View original document

The full text on this page is automatically extracted from the file linked above and may contain errors and inconsistencies.

June 15, 1992

eOONOMIC
GOMMeNTORY
Federal Reserve Bank of Cleveland

Auctioning TVeasury Securities
by E. J. Stevens and Diana Dumitru

X he U.S. Treasury expects to sell
about a trillion dollars of new securities
this fiscal year to finance a projected
$400 billion budget deficit and to
refinance maturing debt. Most of the
securities will be issued through public
auctions, where competition among bidders might be expected to minimize interest payments on the debt.
The competitiveness of Treasury auctions was called into question last
August, however, when Salomon
Brothers, a large securities dealer, admitted to having placed unauthorized
bids in the names of customers during
eight auctions. For example, in the
May 22, 1991 note auction, the firm
controlled more than 90 percent of the
issue, far exceeding the 35 percent
limit set by the Treasury. Rumors of a
market "squeeze" had surfaced even
before the notes were issued on May
31. Disappointed bidders, with contracts to deliver the security after it was
issued, had to pay an unexpectedly high
price for the issue in the secondary
market, where Salomon controlled
most of the supply.
Two causes for concern emerge from
this incident. First, of course, is simply
that the market is not fair when auction
rules are broken. Some investors might
shun the Treasury securities market
rather than be exposed to losses resulting from market manipulation. Thus,
any short-term gain to the Treasury
from an artificially high price at a
single auction could be outweighed by
lower demand and prices in all auctions. Second, as suggested here, is the
possibility that the auction process

ISSN 0428-1276

itself may be at fault. Perhaps a different system of selling new issues of
Treasury debt would reduce incentives
for manipulation.
Following Salomon's admissions and
other reported irregularities, the
Treasury Department, the Board of
Governors of the Federal Reserve System, and the Securities and Exchange
Commission conducted a study in 1991
leading to a joint report on the government securities market. The report reaffirms that public auctions are the best
means of issuing new Treasury debt
and recommends some minor adjustments that have already been adopted
to ensure public access to the existing
auction process. It also recommends a
more thorough exploration of alternative methods of conducting public auctions. This Economic Commentary examines the rationales for adopting a
different system for determining the
price paid by a winning bidder and a
new technology for bidding. Both of
these changes could make public auctions of Treasury securities less susceptible to manipulation.
• Today's Auctions:
Multiple-Price, Sealed-Bid
The Treasury maintains a regular
schedule of auctions in which it sells
bonds, notes, and bills. Bonds and notes
are sold in $1,000 denominations and
pay interest every six months until
maturity, which ranges from 10 to 30
years for bonds and from one to 10
years for notes. The recent schedule
has included monthly auctions of twoand five-year notes and quarterly

The current method of auctioning
Treasury securities contains both incentives and opportunities for market
manipulation. Two suggested changes
in the auctions might eliminate these
problems and thus reduce the need to
police arbitrary auction rules.

auctions of three-, seven-, and 10-year
notes and 30-year bonds.
Treasury bills, in $10,000 denominations, have no coupon. An investor's
return comes from the difference between the maturity value and the price
paid — the "discount." Bills maturing
in 13 weeks and 26 weeks are auctioned
every Monday. Bills maturing in 52
weeks are offered every four weeks.
About a week before each auction, the
Treasury announces the auction day,
size, maturity, and the settlement day
when successful bidders must make
payment. From this announcement
until settlement, the impending security
actually trades in the market on a
"when-issued" basis, with the promise
of delivery on settlement day. Whenissued trading thus may provide potential bidders with information about the
likely price in the auction.
By 1:00 p.m. on auction day, bidders
must submit tenders (written, sealed
bids) at Federal Reserve Banks or their
branches. Two kinds of tenders can be
used: Competitive tenders state the
amount of securities desired (as much
as several billion dollars) and either a

