Auction Basics: Background for Assessing Proposed Treasury Purchases of Mortgage-Backed Securities

Auction Basics: Background for Assessing
Proposed Treasury Purchases of Mortgage-
Backed Securities
October 14, 2008
D. Andrew Austin
Analyst in Economic Policy
Government and Finance Division



Auction Basics: Background for Assessing Proposed
Treasury Purchases of Mortgage-Backed Securities
Summary
To address the turmoil in financial markets, the Emergency Economic
Stabilization Act (EESA; H.R. 1424, P.L. 110-343), enacted on October 3, 2008,
authorizes purchases of “troubled assets.” The act passed the Senate on October 1,

2008, passed the House on October 3, 2008, and was signed into law the same day.


The Administration proposed using reverse Dutch auctions to purchase troubled
assets — primarily mortgage-related securities from financial institutions. In reverse
Dutch auctions, a buyer purchases multiple objects from private parties at a price set
by the last accepted bid. The government has used reverse auctions since the
Revolutionary War. Designing efficient reverse Dutch auctions may present some
tradeoffs between enhancing competition among bidders and overpaying for assets
relative to their quality. Careful auction design, however, can help minimize these
problems.
Auctions are especially useful for selling assets whose value to potential owners
is unknown to the seller. Reverse auctions are useful when a buyer does not know
what value sellers place on assets. Auction results could clarify the market value of
troubled assets. The price discovery properties of auctions could stimulate trading
by reducing private traders’ uncertainty about the value of troubled assets.
A reverse auction program essentially swaps Treasury securities for troubled
mortgage-backed securities. If Treasury securities are exchanged for troubled assets
at prices close to those assets’ current market prices, costs to the taxpayer would be
minimized. Financial institutions, however, may gain some liquidity, but might not
receive much additional capital. Some economists have argued that other means of
injecting capital into the financial sector, such as purchases of preferred stock or
capital injections balanced by equity warrants (i.e., options to claim an equity stake),
might be a better strategy. Since passage of EESA, the U.S. Treasury has been
working to design methods to inject capital into firms and restore market liquidity.
The heterogeneity of troubled assets may present challenges to the Treasury
auction program. The reverse Dutch auctions would need to be adapted to buy highly
diverse and relatively small-volume securities, in a way that minimizes risks of
trading manipulation. Reverse Dutch auctions may be vulnerable to adverse
selection, meaning that the average credit quality of submitted assets of a given type
may be systematically worse than the average credit quality of all assets of that type.
Auction mechanisms might be designed that could mitigate these problems.
Recent academic research in auction theory and in experimental economics has
examined how various types of auctions work. Auctions may capture higher
revenues for governments and can often allocate scarce resources more efficiently
than traditional methods of selling or purchasing. Different policy problems,
however, call for different types of auctions. Government economists involved in
designing reverse auctions to buy troubled assets have drawn upon academic research
and internal Treasury research. This report will be updated as events warrant.



Contents
Dutch Auctions and Reverse Dutch Auctions............................2
Quality Differences of Troubled Assets Presents Challenges for Auctions..4
Adverse Selection and Firms’ Asset Holdings...................6
Reputational Issues and Participation..........................7
Sequencing Issues.............................................7
Will Auctions Unlock Credit Markets?.............................8
Costs and Risks to Taxpayers...................................10
Auction Design..................................................10
Single-Unit Auctions..........................................10
Why Are Treasury Auctions Called Dutch Auctions?.............11
Imperfect Information: Common Values and the Winner’s Curse...11
Revenue Equivalence......................................12
Multiple-Unit Auctions........................................13
Manipulation and Demand-Reduction Strategies................13
Complementarities ........................................13
Dutch Auctions with Multiple Units..........................13
Uniform-Price vs. Individual Price Auctions....................14



