A liquidity vade mecum

The Bank of England appears to be gearing up for new initiatives to inject additional liquidity into the sterling money markets. Governor Mervyn King’s appearance before the House of Commons Treasury Select Committee made this clear. I interpret the aim of these imminent actions to be lower liquidity risk premium components of the Libor – OIS (overnight indexed swap rate) spreads and, beyond that, the re-opening of many of the wholesale markets, especially markets for MBS (mortgage-backed securities) and other ABS (asset-backed securities) that have been effectively closed since the crisis started in August 2007.

Governor King was keen to ensure that taxpayers would not be left with the bad debts or stuck with the bad assets of the banks. What are the means at his disposal for extending liquidity to the banking system without taking on credit risk? What are the means at his disposal for extending liquidity to the banking system without taking up credit risk at too low a price?

(1) Outright unsecured lending to commercial banks

The central bank can directly take on its balance sheet the credit risk or default risk of a commercial bank, by making an unsecured loan to that bank or by purchasing outright a bond or other tradable debt instrument issued by that bank. The default risk that remains with the Bank of England in this case is just the probability that the borrowing bank defaults, P(B), say.

One way to do this that would directly address the Libor -OIS spreads would be for the Bank of England to guarantee unsecured interbank transactions up to a certain limit. This could be restricted to unsecured interbank lending at just one or more maturities. If the Bank of England did not want the credit risk of the commercial banks on its portfolio, it could purchase insurance against this default risk. The Credit Default Swap (CDS) market offers one set of instruments for insuring against commercial bank default risk. Another one is provided by the monolines.

With this default insurance in place (and charged to and paid by the borrowers, that is, to the commercial banks), the default risk that remains on the Bank of England’s balance sheet is the joint probability that the borrower will default and that the provider of the default insurance also defaults. Let P(I) be the probability that the bank default insurance supplier defaults and P(B∩I) the probability that both the borrowing bank and the insurer default at the same time. P(B|I) is the probability that the borrowing bank will default given that the insurer defaults and P(I|B) the probability that the insurer defaults given that the borrowing bank defaults. From the definition of conditional probability we know that P(B∩I)=P(I|B)P(B).

Obviously, P(I|B) ≤ 1, and it is in fact safe to assume that the insurance company (or the issuer of the CDS) does not default every time the bank defaults, so P(I|B) < 1. This means that the Bank of England can reduce the default risk it takes on its balance sheet by insuring that default risk: P(B∩I) < P(B). However, as long is there is some risk that the borrowing bank and the insurer will both default, the Bank of England cannot reduce the default risk it takes on its balance sheet to zero. For instance, if borrower default risk is independent of insurer default risk (admittedly an unsatisfactory assumption in a financial crisis), then: P(B∩I)=P(I)P(B).

The Bank of England could take out further default risk insurance against the chance that the original default risk insurer would default. With obvious notation, the remaining default probability that the Bank of England would carry on its balance sheet would be P(B∩I1∩I2), which is the probability that the borrowing bank and both insurers would go belly-up at the same time.

Let me emphasize right here, however, that it is not, in my view, an appropriate objective for the central bank to minimize the default risk it takes onto its balance sheet. What is key is that any default risk it does take onto its balance sheet has been properly priced, so the tax payer gets an expected return that compensates for the default risk.

(2) Outright purchases of securities held by commercial banks

The central bank can directly take on its balance sheet the credit risk of securities currently held as assets by commercial banks. Let P(S) be the probability that the security will default. With an outright purchase of the security by the central bank, P(S) is the default risk the central bank would put on its books. There is nothing wrong with that, as long as the price paid for the security (the valuation put on the MBS, for instance) appropriately reflects that default risk.

It obviously means that risky MBS and other ABS should and will not be sold at par or at their notional value. Their valuation will reflect an appropriate default risk-related discount. This valuation then also should be subject to the standard liquidity haircuts that central banks apply to this valuation.

(3) Outright purchases of securities with the default risk on the security insured by the bank selling the security (that is, with a bank guarantee)

Here the central bank is left with on its balance sheet the joint probability that the security will default and that the bank will also default (on the guarantee), that is, P(B∩S). The risk that the guarantor bank defaults could also be insured (at a price), and the Bank of England would just put on its balance sheet the risk that the security, the guarantor bank and the insurer of the guarantor bank would all default at the same time: P(B∩S∩I).

