Undercover Economist

Why banks are going to auction

In 1873, Walter Bagehot famously argued that in a banking crisis, the Bank of England should be willing to lend “freely and readily” in exchange for good collateral. This view seemed quaint five years ago, when international capital markets were willing to pay cash for the most illiquid-seeming assets, but it quickly became relevant during the credit crunch.

Yet it is a challenge to support banks that are temporarily embarrassed by a lack of liquidity, without bailing out bankrupt banks or relieving them of all responsibility for their own liquidity management.

Paul Klemperer – the Oxford University theorist behind the auctions for “3G” mobile phone licences 11 years ago, and less gloriously my thesis supervisor – has been working with the Bank of England to design an auction to solve the problem. The design exploits a long tradition in auctions: that of proxy bidding.

In a standard auction, a proxy bid of, say, £260 for a case of champagne means that the buyer is happy to outbid all-comers until the bidding reaches £260.

The Klemperer auction design allows bidders to submit multiple proxy bids for different kinds of object. Imagine an auction in which both vintage and non-vintage champagne is available, and several cases of each are on offer. A package of bids might include one of £250 for a case of non-vintage champagne, plus a “paired” bid of either £300 for vintage champagne or £200 for non-vintage, plus a third, bottom-feeding, bid of £150 for a case of vintage or £100 for non-vintage, useful if prices are low. Depending on how prices in the auction work out, this package of bids might win up to three cases of champagne, and the quality would vary depending on which kind of champagne was more in demand.

Although it may need a computer to figure out exactly who gets what, this basic idea is not too troubling, and the multiple proxy bids enable a bidder to specify, quite intuitively, how much booze they’d like to buy, and what sort of price-differential would tempt them to favour one type of champagne over another.

This is roughly what the Bank of England is now doing to supply liquidity to banks. The banks bid different interest rates to borrow money from the Bank, and offer either higher- or lower-quality collateral, depending on the interest rate premium required for the lower-quality stuff.

(It would be possible to run two separate auctions or combine the two auctions into one, with some fixed premium for low-quality collateral. But the first alternative dilutes the competitiveness of the auctions, while the second will malfunction if the Bank misjudges the level of the premium.)

The Bank has been running these auctions every month since last summer, and seems pleased with the innovation. The auctions may seem fiddly, but “all the complexity lies on our side”, says Chris Salmon, an executive director and chief cashier at the Bank of England. While they haven’t yet been tested under stress, they offer something quite attractive: a flexible tool, ticking away in the background, with the ability to prevent old-fashioned, self-fulfilling bank runs.

Klemperer, meanwhile, has his eye on wider horizons. His auction design could, in principle, be used to buy – or sell – dozens, even hundreds, of substitute products at the same time.

This is an idea with many uses. It could be used to run electricity networks by selling intermittent power and baseload power in the same auction, or to sell planning permission on green-field and brown-field sites – a prospect to gladden George Osborne’s heart. And in a future banking crisis, it could be used by a government wishing to establish a proper price for toxic assets. The Bank of England, I suspect, is hoping that particular use remains theoretical for many decades yet.

Also published at ft.com.