Auction Basics: Background for Assessing
Treasury Purchases of Mortgage-
Backed Securities
To address the turmoil in financial markets, Congress passed and the President
signed the Emergency Economic Stabilization Act (EESA; H.R. 1424, P.L. 110-343)
on October 3, 2008. The act authorizes the Secretary of Treasury “to restore liquidity
and stability to the financial system of the United States” (Sec. 2) by purchasing or
insuring “troubled assets.” The Secretary will design the mechanism for purchasing
the troubled assets and methods for pricing and valuing these assets. The act broadly
defines troubled assets as (1) mortgages and any securities based on such mortgages
whose purchase the Secretary determines will promote financial market stability, and
(2) any other financial instrument whose purchase the Secretary and the Federal
Reserve Board Chairman determines will promote financial market stability.
The act authorizes Secretary Paulson to use $700 billion to enhance liquidity
and to inject capital into financial markets by purchasing (1) mortgages and pools of
mortgages, (2) preferred stock in ailing financial institutions, and (3) troubled
mortgage-backed securities (MBSs).1 The Secretary has indicated that reverse
auctions will be used to purchase mortgage-backed securities from troubled financial
institutions. A well-designed auction could help determine the price and value of
these assets in an efficient manner.
The act establishes a Congressional Oversight Panel to review the current state
of financial markets, the regulatory system, and the administration of the Troubled
Asset Relief Program (TARP) by the Secretary. This report provides Congress with
information on the uses, design, and functions of auctions. It reviews some common
types of auctions used by the federal government and some of the complexities of
auction design that the Secretary will have to deal with in the TARP.


1 MBSs are debt obligations representing claims to the cash flows (principal and interest)
from pools of mortgages primarily on residential property. The main use of MBSs is to
transform relatively illiquid financial assets (mortgages owned by mortgage originators) into
liquid and tradable capital market instruments. Given the problems with the housing market
and subprime mortgages, it is difficult to know the true value of MBSs. Consequently,
MBSs have become illiquid.

Dutch Auctions and Reverse Dutch Auctions
The Administration has proposed using reverse auctions to purchase mortgage-
related securities from financial institutions, which are similar to the multiple-unit
Dutch auctions that the Treasury uses to sell government securities. In a reverse
auction, a buyer accepts bids from multiple potential sellers. In reverse multi-unit
Dutch auctions, a buyer (e.g., the U.S. Treasury) buys a given number of units from2
private parties (e.g., financial institutions) at a price set by the last accepted bid.
Figure 1 below contains a hypothetical example of a reverse Dutch auction.
Auctions provide a means of selling objects whose value to potential owners is
unknown to the seller. Many different auction mechanisms are in common use. A
large research literature in economic theory and experimental economics examines
how different types of auctions work. On the one hand, well-designed auctions can
provide an expeditious and efficient method for selling or acquiring objects. On the
other hand, poorly designed auctions have caused governments to forego large
amounts of revenue. Moreover, poorly designed auctions may increase the likelihood
that valuable resources are allocated to buyers who value those resources less than
others — a source of inefficiency.3 In particular, auctions suitable for some
applications may be unsuited for other applications.
The American government has used reverse auctions since Robert Morris, who
headed the Treasury Department, used them to procure supplies for troops in the
Revolutionary War.4 In recent decades, the Treasury Department has used auctions
to sell federal securities.5 Over time, the federal government has gained valuable
experience in conducting and designing auctions, which has reduced costs and
increased revenues compared to other methods.


2 Dutch auctions usually refer to falling-price auctions, which are described in more detail
below. The reverse multi-unit sealed-bid auction outlined by the Treasury is strategically
equivalent to a certain falling-price multi-unit auction, which is why the former is usually
called a Dutch auction.
3 Paul Klemperer, “Using and Abusing Economic Theory — Lessons from Auction Design,”
Journal of the European Economic Association, 2003, available at
[http://www.paulklemperer.org/]. In some situations, auction designers may face a tradeoff
between maximizing expected seller revenue and maximizing the likelihood that resources
are allocated to those who value them the most.
4 Robert E. Wright, One Nation Under Debt: Hamilton, Jefferson, and the History of What
We Owe, (New York: McGraw-Hill, 2008), p. 61.
5 Various federal agencies have used different auction mechanisms for many years. The
Department of Interior has run auctions to distribute Outer Continental Shelf (OCS) oil and
gas leases since 1954. Since 1994, the Federal Communications Commission has used
complex auction mechanisms to license slices of electromagnetic spectrum for use in
wireless communications.

