Wanted: tough love from the central bank

The heads of the five biggest UK banks met with Governor Mervyn King of the Bank of England on Thursday 20 March 2008 to ask for more liquidity support. They should get it, but in the right manner and on the right terms. Liquidity should be provided on demand, against a wide range of collateral and at maturities of up to a year. But it should be provided in the right way, and on the right terms. The same applies to the Fed and the ECB. This is the time for tough love, if the resolution of the current crisis is not to sow the seeds for a worse crisis five years from now. Flexibility and responsiveness, by all means. Generosity, no.

The inherent fragility of leveraged financial institutions

Financial stability ultimately rests on confidence and trust. Confidence and trust are always fragile. Any highly leveraged institution can be brought down by a ‘run’. It is the economic function of highly leveraged financial institutions to invest in long-maturity illiquid assets and to fund this with liabilities that are of short maturity. These liabilities are often liquid during normal times but can become illiquid during panics, when depositors withdraw their deposits and other creditors are unwilling to roll over their maturing loans, let along extend net additional credit.

With traditional banks, which fund a portfolio consisting largely of illiquid, non-traded loans with deposits withdrawable on demand and subject to a sequential service (first-come, first-served) constraint, we get the depositors’ bank run, familiar from the Great Depression in the USA and Continental Europe, and recently reprised in the UK with Northern Rock after an absence from these shores of more than a century.

Clearly, it makes sense to try to withdraw your deposit if you suspect the bank’s loan portfolio is of poor quality. It also may make sense, however, to try to withdraw your deposit even though you know that the bank’s loan portfolio is of the highest quality if only the loans could be held to maturity. This will be the case if you fear that a sufficiently large number of other depositors may try to withdraw their deposits, that is, if you fear a withdrawal of deposits so large that the bank runs out of liquid reserves and has to start liquidating its illiquid loan portfolio. If this can be done at all, it will be at ‘fire sale prices’, which represent a severe discount on the fundamental value of these assets. There are therefore always two kinds of self-fulfilling, individually rational equilibria. The first, efficient, equilibrium is where no one runs because no-one expects a sufficiently large number of other depositors to run. The second, inefficient equilibrium is where everyone runs because everyone expects a sufficiently large number of other depositors to run.

If there is no other way for banks to fund themselves (replacing the departing depositors with other creditors, or issuing asset-backed-securities or covered bonds secured against their loan portfolio) the bank that falls victim to a run equilibrium is dead. A fundamentally solvent institutions can be made insolvent through lack of liquidity.

There are several solutions to the classical bank run problem. One is deposit insurance. To be credible in the face of a potential banking sector wide run, this must ultimately be backed by the tax payer, that is, the Treasury. Another solution is a ‘standstill’ or deposit withdrawal moratorium. The simplest version is simply to close the bank or banks that are victims of a bank run, and to wait till the panic dissipates. The third solution is the central bank acting as lender of last resort: lending freely at a penalty rate against collateral that would be good in normal times (when markets are orderly) but that may become impaired in abnormal times (when markets are disorderly).

The same holds for ‘bank runs’ on other highly leveraged institutions, e.g. non-depository banking institutions like investment banks (Bear Stearns) or other highly leveraged institutions in the shadow banking sector like hedge funds and private equity funds. Here the rush of depositors wishing to withdraw their deposits becomes a credit market and wholesale markets strike. Existing non-depositor creditors refuse to renew maturing credit. New creditors cannot be attracted on any terms. The wholesale markets that willingly bought the ABS and asset backed commercial paper (ABCP) used to fund the illiquid investments of the banks and quasi-banks, now refuse to buy those securities at virtually any price. This is what killed Northern Rock and Bear Stearns. Here too, a rational refusal to extend or renew credit to an insolvent institution, which is not something the central bank should concern itself with, has a parallel in a world where the bank is solvent if it is not forced to sell its illiquid assets prematurely and in a hurry, in the form of a credit and wholesale markets strike.

