identical to those if the Treasury were simply to have borrowed up front a sum equal to the difference between the amount that the Fed lends and the amount that it is repaid.
In other words, the Fed seems to be committing the Treasury to cover any losses that may be incurred.
Even in the worst possible outcome, the ultimate increase in outstanding Treasury debt would be substantially less than $400 billion, because the collateral is far from worthless. And I would trust the Fed to be taking a smaller risk on behalf of the Treasury than I would expect to be associated, for example, with congressionally mandated expansion of FHA insurance, or the unclear implicit Treasury liability that results from increasing the assets and guarantees from Fannie or Freddie. Nevertheless, the doubters seem to me to be correct that the risks currently being absorbed by the Federal Reserve are substantially greater than zero.
You don't get something for nothing.