Figure 1. A Hypothetical Reverse Dutch Auction
Suppose a government agency wished to purchase 1,000 desktop computers of a given
specification. Potential sellers would submit sealed bids with a price and a number of units.
The agency would then sort bids by price in ascending order. The agency would accept
bids, starting with the lower price, and continuing until units offered in bids reached 1,000.
Hypothetical bids are shown below.
Bids for a Hypothetical Government Procurement Reverse Auction
Hypothetical FirmBid Price (Per Unit)Number of Units
Hang-10 IT $450400
Crash Computers$470500
Macrohard Inc.$510200
HAL Systems$525300
The agency would accept the bid from Hang-10 IT and buy 400 units because it had
the lowest bid price, although Hang-10 IT’s bid price ($450) would not be the purchase
price. The agency would also accept Crash Computers’ bid for 500 units, because its bid
price was the second lowest. Finally, the agency would buy 100 computers from Macrohard
Inc. to reach the total of 1,000 computers. Because Macrohard was the last successful
bidder, all computers would be bought at a unit price of $510. HAL Systems would sell no
computers to the agency.
Some may wonder why the government agency would not force firms to sell
computers at their bid prices, which might seem to save the agency money. Such a strategy
would probably be unsuccessful in saving money because firms would change their bidding
strategies if they knew that they would have to sell at their bid prices. In particular, sellers
would tend to shade their bids upwards, which might raise costs to the agency.
Specific rules and operation details for reverse Dutch auctions used by governments
and private firms have varied. Some central banks use multiple-price or pay-your-bid
auctions to sell securities, others use uniform-price auctions, and some have used both.
Differences between multiple-price and uniform price auctions are discussed in more detail


Current U.S. Treasury auctions, which use a multi-unit Dutch auction
mechanism to sell securities, have apparently provided a starting point for
Administration proposals to buy “troubled assets” via reverse Dutch auctions.
Because the reverse Dutch auction process would, in approximate mirror fashion,
resemble Treasury auctions of government securities to primary dealers, key market
participants could quickly familiarize themselves with new bidding procedures.
Those administering Treasury reverse auctions, however, recognize that differences
between Treasury securities and troubled assets have consequences for auction
design.6 Designing reverse Dutch auctions may present some tradeoffs between
enhancing competition among bidders and overpaying for assets relative to their
quality. Careful auction design, however, can help minimize these problems.
6 Discussion with government official, Sept. 26, 2008.

In the Dutch auction mechanism used to sell U.S. Treasury securities, all
successful buyers pay the same price; in other words, it is a uniform-price
mechanism. Uniform-price mechanisms, according to some experts, may encourage
more aggressive bidding, which raises expected revenues.7
While details of the Treasury reverse Dutch auctions remain unspecified, the
U.S. Treasury would probably announce that it wished to buy a given amount of
mortgage-related securities (MBSs) of specific types or issues for a given auction.
Bidders would then list securities they wish to sell and specify prices. The Treasury
would then buy the securities listed at the lowest prices until the specified amount
was reached. The price offered by the last successful bidder would then be paid to
all successful bidders.
Quality Differences of Troubled Assets
Presents Challenges for Auctions
The diversity or heterogeneity of troubled assets may present challenges to the
Treasury auction program. The mortgage-backed securitization process was intended
to create marketable assets with credit characteristics that could be readily assessed
by credit agencies and buyers. Credit rating agencies claimed that the rating process
sorted asset-backed securities and other assets into categories with essentially
homogenous profiles. All assets of a specific type receiving a given rating were
supposed to embody essentially similar risk characteristics.
Confidence in the credit rating process has deteriorated over the past two years,
however. Credit rating agencies, according to the Securities and Exchange
Commission (SEC), struggled to keep up with more complex types of securities and
had difficulty assessing risks embedded in subprime mortgage-backed assets. In
addition, according to the SEC, none of the rating agencies examined had specific8
written procedures for rating residential MBSs and collateralized debt obligations.
Thus, some buyers may doubt that credit ratings represent a true gauge of asset
quality. Tightened market liquidity in capital markets since August 2007 may stem
in part from potential buyers’ uncertainty about the credit quality of MBSs and other
assets. Buyers may worry that sellers could have incentives to sell assets with
difficult-to-detect risks before assets without such hidden hazards.
Similar problems may occur with reverse Dutch auctions used to buy troubled
assets. A typical Treasury auction sells a large volume of identical government
securities whose characteristics are well understood. The reverse Dutch auctions
used by the Treasury would need to be adapted to buy highly diverse and relatively
small-volume securities, whose characteristics may not be well understood by many


7 The U.S. Treasury has run multiple-price (in which buyers pay their bids) and uniform-
price auctions in the past. In particular, multiple-price auctions may be more susceptible to
the winners’ curse, which may cause bidders to shade their bids downwards. The winners’
curse and other issues relating to differences between multiple-price and uniform-price
auctions are discussed below.
8 U.S. Securities and Exchange Commission, Summary Report of Issues Identified in the
Commission Staff’s Examinations of Select Credit Rating Agencies, July 2008.