When faced with a credit markets and wholesale markets strike, or seizure, central banks like the Bank of England, the Fed and the ECB should act as market maker of last resort. The central bank should either accept as collateral for loans or purchase outright, securities that have become illiquid as a result of fear, mistrust, lack of confidence and widespread panic. It should do so at a price. The valuation of the illiquid securities offered for sale or as collateral should be sufficiently punitive, that the expected rate of return to the central bank (and thus to the tax payer) compensates for the increased credit risk (default risk) the central bank takes onto its balance sheet.

A variety of auction mechanisms (e.g. the reverse Dutch auction) exist to make sure that the central bank does not subsidize the banks to whom it lends or whose assets it purchases outright. A reverse auction is a type of auction in which the role of the buyer and seller are reversed, with the primary objective being to drive purchase prices downward. In an ordinary auction (or forward auction), buyers compete to obtain a good, service or asset. In a reverse auction, sellers compete to obtain business. I a reverse Dutch auction, there is one buyer (in this case the central bank) and multiple sellers, and the auctioneer raises the price from a low starting point until a bidder agrees to sell at that price. If the buyer has a predetermined amount of the asset he wishes to buy or a given amount he wishes to spend on the asset (say, in the case of the Bank of England, it wants to buy either a given face value of asset-backed securities (ABS) or it wishes to inject a given amount of sterling liquidity by purchasing ABS) it will, after satisfying all sellers willing to buy at his lowest bid price, raise his bid price and buy whatever is offered at that price, up to the predetermined limit of the auction. The price is raised progressively until there either are no would-be sellers left or until the buyer meets his quantity objective.

What central banks should do

(1) Broaden the range of eligible collateral in repos

The Bank of England in particular should accept both in sale and repurchase operations (repos) a much wider range of collateral than it has hitherto. It should accept as collateral a slightly extended version of what the ECB currently accepts. The ECB accepts, subject to some minor qualifications, euro-denominated securities, public and private, issued in the European Economic Area (the EU plus a few bits and pieces), that are rated at least in the single A category. This includes ABS such as residential mortgage backed securities (RMBS), and indeed pools of mortgages themselves. For some reason the ECB restricts collateral to be euro-denominated and does not accept ABS where the cash-flow generating assets backing the ABS consist, in whole or in part, actually or potentially, of credit-linked notes or similar claims resulting from the transfer of credit risk by means of credit derivatives.

The Bank of England should accept as collateral in its repos any security rated at least in the single A category. The Bank of England already accepts (in exceptional circumstances) US Treasury bonds (denominated in US dollars) as collateral. It should accept collateral denominated in any freely convertible currency, but should charge the would-be borrower offering such collateral the full cost to the central bank of the necessary foreign exchange cover. There is no need to exclude ABS whose cash-flow generating assets backing the ABS consist, in whole or in part, actually or potentially, of credit-linked notes or similar claims resulting from the transfer of credit risk by means of credit derivatives. That is an arbitrary and silly restriction. Risk is risk. As long as the central bank gets compensated adequately (through penalty interest rates, conservative valuations for the securities and sizable liquidity haircuts on those conservative valuations) for taking the risk, they should take the risk.

(2) Increase range of maturities at which the central bank injects liquidity

Until the crisis that started in August 2007 the Bank of England had not tried to influence, through open market operations, any interest rate other than the overnight rate in the sterling interbank market. It is not quite true that Bank Rate is the target for the overnight sterling interbank rate, the way the Federal Funds rate target is the target for the overnight rate in the US dollar interbank market – it is in fact the rate the Bank of England charges eligible borrowers on its reserves. For the purposes of this blog, we can take Bank Rate to be the target for the (unsecured) overnight sterling interbank rate. The Bank of England supplies the amount of short-term liquidity that it expects will keep the effective overnight interbank rate close to Bank rate.

The Bank has historically provided liquidity through repos at maturities ranging from intraday to a year or longer, but at all maturities beyond overnight, it has effectively taken market rates as given, that is, driven by market expectations about future official policy rates and by the market’s perception of bank default risk. It has not tried to influence interest rates at these longer maturities by varying the amount of liquidity it supplies at these longer maturities.