yield bid (for bonds and notes) in onebasis-point increments or a price bid
(for bills) on the basis of 100 (for example, 98.995). Noncompetitive tenders
can be placed only for smaller amounts
(up to $5 million for notes and bonds,
or $1 million for bills), with the yield
or price determined by the average of
awards in competitive bidding.
At 1:00 p.m. on the day of an auction,
bids are tabulated and transmitted to
the Treasury. Securities are awarded to
all noncompetitive bidders; the
remainder of the issue is awarded to
competitive bidders in descending
order of price bid (increasing yield
bid).' A bidder may submit tenders up
to a maximum of 35 percent of the
amount being auctioned, with bids at
various prices. The exception is that if
too many bids are received at the
lowest accepted "stop-out" price, these
awards are made in proportion to the
total of all bids received at that price.
The process can be summarized as a
sealed-bid auction, open to anyone, with
awards at multiple prices. Modifications
already adopted address the openness of
the process, to ensure that any financially responsible party can participate in a
Treasury auction. Some drawbacks are
nevertheless associated with multiple
prices and sealed bidding.
• The Winner's Curse
Making auction awards at multiple
prices means that the highest bidders
must follow through, actually paying
the prices they offered even though
others are paying less for the identical
security. This may seem only fair. After
all, if the high bidders didn't think the
securities were worth so much, they
shouldn't have bid so much. This view
assumes that the item's value to the
winner may be largely independent of
the lower value placed on the item by
unsuccessful bidders.
In the case of Treasury auctions, however, the value of a security is not independent of the market. A dealer wants
the security being auctioned only in
order to sell it, but must compete with
other dealers who may have paid less in

the auction; an investor wants the securities for portfolio purposes and could
wait to buy in the post-auction market.
No matter how fair it may seem, paying
a high price for an award of securities
that others have bought more cheaply
truly involves a "winner's curse."
The winner's curse has serious implications for Treasury auctions. While a
high bid will increase the chances of an
award, it also raises the possibility of
paying more than the post-auction
market value of the security. The auction process might reveal something
about that value, but only the smallestvolume bidders can benefit by submitting noncompetitive tenders; competitive bidders cannot take advantage of
the information revealed in the auction.
The winner's curse dampens the aggressiveness of their bidding, resulting in
a lower auction price received by the
Treasury. Bids may be lowered to cover
the costs of gathering information about
what others are likely to bid. Alternatively, customers may submit their
bids through a small number of wellinformed dealers, where bidding might
tend to become concentrated. Thus, a
winner's curse may dampen all competitive bidding and lower multiple auction prices relative to market values.
Sealed bidding, on the other hand,
when coupled with multiple prices,
enables a single bidder to corner the
post-auction market, transforming the
auction-winner's curse into a postauction blessing. A well-informed and
well-financed bidder or group of bidders could deliberately submit a high
bid to ensure receiving a dominant
share of auction awards. Because of the
substantial volume of when-issued trading, controlling the post-auction supply
of a Treasury security places the high
bidder in a position to squeeze unsuccessful bidders who had contracted
to deliver the security after the auction.
While not a necessary outcome, the
current auction setup does contain the
seeds of market manipulation that can
discourage demand. The winner's
curse places a premium on discovering
how others will bid, creating the basis

for a bid that will corner an auction and
squeeze the post-auction market.
• Safeguards against Manipulation
Treasury auctions contain an important
safeguard against cornering the supply
of a new issue: A single bidder is prohibited from acquiring more than 35
percent of an issue. As the Salomon example attests, enforcing this rule is not
easy. For example, winning bidders
must be contacted to ensure that customers' names are not used improperly by a
dealer trying to control more than his
share of an issue. Moreover, 35 percent
is an arbitrary limit that may restrict
demand and unnecessarily lower auction prices when there is no threat of
manipulation.
One alternative to this rule is to reopen
the issue whose supply has been
cornered. Augmenting supply would
drive down an artificially high price
and eliminate profits expected by the
perpetrators of the squeeze. Knowing
that the Treasury would respond in this
way would eliminate the incentive to
manipulate the market. Of course, this
approach is simply the price equivalent
of the 35 percent rule, reducing the
high bidder's market share by increasing total supply. It would involve an
equally arbitrary judgment about the
permissible range within which the
post-auction price could vary without
triggering a reopening, as well as arbitrary judgments about whether prices
reflected manipulation or a change in
market fundamentals.
• A Single-Price Auction
Changing to a single-price auction has
been proposed as an alternative to enforcing arbitrary rules on the current
auction process. This procedure would
be identical to the current one, except
that all competitive awards would be
made at the stop-out price, which is the
price that clears the market. The winner's curse would disappear because
winning bidders would receive their
awards at the stop-out, even if they bid
higher.
Resistance to a single-price auction
centers around its implication for the