buyers. A typical mortgage-backed security issue, while enormous relative to a
single housing mortgage, is small when compared to the size of a typical Treasury
security issue, and individual tranches (slices) of MBSs are smaller still.9 Different
MBSs and related structured finance assets are very diverse, although their structures
and pricing follow some general principles.10 Thus, selling Treasury securities is like
selling commodity steel; buying mortgage-backed securities is like buying used cars.
If the Treasury were to run a large number of narrowly defined reverse auctions,
it would be more difficult to prevent certain types of market manipulation.11 For
example, a bidder who gains control of a large proportion of an issue might exert
influence on auction outcomes. But, if a wider variety of securities or assets were
allowed in the same auction, which would sharpen competition, sellers may have an
incentive to submit bids for those assets with hidden flaws.
Reverse Dutch auctions may therefore be vulnerable to adverse selection,
meaning that the average credit quality of submitted assets of a given type may be
systematically worse than the average credit quality of all assets of that type. In
addition, if assets submitted to an auction were on average worse than other assets
of a given type, then auction prices might be biased downwards. Some empirical
research has found that adverse selection problems can lower prices on eBay
auctions.12 If reverse auctions for troubled assets generated downwardly biased
prices, that could affect valuations of assets held by other firms through mark-to-
market accounting requirements. Thus, a downward bias in auction prices due to
adverse selection could affect the market value or even solvency of some firms,
whether or not they participated in auctions.
Moreover, managing a portfolio bought via reverse Dutch auctions susceptible
to adverse selection could present financial risks to the federal government. While
adverse selection could push prices down (relative to prices appropriate for the
average of all assets of a given category), the quality of assets accepted by the


9 Most Treasury issues are generally sized in the tens of billions of dollars range, although
cash management and other concerns may lead to smaller issues (U.S. Department of
Treasury, Treasury Bulletin, various issues). Structured product issues (i.e., asset- or
mortgage-backed securities, or collateralized default obligations (CDOs)) are generally sized
in the hundreds of millions of dollars range, and are often divided into a half dozen or so
tranches. Tranche comes from the French verb tranchez meaning to slice or to carve.
10 The Bond Market Association and American Securitization Forum note that “(e)ach
product sector is distinct based on issuers, investors, origination, servicing and collateral
management, trading and pricing systems, and type and volume of available information.
These differences lead to distinct pricing and valuation conventions, liquidity levels and risk
exposures and sensitivities.” An Analysis and Description of Pricing and Information
Sources in the Securitized and Structured Finance Markets, Oct. 2006, p. 1, available at
[ ht t p: / / www.mi acanal yt i c s.com/ f s / About MIA C / P R / P r i c i n g-In f o r ma t on_Sources_Study_

1006.pdf].


11 Manipulation in auctions is discussed below.
12 Sanjeev Dewan and Vernon Hsu, “Adverse Selection in Electronic Markets: Evidence
from Online Stamp Auctions,” Journal of Industrial Economics, vol. 52, no. 4, pp. 497-516,
Dec. 2004.

Treasury through the reverse auction mechanism could also be lower (relative to
average quality of all assets of a given category). To the extent that Treasury
overpays for assets relative to their quality due to adverse selection, costs to the
taxpayer rise. Whether this effect presents a significant financial risk to the taxpayer
is difficult to determine before the reverse Dutch auctions have been running for
some time.
Auction mechanisms, however, might be designed that could mitigate these
adverse selection problems.13 Analysts close to government auction design
discussions have outlined a design in which several similar securities could be listed
for an auction.14 Price offsets or “handicaps” could be applied relative to a
benchmark security. For instance, an auction might encompass a specified set of
MBSs, underwritten by the same investment bank in the same month. Bids offering
a specific MBS issue with a higher average default rate on the underlying mortgages
would be reduced by an offset calculated using a financial asset pricing model.
Other auction mechanisms might also mitigate these problems. For example, a
firm could be required to let the government select an asset randomly from all of its
holdings of a specific type of asset if its bid were successful. Or participating firms
could be required to post a performance bond that could be used to compensate the
government if a firm’s assets sold in auctions turned out to be systematically worse
than average.15 Charles Plott, a pioneer in auction design, has designed and tested a
reverse Dutch auction that was able to mitigate adverse selection concerns in another
contex t. 16
Adverse Selection and Firms’ Asset Holdings. Uncertainty about the
composition of firms’ holdings could be an additional source of adverse selection in
the administration of the Troubled Asset Relief Program. Some claim that financial
institutions have avoided offering to sell large amounts of subprime and other
troubled assets out of a concern for the institution’s reputation. Other market
participants might infer that a would-be seller has a large inventory of subprime and
other troubled assets and thus may be a risky counterparty.17 That inference could
adversely affect a potential seller’s stock price and its ability to raise capital, thus
providing a reason for firms to hold troubled assets. In other words, some firms may
worry that attempting to sell troubled assets may damage its reputation, market value,


