|Home > Research > Research Publications|
Current Issues in Economics and Finance
The Treasury Auction Process: Objectives, Structure, and Recent Adaptations
|February 2005 Volume 11, Number 2||
|JEL classification: G18, G28, H63||
Authors: Kenneth D. Garbade and Jeffrey F. Ingber
Treasury auctions are designed to minimize the cost of financing the national debt by promoting broad, competitive bidding and liquid secondary market trading. A review of the auction process—from the announcement of a new issue to the delivery of securities—reveals how these objectives have been met. Also highlighted are changes in the auction process that stem from recent advances in information-processing technologies and risk management techniques.
The stated goal of Treasury debt management is to meet the financing needs of the federal government at the lowest cost over time.1 Since Treasury auctions provide the principal means of financing the federal deficit and refinancing maturing debt, Treasury officials have sought to structure the auction process to minimize the government's costs, both directly by promoting broad, competitive bidding and indirectly by promoting liquid post-auction secondary markets for new issues.2
This edition of Current Issues offers a concise explanation of the modern Treasury auction process. We follow the process from the announcement of a new issue to the delivery of the new securities, looking at key aspects of the process such as the form of bidding and restrictions on tenders and awards.
We also consider why, and how, the auction process has changed in recent decades. Treasury bills were first auctioned in 1929; coupon-bearing securities, in 1970. Given this lengthy history, an observer might reasonably hypothesize that an efficiently administered auction process would have long since converged to some optimal form. Surprisingly, however, the process continues to evolve. Understanding the nature of, and reasons for, this persistent change is important for both public officials and private market participants. We suggest that it reflects adaptations to new developments in the financial services industry, such as futures trading and net settlement, and advances in communications and information-processing technologies.
The Treasury auctions an astonishing quantity of securities. In calendar year 2003, it auctioned $3.42 trillion of securities, including $2.78 trillion face amount of bills, $616 billion principal amount of notes, and $26 billion indexed principal amount of TIPS, in a total of 202 auctions. Most of the auction proceeds went to redeem maturing issues ($2.83 trillion) or to bridge short-term gaps between expenditures and receipts ($229 billion), but $363 billion was new money.
Marketable Treasury securities come in two forms: entries in book-entry accounts at Federal Reserve Banks and entries in book-entry accounts in TreasuryDirect. The Federal Reserve book-entry securities system is operated by the Federal Reserve District Banks in their own capacity and as fiscal agents for the United States. The system maintains the primary records of bill, note, and TIPS holdings of depository institutions and certain other entities. (Depository institutions hold book-entry securities on behalf of their customers as well as for their own account.) TreasuryDirect is a proprietary Treasury book-entry system introduced in 1986 to accommodate retail investors that buy securities at original issue and typically hold them to maturity.
The Auction Announcement and When-Issued Trading
Immediately following the announcement of a forthcoming auction, dealers and other market participants begin to trade the new security on a when-issued basis. Secondary market transactions in outstanding Treasury securities typically settle on the business day after the trade date, when sellers deliver securities to buyers and receive payment. When-issued transactions, by contrast, settle on the issue date of the new security (which can be as much as a week or more after a trade is negotiated) because the security is not available for delivery at any earlier date.
When-issued trading enables market participants to contract for the purchase and sale of a new security before the security has been auctioned. This type of trading is important because public dissemination of the yield at which a new note is trading, or the discount rate at which a new bill is trading, provides valuable information about the market's appraisal of the prospective value of the security. The Joint Report on the Government Securities Market prepared by the U.S. Department of the Treasury, the Securities and Exchange Commission, and the Board of Governors of the Federal Reserve System pointed out that when-issued trading “reduces uncertainties surrounding Treasury auctions by serving as a price discovery mechanism. Potential . . . bidders look to when-issued trading levels as a market gauge of demand in determining how to bid at an auction” (1992, p. A-6). When-issued trading thus contributes to the Treasury's goal of promoting competitive auctions by enhancing market transparency.