FIGURE 1 SINGLE-PRICE
AUCTIONS: GAINS AND LOSSES

Price

Quantity
SOURCE: Authors.

cost of servicing government debt.
With a single-price auction, the Treasury would forgo revenue now received
from the difference between successful
price bids and the stop-out price. Actually, however, Treasury auction revenues might increase, although auction
theory is ambivalent on this matter.
Revenue forgone by shifting from a
multiple- to a single-price auction
might be more than offset by an increase in demand as participants bid
more aggressively in the absence of a
winner's curse, as illustrated in figure 1.
Whether the gain would actually exceed forgone revenues, though, depends
in part on the size of the increase in demand. It could be slight if most participants are already so averse to the risk
of not getting an award that they bid
aggressively in multiple-price auctions,
despite the winner's curse. So, a singleprice auction will increase demand, but
might not reduce the interest cost of
financing the debt.
More to the point, a single-price auction also might or might not produce a
less fertile environment for market
manipulation. A high bid could garner
a dominant market position without requiring the successful bidder to pay
any more than the stop-out price. But
other incentives to corner the market
might be weaker, because bidders
would no longer conduct as intensive a
pre-auction search for information
about others' bids. Thus, a bidder

would have to make special efforts to
gather information; cornering the
market would become more costly.
Chances of pre-auction detection by
other bidders, who might raise their
own bids by enough to defeat an attempted corner, would also increase.
• An Auction by Open Outcry
Perhaps the most intriguing alternative
presented in the joint report is that of
conducting Treasury auctions by open
outcry, rather than by sealed bid. The
disadvantage of sealed bidding is that
participants cannot be certain of the
distribution of all other bids. In the
multiple-price auction, winning with a
bid higher than the stop-out implies
that your estimate was too high, and
the risk of this winner's curse leads to
conservative bidding. In a single-price
auction, your winning bid is lowered to
the stop-out, based on the auctioneer's
information about the actual distribution of bids, all of which will be less
conservative than in a multiple-price
auction. In an open-outcry auction,
which is typically used to sell antiques
or works of art, information about the
distribution of bids is revealed gradually to all bidders before the auction
closes. As the auction process drives
the bidding higher and higher, low
prices become irrelevant.
An attempt to gain a dominant share of
Treasury auction awards seems unlikely
to succeed with this system, because the
auction process reveals more information about the actual distribution of
bids. If the distribution of bids initially
contains an unusually large bid at a high
price, the ascending price reveals that a
large number of bidders underestimated
the value of the security and therefore
can raise their bids. While the final auction price may be unusually high, the
individual who attempted to corner will
not gain a large share of awards.
In the past, formidable obstacles have
stood in the way of conducting Treasury auctions by open outcry, obstacles
mostly absent from the typical auction
of a unique art object. At one time, it
might have been difficult even to assemble all bidding in a single place, but

today, open communication lines to
Federal Reserve offices and to individual bidders could overcome that geographic problem, just as telephone bidding does in major art auctions.
More difficult, and unlike an auction of
a single work of art, bidders must cry
out not just a price, but a quantity at
each price. Bidding must ascend from
low prices, where the demand for
securities exceeds the supply, to the
market-clearing price, where demand
exactly equals supply. As envisioned in
the joint report, bidding would have to
proceed in discreet "rounds," starting
from a low price and rising by small increments. As the results of each round
were announced, participants could
drop out, resubmit the same quantity,
or adjust the quantity in light of information gained from the previous round.
Auctioning Treasury securities by open
outcry would consume too much time
to be feasible with current auction technology. Each round of bidding probably would take more than an hour
using today's paper tenders, manual
tabulation, verification of payment
status, transmission to the Treasury, and
final compilation. Each open-outcry
auction might consume most of a day,
during which unfolding world events
might be moving market fundamentals
enough to change bidding decisions and
to prevent the process from moving
smoothly toward an equilibrium
market-clearing price.
The Treasury expects to automate the
current sealed-bid, multiple-price auction process this year, allowing telecommunication of tenders and computerized
compilation of bids. Going any further
toward auctions by electronic open outcry would require substantial investments, both in hardware and software
and, before that, in design. One critical
design question would involve the information feedback provided at the end of
each round of bidding. Simply announcing the amount of excess demand might
not be sufficient for bidders to detect
an aggressive effort to dominate the
awards. On the other hand, a complete
(anonymous) list of all bid quantities