13 Justin Lahart, “Economists Look at Ways to Structure Auctions,” Wall Street Journal,
Sept. 25, 2008.
14 Discussion with government official, Sept. 26, 2008.
15 Matthew O. Jackson and Hugo F. Sonnenschein, “Overcoming Incentive Constraints by
Linking Decisions,” Econometrica, vol. 75, no. 1, Jan. 2007, pp. 241-257.
16 Charles Plott and P.J. Brewer, “A Decentralized, Smart Market Solution to a Class of
Back-Haul Transportation Problems: Concept and Experimental Test Beds,” Interfaces, vol.

32, no. 5, pp. 13-36, 2002.


17 Jonathan Carmel, “Pitfalls of the Paulson Plan,” Economists’ Voice, vol. 5, no. 5, Oct. 2,

2008, available at [http://www.bepress.com/ev/vol5/iss5/art14].



and ability to trade. However, many view the former investment bank Merrill
Lynch’s sale of troubled assets as a sound strategic move.18
Reputational Issues and Participation. If financial markets were to
associate participation in the Troubled Asset Relief Program (TARP), including
bidding in Treasury reverse auctions, with financial weakness, then firms might be
less willing to participate. Reputational issues, in the view of many financial experts,
have discouraged some firms from participating in certain federal government or
Federal Reserve programs. For example, in recent years some have viewed firms
using the Federal Reserve’s discount window as “desperate.”19 The Federal Reserve
Term Auction Facility (TAF), in the view of some, was designed to provide liquidity
to firms while avoiding the stigma that some might perceive to be associated with the20
discount window. U.S. Treasury, however, has stressed that TARP is “not targeted
at failing firms,” but instead is designed to attract broad participation among financial21
institutions.
The effectiveness of the reverse auction asset purchase program could be
reduced if reputational issues caused some firms to forego participation in TARP.
Factors that reduce the number of active bidders in an auction can decrease expected
revenues and can dampen competition among bidders. The choice of how auction
results are released and how much detail is disclosed may affect firms’ willingness
to submit bids. For example, if firms believe that market analysts can observe or
infer from announced auction results that the firm holds large inventories of troubled
assets, then that firm may become reluctant to participate in the auction. Designing
Treasury reverse auctions for troubled assets and associated announcements of results
to avoid any such possible reputational effects may enhance firms’ willingness to
participate.
Sequencing Issues
The order or sequencing in which reverse securities auctions take place may
affect the results of auctions. First, the initial reverse auctions might be a learning
process for both bidders and the U.S. Treasury. This might imply that smaller,
simpler auctions should precede larger and more complex auctions. In the past, new


18 Merrill Lynch, “Merrill Lynch Announces Substantial Sale of U.S. ABS CDOs, Exposure
Reduction of $11.1 Billion,” Press release, July 28, 2008, available at [http://www.ml.com/
index.asp?id=7695_7696_8149_88278_101366_103431].
19 Allan Sloan, “The Fed’s Thaw,” Washington Post, Dec. 25, 2007, p. D1.
20 The Term Auction Facility was created in December 2007. For details, see the Federal
Reserve Bank’s TAF webpage, available at [http://www.federalreserve.gov/monetarypolicy/
taf.htm].
21 Neel Kashkari, Assistant Treasury Secretary for International Economics and
Development, “SIFMA Conference Call,” Sept. 28, 2008, available at
[http://www.youtube.com/watch?v=uXmFADm61Rc]. On Oct. 6, 2008, Mr. Kashkari was
named Interim Assistant Secretary of the Treasury for Financial Stability and will oversee
TARP. U.S. Treasury, Press release hp-1184, Oct. 6, 2008, “Kashkari Appointed Interim
Assistant Secretary for Financial Stability, available at [http://www.treas.gov/press/releases/
hp1184.htm].