(4) Repos

From an economic perspective, the outright purchase of a security that makes a single bullet payment K at maturity, with a remaining term to maturity T and a default guarantee from the bank selling the security, is equivalent to a repo (sale and repurchase operation) with maturity T, which repays the lender (the central bank) K at maturity, using that same security as collateral. The default risk exposure of the central bank with the repo is as in (3), the joint probability that the borrowing bank will not repay the loan and that the security used as collateral will default: P(B∩S).

(5) Covered bonds

Covered bonds are backed by specific assets (like mortgages) and by the credit of the borrowing institution (its current and future cash-flow). From an economic perspective it is equivalent to a repo with the same maturity.

(6) Swaps of Treasury securities for (illiquid) private financial assets

Rather than providing cash (Bank of England reserves) in exchange for MBS or other ABS, the Bank of England could provide Gilts in exchange for these illiquid private securities. The Fed did this when they created the Term Securities Lending Facility (TSLF) for primary dealers (investment banks). Under this arrangement, the Federal Reserve will lend up to $200 billion of Treasury securities to primary dealers secured for a term of 28 days (rather than overnight, as in the previous program). Second, it will accept as collateral not just federal agency debt and federal agency residential-mortgage-backed securities (MBS), but also non-agency AAA/Aaa-rated private-label RMBS.

Since then primary dealers have also gained access to the Fed’s primary discount window, albeit just for overnight borrowing, against similar collateral as for the TSLF. This Primary Dealer Credit Facility (PCDF) is of course subject to a 25bps penalty over the Federal Funds target rate. The TSLF is at market prices.

In an earlier blog of mine I discussed the view that the Fed’s swap of Treasuries for MBS with the primary dealers had been motivated by the desire not to increase system-wide liquidity (monetise the MBS), but simply to allow existing pockets of liquidity already in the private sector to be mobilised more effectively.

However, except for some high-frequency anomalies introduced by the Bank of England’s peculiar operating procedures in the overnight money markets (discussed in yet another recent blog of mine), the difference between doing ‘sterilised repos’ (swaps of Treasuries for private assets) rather than non-sterilised repos (swaps of cash for private assets) is unimportant when the central bank uses the short-term interest rate as the instrument of monetary policy.


There is nothing new, therefore, in the statement of the Governor of the Bank of England that he will want the banks that are in need of liquidity to guarantee any securities they may sell to the bank. Repos already satisfy this requirement, and so does borrowing at the standing lending facility (the discount window).

It rules out outright purchases of assets from banks (that is, purchases without a guarantee by the seller). It does not rule out outright purchases of securities with the credit risk on the securities insured in some other way than through a guarantee from the seller of the securities. The CDS market or the monolines are possible insurance vehicles.

Guarantees by the Bank of England of unsecured interbank lending involve quite different risks, as the borrower’s guarantee is there, but there is no collateral. There is a market for secured interbank lending as well. A Bank of England guarantee of a secured interbank loan would end up looking very similar to one bank engaging in a repo and another bank in an equal size reverse repo with the Bank of England.

The key issue is not how much credit risk the Bank of England takes on its balance sheet. The key issue is that whatever credit risk the Bank of England takes on its balance sheet has been properly priced ex ante. The Bank of England, and ultimately the tax payer, must be compensated through an appropriately enhanced expected rate of return for any default risk acquired as a result of the liquidity-enhancing actions of the Bank.

Finally, the Treasury should always stand ready to recapitalise the Bank of England, should a default of sufficient scale on the Bank’s holdings of private securities blow a hole in its balance sheet that threatens the Bank’s ability to pursue its inflation target.

Default and insolvency are never a problem for a central bank if its liabilities are domestic-currency denominated nominal instruments. The central bank can always ‘print money’ to get out of an insolvency hole. But the scale of base money creation required to restore the capital position of the Bank of England could be greater than is consistent with the achievement of the inflation target.

In such circumstances, the Treasury has to stand ready to perform a capital injection into the Bank – effectively a ‘Treasury bond drop’ on the Bank by the Treasury, financed through a capital transfer.But it is unlikely to come to that.

It is my hope and expectation that the private assets the Bank has acquired and will acquire during the rest of this crisis will turn out to be a great investment for the Bank and the taxpayer. The capital gains could be large enough to justify an extraordinary dividend by the Bank of England. This should not go to its single, boring shareholder: HM Treasury. Instead it should be disbursed through a helicopter drop of money to all the citizens of the realm. I can’t wait to see that.

Maverecon: Willem Buiter

Willem Buiter's blog ran until December 2009. This blog is no longer active but it remains open as an archive.

Professor of European Political Economy, London School of Economics and Political Science; former chief economist of the EBRD, former external member of the MPC; adviser to international organisations, governments, central banks and private financial institutions.

Willem Buiter's website