The spread between Bank rate and, say, 3 month Libor is the sum of the expected change in the policy rate over the three month horizon, a default risk premium (the 3-month Libor rate is for unsecured (non-collateralised) interbank lending and a liquidity risk premium. The 3-month overnight indexed swap (OIS) rate represents the market’s best guess at the average policy rate over the three month horizon. The spread between 3-month Libor and the 3-month OIS rate is therefore the sum of the banks’ default risk premium and the liquidity risk premium.

The Bank of England holds the view that it should not try to bring down the market’s assessment of bank default risk. It does now, I believe, accept the proposition that it ought to try and bring down the liquidity risk premium, if it can. It believes, using data from the credit default swap (CDS) markets to get a sense of the banks’ default risk premium, that although in August and September 2007 the bulk of the Libor-OIS spread at longer maturities reflected liquidity risk, by November-December 2007, most of the (smaller) spread reflected default risk (see Bank of England, Inflation Report February 2008, Chapter 1, Money and asset prices, and especially Chart 1.4 on page 11).

I actually agree with the Bank of England’s judgement about the relative contributions of default risk and liquidity risk qualitatively, but I don’t believe their quantitative, numerical decomposition. After all, if you are illiquid for long enough, your risk of insolvency increases. It is not credible that the CDS markets reflects only pure default risk (that is, the default risk that would exists even if markets were liquid and orderly), rather than pure default risk plus the effect of market illiquidity on default risk.

I agree with the Bank of England that, when markets are orderly, the Bank of England cannot set more than one interest rate, unless they happen to pick two rates that are consistent with the fundamental term structure of interest rates driven by market expectations of the Bank of England’s future policy actions. But when markets are disorderly and illiquid, there is a term structure of liquidity risk premia that the central bank can and should try to eliminate by injecting as much liquidity at all the economically important maturities at which these liquidity risk premia occur. This means in practice, under current market conditions, that the Bank of England should aggressively inject liquidity at 1-month, 3-month, 6-month, 9-month and 12-month maturities.

(3) Engage in outright purchases of a wide range of systemically important private securities

When it is clear that the illiquidity of a certain class of securities (and the effective closure of the markets they are traded in) is the main proximate cause of the funding problems faced by key players in the financial system, and that the solvency of systemically important financial sector institutions is threatened by this market illiquidity, the central bank should take the credit risk (default risk) of these securities directly onto its balance sheet by outright purchases of these securities. Open market operations routinely involve outright sales and securities of sovereign debt instruments. In times of crisis, this practice should be extended to systemically important classes of private securities.

Residential mortgage backed securities (RMBS) other mortgage backed securities (MBS) and a variety of other mortgage-backed securities (ABS) including possible asset backed commercial paper (ABCP) fall into the ‘candidate for outright purchase by the central bank’ today. Such outright purchases of illiquid private securities are likely to happen on a significant scale in the UK, the US and the Euro Area before this crisis is over, possibly quite soon.

It is essential that the price paid by the central banks for these risky securities fully reflect the best guess at the underlying fundamental default risk. I am sure that the heads of the five largest UK banks that met with Mervyn King last Thursday, would love the Bank of England to value their risky mortgages and MBS as if there were no default risk attached to them. They should be disappointed. Reverse auctions, of which one nicely aggressive example is the reverse Dutch auction discussed earlier, should be used to extract an adequate return for the central bank and for the tax payer for their exposure to these risky securities.

(4) Broaden the range of eligible collateral at the discount window (standing lending facility)

Eligible collateral at the discount windows should include a wider range of securities than those eligible for repo in liquidity-oriented open market operations. I would propose that the Bank of England accepts any investment grade securities as collateral at the standing lending facility. By this I mean that any security that has a credit rating of BBB- or higher by Standard & Poor’s or Baa3 or higher by Moody’s should be collateralisable (at an appropriate price). This is less demanding than my proposed standard for collateral in repos, which is A- or higher according to S&P or A3 or higher according to Moody’s. It is conceivable that speculative grade or junk securities may have to be added to the menu of assets collateralisable at the discount window, but there ought to be enough investment grade stuff in the right places in the banking sector to do the job.

