December 15, 2007
Good news! Central banks don’t have to be smart to run auctions for illiquid securities without becoming moral hazard patsies
I have written quite extensively, starting in August 2007, with much of it appearing in this blog, about how the central bank can make a market in illiquid securities for which there is no market price or any other verifiable benchmark available, without providing a subsidy to the private sellers of these securities. It’s not hard, and does not require the central bank to know anything more than anyone else about the fair value of the security. Indeed, it does not require the central bank to know anything at all - which is probably just as well.
In response to comments from Martin Wolf on an earlier blog of mine I will here just give a brief summary of these earlier scribblings on how central banks can organise auctions for loans against against illiquid collateral for which there isn’t a market price, without rewarding reckless lending and borrowing and thus encouraging a repeat of past excesses in the provision of credit.
Well-designed auctions will act as reservation price-revealing mechanisms for the central bank. The simplest auction is a reverse Dutch auction with the central bank as the only buyer. The central bank has to have a list of securities eligible for access to the auction. I would recommend the central bank only include simple structured investment products on the list, to provide incentives for private financial innovators to keep their wilder horses under control. It could also restrict access to the auction to sellers that are subject to a regulatory regime approved by the central bank. Unregulated financial institutions would have to try their luck with those regulated institutions that were successful obtaining liquidity from the auctions.
The central bank would have to decide for each auction an upper bound on the amount it is willing to purchase at the auction. It could do so by setting an upper bound either on the market value or on the notional or face value of the securities it is willing to purchases. Assume for the sake of argument it sets an upper limit of £10bn for the face value of the securities it is willing to buy. The central bank starts by offering to buy any amount up to £10bn at the lowest possible price, say, 1 penny for each pound sterling of face value. Assume £2bn worth of face value is sold at 1 penny. Next it offers to pay 2 pennies for each pound sterling worth of face value for up to £8bn worth of securities. If the cumulative sales at 1 penny and 2 pennies don’t add of to £10bn, there will at least be a third round, say, at three pennies for each pound sterling of face value. The auction continues until the pre-set upper limit, £10bn is reached, or until the price reaches 1 pound sterling for 1 pound sterling worth of face value.
In such a reverse Dutch auction, those desperately short of liquidity will offer to sell first, at very low prices. The less panicked would-be sellers hold out for higher prices, but risk missing out altogether.
The central bank would take the securities it had bought at the auction onto its balance sheet. If the markets for the securities bought at the auction by the central bank were to normalise later, the central bank could opt to sell the securities at that time. There is, however, no need to do so. The central bank can simply hold all the securities it buys at the auction until maturity. That way it never has to form a view on what they really are worth. There are advantages to never being liquidity constrained – the happy condition of the central bank.
The central bank would, of course, take credit risk onto its balance sheet if it buys private securities at the auction. The reverse Dutch auction with a monopolistic buyer is, however, so hard on the sellers, that the central bank could expect to make a profit out of the activity. Should it be hit by an unexpected wave of defaults on these securities, it would have to be recapitalised (bailed out) by the Treasury. This, of course, is also the situation the Bank of England faces today with its exposure to Northern Rock through the Liquidity Support Facility.
To encourage those private financial institutions that do not wish to mark to market their illiquid assets to participate in the auction, it could be made a legal or regulatory requirement that all securities for which auctions are organised, even those not offered for sale, be marked-to-market at the prices established in these auctions (in the case of the reverse Dutch auction, you could use the average price, say, for this).
The Fed, in its Term Auction Facility, could extend the range of eligible collateral further. There is no reason why illiquid junk could not be auctioned and bought by the Fed using the mechanism outlined here. There are many other kinds of auctions with desirable properties, that don’t require the auctioneer (or the monopolistic buyer), to know much if anything about the fundamental value of the securities that are being auctioned. Economists like Paul Milgrom in the US, or Paul Klemperer and Ken Binmore in the UK, should be able to get a suitable set of auctions up and running in no time.
So, no, the central bank does not have to establish what a competitive market price would be. It does not have to buy at a price above ‘the market price’. My reverse Dutch auction will generically not have a single market price. Securities will be bought at different prices, starting at the lowest. In its purest form, this would be a perfect price discrimination mechanism that creams off all the surplus over the seller’s reservation price for the monopolistic buyer. It would be tough on the sellers, but it would be efficient. There would not be even the faintest whiff of moral hazard in the air.
What it takes to make these auctions successful is not a central bank that knows more than the private sector about the fundamental value of the securities that are being auctioned. There need never to have been a market for the security in question – they could be (standardized) OTC instruments. The central bank could be a bear of very little brain. All it needs is deep pockets and the ability and willingness to take the long view. Those do not seem of to be unreasonable demands to place on the central bank.