A particularly important part of pre-auction when-issued trading involves purchases by private investors from dealers.3 The purchases have two significant distributional effects. First, they facilitate distribution of a new issue ahead of its auction. The Joint Report noted that when-issued trading “benefits the Treasury by . . . stretching out the actual distribution period for each issue, . . . allowing the market more time to absorb large issues without disruption” (p. 9). Second, when-issued sales by dealers to private investors leave the dealers with a need to make offsetting purchases in the auction, concentrating bidding interest in the hands of market participants that have a substantial financial incentive to identify correctly the price that balances demand with supply.
The Treasury conducts note auctions in a “single-price” format. After the close of bidding, it subtracts the noncompetitive bids from the total quantity of securities offered and then accepts competitive bids, in order of increasing yield, until it has exhausted the offering. The highest accepted yield is called the “stop.” Bids specifying yields below the stop are filled in full, bids above the stop are rejected, and bids at the stop are filled on a pro rata basis.6 All auction awards are made at a single price, computed from the yield at which the auction stopped. If the note does not reopen an outstanding note, the coupon rate on the note is set at the highest level, in increments of 1/8 percent, that does not result in a price greater than 100 percent of principal (see the example in Box 1). Bill and TIPS auctions are similar, except that competitive bids for bills are specified in terms of discount rates rather than yields and there is no coupon rate on a bill.
The 1990s witnessed several important changes in auction structure, including a change from a multiple-price format (in which successful competitive bidders paid prices computed from their own bid yields rather than from the stop), a significant elaboration of restrictions on auction awards, and an increase in the arithmetic precision with which bids are expressed. Each of the changes was motivated by the Treasury's interest in promoting competitive bidding and liquid secondary markets.
Auction format. The best-known feature of a Treasury auction—the single-price format—was introduced in 1992. The change from the earlier multiple-price format was part of a major overhaul of the auction process that followed several violations of auction rules in 1991.
The Treasury first adopted the multiple-price format when it initiated bill auctions in 1929 and it continued to use that format when it introduced auctions of coupon-bearing securities in the early 1970s. However, when the auction process came under scrutiny in 1991, public officials became interested in alternative formats that might appeal to more investors and that might lead to lower financing costs. Several academics had suggested earlier that single-price auctions might reduce financing costs (see Carson , Friedman [1960, 1963], and Smith ). In a single-price auction, a participant can bid its actual reservation yield for a new security, that is, the minimum yield at which it is willing to buy the security. The bidder certainly has no reason to bid a lower yield, but if the auction stops at a higher yield it will get the full benefit of buying at that higher yield. In contrast, the multiple-price format encourages a participant to bid higher than its reservation yield in hopes of getting the security on more favorable terms.
Whether the Treasury would be better off selling securities in a single-price format or a multiple-price format was a matter that could only be resolved by empirical analysis.7 In September 1992, the Treasury announced that, in an experiment, it would begin to auction two-year notes and five-year notes in the single-price format. It subsequently produced two empirical studies analyzing the results of the experiment (see Box 2). Although the evidence was not unambiguous, the Treasury decided in October 1998 that it justified extending the single-price format to all auction offerings.
Restrictions on auction awards to competitive bidders. In the interest of fostering a liquid post-auction secondary market for a new issue, the Treasury limits the maximum auction award to a single bidder to 35 percent of the offering, less the bidder's “reportable net long position” in the security. A bidder's net long position is the sum, as of one-half hour before the close of bidding, of
A bidder must report its net long position along with its auction bids if the sum of its net long position and its bids exceeds 35 percent of the offering.
Limitations on auction awards go back more than forty years. A limitation was first imposed following an auction of thirteen-week bills in August 1962, when Morgan Guaranty Trust Company bid for half of the bills offered. To avoid a market “disruption,” Secretary of the Treasury Douglas Dillon exercised his right to reject any tender in whole or in part and reduced Morgan's award to 25 percent of the amount offered. He stated that, going forward, no bidder would be awarded more than 25 percent of a bill offering.9 Limitations on auction awards were subsequently understood to apply to all offerings of marketable Treasury securities.
Limitations on auction awards have been modified several times since 1962 in response to innovations such as futures trading and STRIPS. Box 3 contains a summary of the modifications.