The discount window should only be used by banks that find themselves seriously liquidity-constrained and may not have the collateral that enables them to borrow overnight at the policy rate or at regular market-rate open market operations at longer maturities.. The penalty rate for discount window access should reflect this. It does for the ECB and the Bank of England, where the discount window penalty spread is 100 basis points, but no longer in the US, where the Fed has, for no good economic reason, reduced the discount window penalty spread over the policy rate to 25 basis points.

(5) Lengthen the maturity of discount window borrowing.

Since this crisis began, the Fed has increased the maximum maturity of its discount window loans first to 30 days (in August 2007) and more recently to 90 days (in March 2008). The Bank of England should follow suit. Rolling over on a daily basis overnight discount window loans from the central bank is not a mode of existence likely to install confidence in would-be private counterparties for the troubled financial institution that is accessing the discount window.

(6) Broaden the range of eligible counterparties in the full range of financial stability arrangements provided by the central bank.

The current eligible counterparties for the Bank of England’s various liquidity- and stability-oriented arrangements are most easily understood by looking at the following Chart from the Bank of England’s ‘Redbook’, The framework for the Bank of England’s operations in the sterling money markets. (Either I or/and the blogging software is (are) pants at inserting graphics; you will have to look in the Redbook yourself).

It is still the case in the UK and the Euro Area that effectively only deposit-taking institutions – commercial banks and universal banks – can access the discount window. Open market operations have a larger set of counterparties. I will mainly consider discount window facilities here.

In the US, the Fed has just extended access its discount window to the 20 (if we still count Bear Stearns) Primary Dealers, non-deposit-taking banks with a wholesale clearing and dealing role. The collateralised loans that can be obtained at the new Primary Dealers Credit Facility are only for the overnight maturity (the regular discount window now provides loans for up to 90 days) and will be made through the agency of a conventional clearer, but it is an important step.

With institutions eligible to access the discount window still hoarding liquid assets rather than passing the liquidity on to parts of the financial system without direct access to the discount window, it is desirable to extend access to the discount window to all highly leveraged financial institutions that meet two criteria. First, they should be listed, that is, public companies. Second, they should be regulated to the satisfaction of the central bank.

(7) Do not reduce the penalty rate for borrowing at the discount window.

In the UK, banks can borrow at the standing lending facility on demand against eligible collateral at a rate 100 basis points above Bank Rate. In the Eurosystem, banks can access the marginal lending facility against eligible collateral at the same 100 basis points spread over the policy rate. In the US, the Fed reduced its discount window penalty (for overnight borrowing) over the policy rate from 100 basis points first to 50 basis points on August 17, 2007 and to 25 basis points in March 2008. This was a mistake – an inframarginal subsidy from the tax payer to the shareholders of banks that have the collateral to borrow at the discount window. It doesn’t do anything for liquidity, but it is a nice Easter egg gift from the Fed to Wall Street.

(8) Minimize moral hazard: pay a bad price for bad assets

Minimising the damage to the real economy from a financial crisis through ex-post assistance and bail-outs always creates moral hazard. A key policy issue is to minimize the damage. There are two reasons for this. The first is efficiency, the second fairness. Unless those who brought us the crisis (the irresponsible lenders and borrowers in the financial sector and the household sector) are made to pay a price, the minimisation of the short-run damage caused by the present crisis will sow the seeds for an even larger future financial disaster a decade or less from now.

The first rule is simple: don’t pay good prices for bad assets.

Collateral accepted in repos and at the discount window (or similar facility for non-banks) should be valued aggressively to compensate the central bank for the default risk it is assuming.

Any private securities purchased outright by the central bank should likewise be valued not at par or notional value, or at book or historic cost, but at a price that is sufficiently discounted to offer the central bank, and ultimately the tax payer, adequate ex-ante compensation for the credit risk it is taking on its balance sheet.

The penalty rate applied at the discount window should not be at least 100 basis points. The Fed set a bad precedent by cutting the penalty to 25 basis points.

The second rule is equally simple: save systemically important institutions, not their shareholders, creditors or management.