auction designs have sometimes presented unanticipated operational problems or
unforeseen strategic vulnerabilities. Careful design and testing of auction
mechanisms, however, can minimize such problems. Second, if the Treasury
interventions in financial markets do start to unfreeze credit markets as intended, then
asset prices will change. Bidders’ anticipation that asset prices will rise after the
initial auctions (or because of other types of government intervention) could induce
bidders to submit fewer assets in earlier auctions. Third, sequencing may raise
operational issues because of the large number of asset types, and because of the
intrinsic complexity of some mortgage-related securities.
Will Auctions Unlock Credit Markets?
The September 20, 2008, Treasury proposal suggested that reverse auctions
would play a central role in restoring liquidity to credit markets. It argued that
auctions could help stabilize asset and credit markets in two ways. First, firms with
illiquid assets would sell them at prices determined by a competitive process, which
would supply firms with liquid proceeds of those sales. Second, auction results could
provide pricing benchmarks that might stimulate trading in other assets. Some
believe that this could improve liquidity conditions in credit markets; others,
however, are skeptical.22
The reverse auction program essentially swaps Treasury securities for troubled
mortgage-backed securities. If the prices at which Treasury securities are exchanged
for troubled assets are close to current market prices for those assets, then financial
institutions may gain liquidity, but might not receive much additional capital.
What prices the “troubled” mortgage-related assets will sell at is, therefore, a
key question. Merrill Lynch, for example, sold a large stake of senior mortgage-
related collateralized default obligations (CDOs) to Lone Star Funds, a private capital
fund, for about 22 cents on the dollar.23 If other assets sold at Treasury reverse
auctions at prices reflecting similar discounts, the solvency of some financial
institutions might be put in doubt.
Some commentators, however, have suggested that Treasury might believe that
such assets are underpriced in current conditions.24 Assets might be underpriced,
relative to fundamental factors, because of the rapid deleveraging of financial


22 Paul Krugman claims that Treasury Secretary Paulson has “never been able to explain
clearly why buying up bad mortgage assets at market prices will solve the credit crunch.”
“Bailout Narratives,” New York Times blog, Oct. 1, 2008, available at
[http://krugman.blogs.nytimes.com/].
23 Merrill Lynch, “Merrill Lynch Announces Substantial Sale of U.S. ABS CDOs, Exposure
Reduction of $11.1 Billion,” Press release, July 28, 2008, available at
[ ht t p: / / www.ml .com/ i ndex.asp?i d=7695_7696_8149_88278_101366_103431] .
24 Paul Krugman, “A $700 Billion Slap in the Face,” New York Times blog, September 24,

2008, available at [http://krugman.blogs.nytimes.com/2008/09/24/


a-700-billion-slap-i n-the-face/].

institutions, which increased the supply of assets and the demand for liquidity.25 As
the price of liquidity rises, measured as the cost of overnight interbank borrowing
relative to comparable Treasury rates, highly leveraged financial institutions may
come under additional pressure to sell assets. Thus, deleveraging and falling asset
prices may create a self-reinforcing spiral. To the extent that buying assets via a
reverse auction process might strengthen the link between assets and their underlying
values, more normal market conditions might be restored. But others argue that the
low prices of “troubled” assets reflect their intrinsic value, and that previous prices
were above those justified by fundamentals.26
Federal Reserve Chairman Ben Bernanke spoke of “hold to maturity” valuations
of assets, which would seem to imply that prices above current market levels might
be paid.27 While paying above-market prices for assets would inject capital into
financial institutions, it would also increase the costs and risks to taxpayers. How
auctions could be designed to ensure “hold to maturity” valuations of assets is
unclear. One economist with knowledge of auction design discussions has said that
reverse auctions would be designed in a hard-nosed manner to minimize taxpayer
costs. But, if current asset prices accurately reflect fundamental value and if the
Treasury reverse auctions are run efficiently, comparatively little capital would be
injected into the financial sector. While this would minimize costs and risks to
taxpayers, such an auction program might provide limited support for financial
institutions.
Some economists have argued that other means of injecting capital into the
financial sector, such as purchases of preferred stock or capital injections balanced
by equity warrants (i.e., options to claim an equity stake), might be a better strategy.28
On October 8, 2008, the U.S. Treasury emphasized that EESA gives it authority to
directly inject capital into firms, and is developing strategies to do so.29
Asset purchases and direct capital injections may have different implications for
affected firms. Proceeds of asset purchases would presumably be counted as trading
profits, which firms can use without restriction. Capital injections, however,