I can immediately think of a couple of practical problems with this proposal. First, the face or nominal value of securities is relatively clear, but would not be useful for these purposes, because their market value depends on maturity as well as credit quality and liquidity. If the central bank used face value in such an auction, it would tend to get offered short maturity but highly toxic securities. So the auction would have to be conducted in terms of market value, which is of course unknown. Second, an auction depends on competition, but part of the problem with these “structured products” is that they are highly idiosyncratic. There may be only one holder of any particular issue, especially among the narrow set of banks with access to the central bank. If, as you suggest, the central bank restricts these auctions to a limited list of simple securities, they will contribute little to solving the problem.
I have an alternative idea. We are told that banks are reluctant to lend to each other because they doubt their counterparties’ creditworthiness. That rings true. One reason for the securities’ idiosyncratic nature mentioned above is that it allowed the investment banks to construct securities that could be sold for more than the sum of their parts. Since they all did this, the investment banks know that any of their number that have been forced to take such securities back may well have lower net worth than the accounting values recorded on their balance sheet would suggest. What is needed most, therefore, is not central bank liquidity, but capital to restore confidence. My proposal is therefore that the investment banks should agree to pay all dividends and staff bonuses due in the next year or so in stock, not cash. Given the size of bonuses that these supposedly stressed organisations are reportedly intending to pay in the next few weeks, this measure should be rapidly effective.
Posted by: Tim Young | December 15th, 2007 at 2:51 pm | Report this commentTim, your first objection is due to a misunderstanding about my statement: “It could do so by setting an upper bound either on the market value or on the notional or face value of the securities it is willing to purchases”. This just means that the central bank could either set an upper limit on the aggregate size of the auction in terms of an upper limit on the aggregate notional value of the collateral it would acccept, or in terms of an upper limit on the on the aggregate market value (as established at the auction) it would accept. For brevity’s sake, I only elaborated on the case where an upper limit on the aggregate face value is set. In this case, even if the constraint ends up binding, we will only know what the value of the loans extended is if we know how each bit of collateral purchased by the central bank ended up being priced in the auction.
Let’s assume that instead an upper bound is set of £10bn for the value of the loans, rather than for the notional value of the collateral. Assume the auction is successful and the constraint ends up binding. We know that £10bn worth of loans have been auctioned off by the central bank. But we will only know how much notional value of collateral will have had to be offered as collateral for these loans, by adding up, once again, all the collateral successfully offered (at a range of prices ) in the auction.
The authorities don’t have to know in advance what the price is going to be. They will simply halt the auction once, at the rising sequence of prices established in the auction, the cumulative auction price value of all the collateral successfully offered exceeds £10bn.
To the extent that capital adequacy is a problem rather than just illiquidity, some way of raising or retaining capital is required. Not paying a dividend is one of these ways. Paying a given amount of dividends in stock rather than cash would not help.
Posted by: Willem H. Buiter | December 16th, 2007 at 2:16 am | Report this commentI am sorry that you have not had more comments on this subject, Willem!
The key point that is not clear here is what the auction terms are, and hence how the procedure ensures that the central bank is not getting ripped off. As written, the central bank seems to be inviting offers according to price per unit of face value - eg it offers to pay 2p per pound of face value. Presumably that means it would get offered securities with the market value most below face value. Since it is easy to recognise and avoid securities that have low market value because of long duration, it seems to me that the central bank would end up with the least creditworthy and most illiquid.
You did not answer my point about the highly idiosyncratic nature of the securities that are the problem - the value of big, widely held issues is not in question. For example, according to Larry Tabb in last Tuesday’s FT, there are more than 2,600,000 mortgage backed securities in existence. I doubt that the central bank would be protected by competitive supply.
If you are saying that paying a dividend in stock would not help any more than not paying a cash dividend, I agree. It is just a presentational device which indicates the size of the dividend foregone.
Posted by: Tim Young | December 18th, 2007 at 11:14 am | Report this commentMight it be in the interest of many hedge funds, institutional investors, and even banks to not have any price for valuing these structured finance securites other than the one their own models suggest is most relevant? Or to put it bluntly, if you are concerned a MMLR operation by a central bank might reveal you are insolvent with respect to you remaining holdings, why participate in the MMLR auction? Tim Young has it right, I suspect. This is a solvency/recapitalization issue, much more than it is a liquidity/price discovery issue. It may serve many interests for these structured finance vehicles to remain, literally, priceless.
Posted by: Rob Parenteau | December 18th, 2007 at 6:32 pm | Report this comment