If neither the repo market nor the discount window can keep a vulnerable financial institution alive (despite the much wider range of collateral that should be on offer), systemically important institutions will be supported with more heavy-duty lender of last resort arrangements, like the Liquidity Support Facility for Northern Rock. As soon as a financial institution goes to drink at such an idiosyncratic well, it should be subject immediately to a Special Resolution Regime (SRR), akin to what the FDIC has for the deposit-taking institutions. Its assets and liabilities would be ring-fenced. A special administrator would be appointed by the regulator (currently the SEC for investment banks, but probably not for much longer). The incumbent top management (the top 12, say) and entire board would automatically be fired without any kind of handshake, leaden or golden. The shareholders would have no say in the running of the institution any longer. If the institution is not liquidated, they will still own whatever it is worth when it comes out of the SRR. If it is liquidated while in the SRR, the shareholders will be last in the line of claimants on the realisation of the defunct institution’s assets.

In the case of Bear Stearns, the Fed did not prevent, and may have tacitly encouraged, the low bid price for its equity (2 dollars per share) by JP Morgan. No shareholder moral hazard therefore for Bear Stearns shareholders (I am not so sure about JP Morgan shareholders, however…). However, the management and board of Bear Stearns are still in place, as far as I know. That makes no sense. Also, the Fed seems to be trying to convey the message that while the shareholders of investment banks and other financial institutions are not protected, all other creditors are. That seems a tad over the top. This sharp discontinuity in the protection offered by the Fed when shareholder equity is exhausted, encourages reckless lending to banks.

The Treasury as the recapitalisor of last resort, must visibly and audibly stand behind the central bank

As long as the securities the central bank purchases outright or accepts as collateral in loans or in repos are denominated in domestic currency, there is never any risk of central bank default or insolvency, no matter how large the stock of dodgy assets on its balance sheet becomes. After all, the central bank can ‘print money’ that is issue currency and increase commercial bank reserves at will and virtually costlessly.

These monetary (base money) ‘liabilities’ of the central bank are not in any meaningful sense liabilities, because they are irredeemable: owning a given amount of base money does not give the holder a claim on the central bank for anything else than that same amount of base money. If you are lucky you can get two fivers for a tenner, but that is all. Since currency (the bulk of base money under normal circumstances) is non-interest bearing, issuing currency is a truly good business. With base money also legal tender in the relevant national or (in the case of the euro, multi-national) jurisdiction, a central bank can always honour its domestic-currency obligations by printing money.

In the midst of a liquidity crisis, large increases in the stock of nominal base money (central bank injections of liquidity) are unlikely to be inflationary. The liquidity crunch means that there is a massive increase in ‘liquidity preference’, that is in the willingness of households to hold currency and commercial banks to hold reserves with the central bank. An increase in the nominal stock of base money that merely accommodates an increase in the demand for real base money at the given price level, and for a given expected rate of inflation, is not inflationary.

So far the good news about central bank solvency. The bad news is that when the panic subsides and orderly and liquid markets return, the demand for real base money will decline precipitously to normal, non-crisis levels. At that point the earlier central bank monetary issuance will become inflationary. So, unless the Treasury at that point (and preferably earlier) recapitalises the central bank, by giving it a capital injection (basically by stuffing its balance sheet with a gift (a capital transfer from the Treasury to the central bank of Treasury bonds and bills), the central bank can only maintain its solvency when faced with a massive capital loss, by engaging in monetary issuance that will be inflationary in the medium and long term.

The credibility of the central bank as a non-inflationary lender of last resort and market maker of last resort therefore depends ultimately on the fiscal backing of the central bank. The Treasury and ultimately therefore the tax payer, have to be willing and able to make up the losses of the central bank.

In the UK this became obvious when the Treasury has to authorise the creation of the Liquidity Support Facility for Northern Rock, through which the Bank of England made collateralised loans (on terms we still don’t know!) to Northern Rock. The credit risk would ultimately be borne by the tax payer, if the Bank of England was to maintain the credibility of its inflation target.