25 Leverage is the value of a firm’s assets relative to its debts. Deleveraging typically means
selling assets or restructuring positions to reduce debt levels.
26 Nikki Kahn “Away from Wall Street, Economists Question Basis of Paulson’s Plan,”
Washington Post, Sept. 26, 2008.
27 Testimony of Federal Reserve Chairman Ben Bernanke, in U.S. Congress, Senate
Committee on Banking, Housing, and Urban Affairs, Turmoil in U.S. Credit Markets:
Recent Actions Regarding Government Sponsored Entities, Investment Banks and Otherthnd
Financial Institutions, 110 Congress, 2 sess., Sept. 23, 2008.
28 Nikki Kahn “Away from Wall Street, Economists Question Basis of Paulson’s Plan,”
Washington Post, Sept. 26, 2008; Paul Krugman, “A $700 Billion Slap in the Face,” New
York Times blog, September 24, 2008.
29 Secretary Paulson, in an Oct. 8, 2008, press statement, mentioned capital injections before
he mentioned asset purchases. Henry Paulson, Secretary of the U.S. Treasury, Press release
hp1189, Oct. 8, 2008; Edmund L. Andrews and Mark Landler, “U.S. May Take Ownership
Stake in Banks,” New York Times, Oct. 8, 2008, p. A1.

generally provide firms with financial resources that are subject to restrictions. For
example, capital injections might provide the U.S. Treasury with the right to demand
management changes or equity warrants. Treasury purchases of preferred equity
shares would probably commit firms to make regular, specified payments back to the
Treasury. Proceeds from asset purchases, however, might be available for dividends,
executive compensation, reducing debt, or other purposes.
Costs and Risks to Taxpayers
Federal interventions to restore more normal conditions to financial markets
would provide substantial benefits to those connected to the financial sector, either
directly or indirectly. Of course, some may receive greater benefits than others from
a resuscitation of the financial sector. A large-scale federal intervention could
impose substantial costs and risks on taxpayers and federal program beneficiaries,
although the scale and nature of those costs and risks may depend on how
interventions are structured and administered.
If Treasury reverse auctions were conducted in a hard-nosed and efficient
manner, direct costs to taxpayers and beneficiaries could be minimized.
Furthermore, the federal government might well eventually sell assets for more than
their purchase price. Such auctions, however, might supply little extra capital to the
financial sector, and thus may fail to achieve a normalization of market conditions.
Other measures, such as debt/equity swaps, purchases of preferred stock, or trading
stock warrants for capital injections, might present the taxpayer with greater financial
risks, but might also be better suited to addressing current financial conditions.
Auction Design
Auctions in recent years have been used to address a wide range of policy issues.
Auctions may capture higher revenues for governments and can allocate scarce
resources more efficiently than traditional methods. Different policy issues,
however, may require different types of auctions to achieve reasonable results. To
provide a basis for evaluating the reverse auction mechanisms that may be used in
TARP, this section discusses potential problems that may arise in using auctions, and
how those problems can be minimized by careful design of auction mechanisms.
Single-Unit Auctions
The most common auction mechanisms used to sell single items are the first-
price sealed-bid auction, the English or ascending-price open-bid auction, and the
Dutch or descending-price auction. In the first-price sealed-bid auction, bidders
submit bids to the seller, who then selects the highest bid when selling an item or the
lowest bid when buying an item. In the English or ascending-price auction, bidders
announce prices that must exceed previous prices by a set amount. The last bidder
to remain receives the object at her last announced price. In flower markets in
Amsterdam and other trading centers in the Netherlands, bidders watch a price clock
that starts at a high price and descends at a constant rate. The first bidder to press a



button buys the lot of flowers at the price indicated on the price clock.30 Similar
descending-price mechanisms are called Dutch auctions.31
Why Are Treasury Auctions Called Dutch Auctions? Treasury
auctions for government securities, which use sealed-bid rules, are often described
as “Dutch” auctions even though they do not use a descending-price mechanism as
in flower auctions. Treasury securities auctions, however, are strategically equivalent
to a particular descending-price (Dutch) auction. An auction mechanism is
strategically equivalent to another auction mechanism when bidders’ incentives, the
identity of the winner, and the final sale price are the same for both. Some auction
rules, however, may be operationally easier to carry out. For example, bidders can
mail in responses for a sealed-bid auction, while English auctions require bidders (or
their agents) to gather in the same place.
A descending-price (Dutch) auction, under certain conditions, is strategically
equivalent to a sealed-bid, first-price auction. A rational bidder calculates what the32
item for sale is worth to her. If a bidder bid her value, however, she would make
zero gain in a first-price auction because the price paid would exactly match the
item’s value to her. A rational bidder therefore shades her bid downwards, trading
off a larger gain (value minus price paid if she wins the auction) against the33
possibility that she would lose the item by lowering her bid. The same calculation
applies to both the sealed-bid, first-price auction and to the descending-price auction,
so the two auctions may be considered strategically equivalent. An English
(ascending-bid) is strategically equivalent to a second-price sealed-bid auction when34
bidders know what the object up for auction is worth to them.
Imperfect Information: Common Values and the Winner’s Curse.
When bidders have imperfect information about the value of an item, an English or
ascending-bid auction may force better-informed bidders to reveal valuable
information to less well informed bidders.35 For example, the value of an antique


