Dutch Sequential Auctions

Julian D. A. Wiseman

Abstract: In February 1999 the Dutch State Treasury Agency consulted Primary Dealers about an auction mechanism, the Dutch Sequential Auction, which was inspired by the ‘auctionette’ mechanism. The following summarises that suggested mechanism, and the objections to it raised in a letter sent by the author on 25th February 1999. Usual disclaimer and copyright terms apply.

Contents: Proposed Dutch Sequential Auctions, Reply

In February 1999 the Dutch State Treasury Agency (which is responsible for the management of the national debt of the Netherlands) sent an email to the Primary Dealers, suggesting a new auction mechanism. The following quotes parts and paraphrases parts of that email.

… The DSTA has traditionally an open mind to new issuing techniques, such as the uniform price tender (“Dutch auction”), now applied in several countries, and more recently the “tap”. The tap has served well in the Dutch market for many years. Relative to the usual auction system, it has proved to have important advantages. Among others, unhedgeable auction risk could be avoided. Also, the spreading out over time has made involvement of end investors easier. For the issuer, it reduced the dependency on market circumstances in a fraction of time. However, logistical limitations and diminished transparency in the issuance and pricing process have recently shown to be some of the drawbacks.

The DSTA is considering to adopt a totally new issuance technique, that combines the advantages of the tap and auction system. The Dutch Sequential Auction is a process where the total volume is split up in parts and issued in a sequence of quick and small auctions (auctionettes, if you like) within a period of 1 to 1.5 hours.

Obviously, this technique could be function ideally in an electronic version that would secure a quick adjudication process. First, however, we wish to build some experience with this technique in the more traditional way of submitting bids by telephone.

With the next issuance of the 10 year DSL, on March 9th, the Dutch Sequential Auction will therefore be applied for the first time. The DSTA plans three auctionettes, according to the following time table:

1. from 10.00 to 10.15am bids can be submitted for the first auctionette;
   results around 10.20am;
   
2. 10.30 to 10.45am bids can be submitted for the second auctionette;
   results around 10.50am;
   
3. 11.00 to 11.15am bids can be submitted for third and last auctionette;
   results around 11.20am.

To make the process manageable, each party is allowed to submit only four bids in each auctionette, with minimum subscription size of euro 25 million and a maximum of the total auctionette amount per bid. This serves to assure fair and orderly allocation. Non-competitive bids are allowed. Although the submission of bids by phone is binding, immediate confirmation by Bloomberg mail is welcomed.

The author interpreted the proposed ‘Dutch Sequential Auction’ as step towards the ‘auctionette’ mechanism; a step designed to avoid the necessity to create of an electronic trading platform.

The following reply was sent on 25th February 1999.

1. My paper on the ‘auctionette’ system was originally written with the UK in mind. Almost fours years after I first discussed this with the Bank of England, I stand by this recommended system, unaltered, for the UK.

2. The UK government debt market has one particular feature not shared by some of the euro government debt markets, including the Dutch. In the UK, a government auction dominates trading activity, and crowds out other trading. So, in the UK, traders are willing to dedicate significant time exclusively to a government auction. Hence my original suggestion that a £2.5 billion gilt auction should be split into 50 auctionettes each of £50 million.

3. But in the euro-denominated government market, the Dutch are ‘merely’ one government amongst many, in a market much larger than the Dutch alone. Because the Dutch do not dominate trading, an individual DSL auction needs to be finished quickly; but because the euro market is much larger than the sterling market, each euro auctionette can be larger. My estimate is that, in principle, a size of about EUR 250 million is probably appropriate, this being about 3 times larger than recommended for the UK government.

4. However, recall that each auctionette has a minimum price, to prevent random windfall gains. For the second and subsequent auctionettes, the minimum price is set with reference to the previous auctionette (I suggested the equivalent of +1bp higher in yield). But the minimum price of the first auctionette is set ‘manually’ with reference to the observed price in the secondary market. Allowing this manually-set minimum to apply to a large auctionette is a retrograde step: the objective is that the market should determine the price to as great an extent as possible.

5. So, my advice is that a EUR 2500 million DSL auction be split into 11 auctionettes: the first of EUR 100 million, the second of EUR 150 million, and the remaining nine of EUR 250 million each. This reflects a judgement about the balance of auction risk, the limitations on traders’ time, and the minimisation of the authorities’ role in price determination.

6. Other than that, the ‘auctionettes’ mechanism is appropriate to the Dutch, without further alteration. In particular, I emphasise that great care should be taken in deciding what information to reveal after each auctionette: see ¶6 of my most recent paper on this subject.

7. The DSTA’s proposed variant, of four auctionettes, conducted by telephone, 15+ minutes apart, concerns me somewhat. If the auctionettes’ results are published immediately, and are closely packed (60 seconds apart), then each would provide pricing information to the next. I believe that your variant is too slow and too discontinuous, and dealers will regard the delay in the publication of the results as a negative step. The 15 minute gap means that previous auction result will have become stale: participation in such an auction may prove to be too much effort and time for a trader. I counsel against.

A few days after a telephone conference between the Primary Dealers and the Dutch State Treasury Agency, the DSTA commented by email:

The telephone conference of last Tuesday has been a good and serious example of the dialogue the Dutch State Treasury Agency wishes to attain with the Primary Dealers.

The DSTA has observed overwhelming support for the present tap issuance method. The commonly held view is that some minor improvements should be sufficient and that logistical flaws are acceptable. We are satisfied with this outcome.

With a view to the development of electronic trading, innovation in issuance techniques will continue to have our attention. We welcome any constructive contribution from primary dealers in this field. Any suggestions should obviously improve and give positive incentives to the functioning of the primary market.

The DSTA has decided that the plans for the Dutch Sequential Auction will remain on the drawing board.