So I would have liked to see US Treasury Secretary Henry Paulson stand next to Ben Bernanke and Timothy Geithner when they announced the $30 bn loan to JP Morgan. This loan, a sweetener to ease the pain to JP Morgan of funding Bear Stearns least liquid assets, looks suspiciously like a $30 bn first-loss guarantee from the Fed to JP Morgan on Bear Stearns’ assets. Indeed, I would have expected to see a more prominent presence of the US Treasury as soon as the Fed started taking significant amounts of illiquid assets with positive but unknown credit risk attached to them onto its balance sheet, back in August 2007.

The US Treasury instead appears desperate to hide the reality that the US tax payer is at risk when MBS are accepted as collateral or bought outright by the Fed or indeed by Fannie Mae and Freddie Mac. This is why rather than the Treasury creating auction mechanism to enable to Treasury itself to acquire MBS, we get the announcement by the Office of Federal Housing Enterprise Oversight last Wednesday March 19, 2008, that it has cut the surplus capital requirement of these two government-sponsored entities to 20 percent from 30 percent. The OFHEO estimates that this reduction, in combination with the release of portfolio caps announced last month, should provide up to $200 billion of immediate liquidity to the MBS market, and would allow Fannie Mae and Freddie Mac to purchase or guarantee about $2 trillion in mortgages this year.

I believe that, provided the Fed and the GSEs value the assets they accept as collateral, purchase or guarantee conservatively, it is entirely right for these public sector entities to take this additional credit risk. But it has to be clear that, even if the default risk on the MBS is priced appropriately by the Fed, Fannie and Freddie, an unexpectedly large number of defaults may well occur. Should some or all of this default risk materialise, these capital losses are the baby not of the Fed, Fannie or Freddie, but of the Treasury, that is, the US tax payer.

Hiding behind the Fed, Fannie and Freddie and hoping that the fiscal exposure inherent in their actions as market makers of last resort and lenders of last resort in the MBS markets may never turn into realised losses but, with a bit of luck, will just go away, is classic quasi-fiscal window dressing and cowardice.

The Euro Area is in the unusual position that there is no Treasury backing the ECB. The fifteen national central banks (NCBs) are no doubt backed by their national Treasuries, but the EU has no budget to speak off (just over 1 percent of EU GDP), no power to tax and no power to borrow (note that the ECB is owned by all 27 EU NCBs, not just by the 15 Euro Area NCBs). We don’t yet have banks incorporated under EU law in the Euro Area, but no doubt that will come. In the mean time, we can only hope that when a systemically important bank in, say, Italy threatens to go belly up, but this bank is wholly owned by one or more other banks not registered in Italy but elsewhere in the EU, the fiscal authorities of Italy will be happy to pick up 100 percent of the tab for a bail-out.

This problem arises of course more generally whenever there is significant cross-border banking activity and shareholders need not be resident of or registered in any of the countries of operation of the bank. It affects US-owned subsidiaries as much as Euro Area banks. The problem is likely to get more acute in the Euro Area, however, because of the lack of any fiscal back-up for the ECB. Let’s just hope there is no major bank failure in the Euro Area, because the Eurosystem is certainly not ready to manage the fiscal fallout.


The following ten commandments are not as impressive as the original set, but they will have to do:

  1. Central banks should accept any private securities rated at least single A as collateral in repos.
  2. Central banks should aggressively hunt liquidity risk spreads at all maturities from 1-month to 12-month.
  3. Central banks should purchase outright a wide range of systemically important private securities rates at least single A.
  4. Central banks should accept any investment grade private securities as collateral at the discount window.
  5. Central banks should lend at the discount window for maturities up to 90 days.
  6. The range of eligible borrowers at the discount window should be extended to all highly leveraged financial institutions that are (a) listed and (b) regulated to the satisfaction of the central bank.
  7. The penalty rate for overnight borrowing at the discount window should not be less than 100 basis points.
  8. Private securities purchased outright or accepted as collateral at the discount window or in repos should be valued aggressively, to make sure the central bank and the tax payer are adequately compensated for the default risk the central bank takes on its balance sheet.
  9. A special resolution regime analogous to that for deposit taking institutions (administered by the FDIC) should be created for all systemically important highly leveraged institutions.
  10. The Treasury has to be ready to recapitalise the central bank if the central bank makes losses because of its exposure to risky private securities.

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