Limitations on the size of a bid. Although the maximum award to an individual bidder has been limited since 1962, before 1990 there was no limit on the size of a bid. Because bids at the stop are filled on a pro rata basis, bidders could get larger awards (subject to the 35 percent limitation) if they bid (at what turned out to be the stop) for more securities. For example, if a dealer wanted $100 million of a new issue at a yield equal to what the dealer expected would be the stop, the dealer might bid for $200 million if it expected a 50 percent allotment. Such “strategic” bidding led to individual bids for more securities than were offered in an auction of four-year notes in June 1990.10
Strategic bidding by some large auction market participants made competitive bidding riskier and more complicated for other participants. Bidders had to anticipate the likely volume of overbidding; if the expected overbidding did not materialize, they would end up owning more securities than they wanted. In the interest of encouraging broad public participation in its auctions, the Treasury announced in July 1990 that it would limit the total bids by a given participant at a given yield or discount rate to 35 percent of the amount offered to the public.
Granularity of bidding. When the Treasury introduced bidding on notes and bonds in terms of yields in 1974, it specified that bids should be expressed to a whole basis point.11 When it introduced bidding on bills in terms of discount rates in 1983, it made a similar stipulation.12 Bidding to a whole basis point continued until 1995, when Treasury officials specified that bids on notes and bonds should be expressed to 1/10 of 1 basis point. The greater precision was intended to “increase participation in Treasury auctions and to conform the auctions to market practice for when-issued trading.”13 Officials refined the bidding increment to 1/2 of 1 basis point in late 1997 for thirteen-, twenty-six-, and fifty-two-week bills and in April 2002 for cash management bills, saying that they expected the change “to promote more efficient and aggressive bidding and lead to marginally higher revenue.” 14
Auction market participants submit bids through a communications system called TAAPSLink®. Institutions other than primary dealers15 (including depository institutions, other dealers, and institutional investors) use an Internet version called TAAPSLink v1. Primary dealers—which submit the largest volume of bids in almost every auction—use an alternative version called TAAPSLink v2. Retail investors with TreasuryDirect accounts submit bids by mail, telephone, and Internet applications that ultimately reach TAAPS through TAAPSLink v1.
The last decade witnessed striking advances in bid submission, bid processing, and announcement of auction results. Until 1993, bids were submitted on paper forms by mail or in person at the Treasury Department in Washington, D.C., or at a Federal Reserve Bank or Branch. Dealers had a lot to lose if they tendered bids early at prices that failed to reflect a late-developing rally or market pullback, so on auction days they stationed employees in the lobby of the Federal Reserve Bank of New York and relayed bidding instructions over the telephone immediately before the close of bidding. Other large bidders submitted bids through primary dealers acting on their behalf. Bids were processed manually and announcements of auction results frequently came out in mid or late afternoon.16 These procedures left auction market participants uncertain—for substantial intervals of time—whether they would be awarded securities. The uncertainty may have led them to enter bids at higher yields than they would have if auction results had been announced more promptly.
The introduction of electronic bid submission and bid processing in the 1990s was stimulated by the 1991 violations of Treasury auction rules noted earlier and resulting demands for a faster, more transparent, and more accessible auction process.17 Electronic processing dramatically reduced the time between the close of bidding and the announcement of results, thus materially reducing bidder risk exposure. In early 2002, the Treasury announced that it intended to release auction results consistently within two minutes of the close of bidding, and in mid-2003 it achieved that goal.18 Electronic bid submission was a necessary precondition to electronic processing. It also made it operationally feasible for auction market participants other than primary dealers to bid directly (rather than through a dealer) right up to the close of competitive bidding—auction access that the Treasury believes can help maximize the breadth of the auction market.19
Delivery and Settlement
Direct deliveries. The simplest example of a direct delivery occurs when a bidder is a depository institution and requests that awards be credited to its account in the Federal Reserve book-entry system. At 9:15 a.m. on the issue date, the Fed, acting as fiscal agent for the Treasury, credits the institution's account for the new securities, debits the institution's reserve account for the cost of the securities, and transfers the payment to a Treasury account at the Fed.
