Better Auctions
and
Better Products

Presented at the UK DMO
Friday 6^th February 2009

www.jdawiseman.com/papers/finmkts/20090206_dmo.html

The price is how ‘good’?

• Hitherto, just before an auction, there has been a self-fulfilling belief that the market price is good in enough billions for an auction to clear near that price.

• For German 10s, this belief has gone.

Germany: unsold proportion

Chart showing ‘Amount set aside for secondary market operations’ ÷ ‘Issue volume’.

Observe that German 10s have fallen into a hole—and gilts might too.

Three Possibilities

• There are only three possibilities:

  • Don’t fall in the hole (= proactive);

  • Climb out of the hole (= reactive);

  • Stay in the hole (= pessimal).

• The Deutsche Finanzagentur is in the hole (= belief that prior market price won’t clear).

• UK DMO can choose a better plan:

  • know that the hole beckons.

Failure of auction mechanism

• Commentators: “too little demand”.

• But if those Bunds had been for sale at a fixed price, say €90, there would have been plenty of demand.

• Rather, the auction mechanism failed to find the price at which there was €6bn of demand.

Why failure to find price?

• Auctions sell stuff, and provide information about the demand.

• But that information, if available to bidders, helps them learn the price.

• Gilt auction reveals that info too late.

• Christies / Sothebys / Bonhams / etc use English auctions, as bids inform bidders. But wrong for divisible goods.

How fix?

• So bidders should see some measure of demand before auction finished.

• That entails splitting a £4bn auction, so that early pieces inform later bidding.

• There are messy possibilities:

  • Tap £200mn every day for a month? Yuck.

  • Unpredictable discretionary taps? Yuck.

How to fix, as an auction?

• Keeping the auction concept:

  • Split a £4bn auction into 40 ‘auctionettes’;

  • Hold auctionettes one minute apart;

  • Too fast for human intervention by seller:

    • must be totally automatic;

    • so there must be a minimum price;

  • Say, for a 10-year, min = £0.08 cheaper than clearing price of previous auctionette.

What to reveal about bids?

• Don’t punish a lone bidder!

  • Don’t reveal total number of bids.

  • Don’t reveal total quantity of bids.

  • Don’t reveal gap average–clearing, hence each auctionette uniform price.

• Be nice not nasty to a lone bidder (unlike current auction system).

What to reveal about bids?

• About this auctionette:

  • The amount sold;

  • The uniform clearing price;

  • Prop’n of clearing-price bids filled = scaledown;

• And about the next auctionette:

  • The minimum price;

  • And the (typically unchanged) amount for sale.

• Not the number or quantity of bids, nor the average of their prices (whether of all or of accepted). Be nice to a lone bidder.

Soft pressure

• Optionally, DMO could state that “Wholesale GEMMs are expected to bid for all of most auctionettes”.

• That doesn’t punish if something dramatic happens, but sets an expectation.

A GEMM’s Strategy

• Previous auctionette cleared at 100.50
  Next £100mn, minimum price 100.42.

• Needs £5mn from each (£200mn total), and wants another £10mn from each (£400mn).

A GEMM’s Strategy

• Optimal strategy? Something like:

  • Bid ≥101 for £5mn;

  • Bid 100.51 for £10mn;

  • Bid 100.49 for £25mn;

  • Bid 100.46 for £30mn;

  • Bid 100.43 for £30mn.

• Even if only bidder, not a bad outcome. If others bid for only £50mn, clearing price 100.46, at which ≈£50mn bought. Also OK.

Details: Deemed Bids

• ∃n GEMMs; size of auctionette = £x;

• Each GEMM not bidding for ≥x/n deemed to bid for deficit at min price;

• Notate bene: NOT additional. Don’t punish a lone bidder! If a lone bidder bids for £10mn, that bidder bids for £10mn, not for £10mn+x/n.

Details: Minimum Price

• Minimum price cannot rise by more than 1×DV01 at each auctionette:

  • A player or players might make a mistake, bidding far too high;

  • Restricting the maximum rise to the same 1×DV01 prevents, in the next auctionette, the deemed bids punishing the innocent by more than a trifle.

Details: Squeezes

• Squeeze game: buy the bond at a price of fair+ε, becoming monopoly owner.

• Impossible with auctionettes.

  • Squeezer buys first few auctionettes;

  • Other bidders respond, price rises;

  • Squeeze costs huge, ≥ monopoly profits.

• But DMO might be redundantly cautious: could superimpose a y% rule.

Auctionettes: Q & A

‘Q’ first; ‘A’ second.

www.jdawiseman.com/papers/finmkts/20090206_dmo.html

Selling calls on long gilts

• Selling calls means the DMO sells gilts into a rising market;

• Perfect zero-discretion transparency;

• What Plenderleithian taps should be;

• Balance-sheet efficient: £2 not £100;

• Promote gilt rather than swap options;

• High implied means HMT will be paid.

Details: Underlying

• A gilt!

• Conventional gilts with maturity much much longer than option expiry. So any gilt with ≥3 years to go;

• As long gilts can and should be much larger, the 4¼ Dec 2055;.

• ILGs later, or never.

Details: Expiry

• Shorter gives more certainty about funding;

• Options always outstanding (ideally several dates), so very short implies very high turnover;

• Compromise: 28 day options:

  • Well-understood sort of horizon;

  • Weekly auctions ⇒ 4 sets outstanding.

Details: Strikes

• Want to sell optionality, so strikes near the money;

• Just ATM strikes would be acceptable;

• But even split of 25%, 50% and 75% δs:

  • Slightly reduces variability of funding;

  • Ensures delta-hedgers stabilise market over good range of prices.

Details: Structure

• OTC?

  • Not transferable.

• ‘Gilt’ with pay-to-exchange rights?

  • At expiry DMO’s counterparty could be anybody, anything, or receiver thereof.

• Listed pay-by-M2M option?

  • Best, but DMO will often be paying margin before receiving premium+strike.

Scale

• Imagine that every week £500mn of each of 25%, 50% and 75% are sold.

• Outstanding would be four series each of three strikes, totalling £6bn nominal. During a year £78bn, being an expected £39bn of funding + ≈£2bn of premium.

• Given size of funding need, about right.

Options as Taps

• Selling call options:

  • The transparent way to do ‘taps’;

  • Slightly market stabilising;

  • HMT should breakeven or better relative to sales of underlying;

  • A little uncertainty about achieved pace of funding, but with 28-day expiries, only little.

Options: Q & A

‘Q’ first; ‘A’ second.

www.jdawiseman.com/papers/finmkts/20090206_dmo.html

— Julian D. A. Wiseman
New York, February 2009
www.jdawiseman.com