received might prove too cumbersome
for rapid rounds of bidding. Another
design question involves the computer
and telecommunications capacity required to receive large numbers of bids
virtually simultaneously. Too little is
known at this time, but automation
could conceivably reduce the turnaround time between successive rounds
of bidding to a matter of minutes, with
an entire auction lasting perhaps less
than an hour. Given the information
benefits of open outcry, serious further
exploration seems warranted.

current incentive of each bidder to discover the price at which other bidders
will make tenders. With less of their activity focused on discovering one
another's intentions, bidders would
have less opportunity and incentive to
corner an issue. Perhaps more promising would be to enlist telecommunications and computer technology in conducting auctions by electronic open
outcry, relying on competition among
the bidders themselves to limit the
share of auction awards controlled by
any single bidder.

• Conclusion
The U.S. Treasury securities market is
the largest and most efficient market in
the world today. Nonetheless, the events
of 1991 demonstrated that concerted
strong bidding in the primary market for
an issue could result in a squeeze in the
secondary market. More aggressive
policing of the current rule against control of more than 35 percent of a new
issue promises to reduce chances of
similar episodes in the future.

• Footnotes
1. A squeeze occurs when there is an unexpected restriction of supply relative to
demand for a particular security, manifested
by an unusually high price of that security
relative to prices of comparable securities.

Embedded in the current sealed-bid,
multiple-price auction process are both
incentives and opportunities for a bidder to seek a dominant share of awards.
New methods of conducting auctions
are worth serious consideration if they
would reduce the need for arbitrary
policing of the market. A single-price,
sealed-bid auction would reduce the

Federal Reserve Bank of Cleveland
Research Department
P.O. Box 6387
Cleveland, OH 44101

Address Correction Requested:
Please send corrected mailing label to
the above address.

Material may be reprinted provided that
the source is credited. Please send copies
of reprinted materials to the editor.

2. See Joint Report on the Government
Securities Market. Washington, D.C.: Department of the Treasury, Securities and Exchange Commission, and Board of Governors of the Federal Reserve System, January
1992. A somewhat more technical examination of these matters appeared recently in
Vincent Reinhart, "Theory and Evidence on
Reform of the Treasury's Auction Procedures," Federal Reserve Board, Finance and
Economics Discussion Series No. 190,
March 1992.
3. Federal Reserve Bank and government
tenders are also awarded in full at the price
established for noncompetitive bidders, but
the amount of these awards is added to the
amount being auctioned.

4. Milton Friedman has been a long-time
proponent of this procedure. For references,
see Joint Report on the Government Securities Market, p. B-22.
5. See Robert J. Weber, "Multiple-Object
Auctions," in Richard Engelbrecht-Wiggins,
Martin Shubik, and Robert M. Stark, eds.,
Auctions, Bidding, and Contracting: Uses
and Theory. New York: New York University Press, 1983.
6. An alternative to rounds of bidding might
be to conduct auctions on an interactive,
open-screen basis: Qualified bidders would
have access directly (or at a Federal Reserve
Bank) to a telecommunications terminal.
Beginning hours (or days) before the close of
an auction, a screen would display an instantaneously updated list of all (anonymous)
bids received and their implied auction stopout price. Bidders would have the right to
alter their bids up until the close of the auction, when the final stop-out price would be
determined, with awards made to all who bid
at or above that price.

E. J. Stevens is an assistant vice president
and economist and Diana Dumitru is a
senior research assistant at the Federal
Reserve Bank of Cleveland. The authors
thank Charles Carlstrom for helpful
comments.
The views stated herein are those of the
authors and not necessarily those of the
Federal Reserve Bank of Cleveland or of the
Board of Governors of the Federal Reserve
System.

BULK RATE
U.S. Postage Paid
Cleveland, OH
Permit No. 385