If an institutional bidder is not a depository institution, it will usually request that securities be delivered pursuant to an “autocharge” agreement with a depository institution that does a custodial business. The agreement provides that the securities will be credited to the custodian's book-entry account and that payment will be collected from the custodian's reserve account. The bidder reimburses the custodian for the cost of the securities, and the custodian credits the securities to the bidder's account on its internal records.
Retail investors that take delivery in their TreasuryDirect accounts commonly pay by check, by a debit entry to a deposit account, or with the proceeds from a maturing security in the same account.
Deliveries through FICC. FICC was organized in the late 1980s to reduce operational costs and enhance risk management practices in settling secondary market transactions in Treasury and related fixed-income securities. Among other things, FICC nets out confirmed purchases and sales, repurchase agreements, and reverse repurchase agreements21 between its members and, in a legal process known as “novation,” steps in as the buyer from every net seller and the seller to every net buyer. Additionally, FICC marks purchase and sale contracts to current market prices every day in order to limit the exposure of its members and itself to credit risk.
Before 1994, the Treasury settled an auction award to a dealer that was an FICC member in the same way that it settled an auction award to any other institutional investor: by delivering securities to the dealer's Federal Reserve book-entry account or to the book-entry account of the dealer's custodian. This process was inefficient because in many cases the dealer had already sold some or all of its award in when-issued transactions. Suppose, for example, that dealer A was awarded $10 million of a new issue and that, in when-issued trading before or after the auction, the dealer sold $10 million of the same issue to dealer B. On the issue date, the Fed would deliver $10 million of the security to dealer A, which would then have to go to the trouble and expense of redelivering the security to dealer B.
In 1991, FICC opened discussions with the Treasury and the Federal Reserve to expand its netting, settlement, and risk management services to include auction awards, or “takedowns.” The 1992 Joint Report on the Government Securities Market encouraged the effort, noting that “the benefits of netting are greater as more trades are included in the net, because a greater number of receive and deliver obligations are reduced to as small a number as possible” (U.S. Department of the Treasury et al. 1992, p. B-76).
FICC initiated its “auction takedown service” in September 1994. The key idea of the service is that auction awards, when-issued purchases, and the starting legs of reverse repurchase agreements are equivalent for purposes of netting and settlement. Additionally, as shown in the table below, if a new issue reopens an outstanding security, auction awards are also equivalent to conventional purchases of securities with the same CUSIP number, as well as to the closing legs of repurchase agreements. On the issue date, the Fed delivers to FICC securities equal to the aggregate awards of its members. FICC redelivers those securities, along with securities received from members with a net short position, to members with a net long position. Thus, the auction awards lose their separate identities and become part of a consolidated net settlement process.22
In addition to enhancing operating efficiency, the auction takedown service has resolved several risk management problems associated with gross settlement of auction awards. First, the unnecessary deliveries to dealers that were not ultimate buyers created risk for FICC because FICC guaranteed settlement of the redeliveries by those dealers. If an “intermediary” dealer became insolvent before redelivering its securities and the price of the securities had risen, FICC, as part of its liquidation of the insolvent dealer's positions, might have to go into the secondary market and buy the securities at a higher price than it would receive upon redelivery. Second, because FICC did not have knowledge of auction awards made to its netting members, it could not guarantee settlement of those awards (as it would for a secondary market trade), thus leaving the Treasury exposed to credit risk. Finally, FICC was unable to assess proper performance guarantees, or margin, on purchasers and sellers, and it could not mark their positions to market accurately. For example, if dealer A was awarded $10 million of a security in an auction and sold $10 million of the same security in a when-issued transaction to dealer B, FICC viewed dealer A as having a $10 million net short position and collected margin on that position, whereas in fact the dealer was net flat and posed no settlement risk. The unnecessary margin reduced the dealer's liquidity. The auction takedown service allows FICC to margin and mark dealer positions on a true net basis. However, because the Treasury is not a member of FICC, FICC does not mark the Treasury's sales contracts to market. As explained in the appendix ( 2 pages / 102 kb), this feature of the auction takedown service complicates marking procedures for auction awards to FICC members.