30 For details, see [http://www.flora.nl/en/AboutFloraHolland/Auctioning/Pages/
default.aspx].
31 Falling-price auctions were invented to avoid Napoleonic-era taxes on rising-price
auctions.
32 “Rational” here only means that a bidder has internally consistent preferences and is self-
interested, so that the bidder can compute bids that best achieve her ends.
33 Bidders here are assumed to keep the object and cannot resell it. In technical terms,
values are private and independent. When values are linked or common, results can differ.
34 When the value of objects are linked, such as when an auctioned object will be resold,
some of these equivalences may not hold. For example, in an English auction among
wholesalers, bidders may learn about an object’s resale value in retail markets through the
bids of others. In that case, the equivalence of an English auction to a second-price sealed-
bid auction would not hold.
35 For example, in eBay auctions, informed bidders often use “sniping” strategies, in which
bids are submitted seconds before the auction closes. This prevents less informed bidders
from gleaning information from the bidding behavior of an informed bidder. For details, see
(continued...)

may depend on who made it, how rare it is, and on who owned it before. A
knowledgeable bidder, who may know more about an item’s value, will attempt to
use bidding strategies that conceal private information.
The value to bidders of some auctioned items may be linked. For example, the
value one energy company places on an Offshore Continental Shelf (OCS) lease that
would allow exploration and extraction of oil and gas will correlate to the value other
energy companies place on the same lease. Different companies might have strengths
and weaknesses in exploration and extraction techniques, so the value of the lease
will not be the same to each company. Any company that got the lease, however,
would sell oil and gas on the same world markets. Auctions that sell items whose
value to bidders is correlated are called common-value auctions.
Bidders in a common-value auction may have different indications of an item’s
value. For example, many energy companies may have private information about the
geological structures of areas covered by an OCS lease. A company holding an OCS
lease in a nearby area that had run seismic tests might have more precise information
about those geological structures, and thus would have a more precise estimate of the
value of the OCS lease up for auction.
When bidders have imperfect signals of value, bidders with overly optimistic
signals are likely to win auctions. Such bidders, however, will suffer losses because
the true value of the item is less than their optimistic estimate. This is the winner’s
curse: such auction winners would have been better off losing. Sophisticated bidders
in common-value auctions shade their bids downward to account for the winner’s
curse. Sophisticated auction designers release as much information as possible about
an item’s value so that revenues are not reduced by bidders who shade their bids
downwards.36 That is, bidders with better information bid more aggressively.
Revenue Equivalence. Certain auction mechanisms, as noted above, are
functionally equivalent to certain other auctions run using different rules. The
auctions discussed above (Dutch, English, first-price and second-price sealed bid) in
theory deliver the same profits to bidders and the same revenues to sellers.
Moreover, any (independent private value) auction that awards the item to the highest
bidders and attracts the same pool of participants also in theory provides the same
profits to bidders and the same revenues to sellers.37 This result, known as the
Revenue Equivalence Theorem, implies that to affect expected revenues requires
changing who participates in an auction. For example, minimum bid rules can raise
expected revenue, but may lower an auction’s economic efficiency. Experimental


35 (...continued)
[ h t t p : / / www.e s ni pe .c om] .
36 John H. Kagel and Dan Levin, Common Value Auctions and the Winner’s Curse,
(Princeton, 2002).
37 This result assumes that bidders are risk-neutral when values are uncertain. Auction
revenues will not in general be equivalent when bidders are risk-averse. Elmar Wolfstetter,
Topics in Microeconomics: Industrial Organization, Auctions, and Incentives, (Cambridge,
1999), pp. 186-188; For a more detailed discussion, see Paul Milgrom, Putting Auction
Theory to Work, (Cambridge, 2004), ch. 3.

research has found that some expected auction equivalences hold, while others do
not.38 Experimental testing of revenue equivalence of auction mechanisms is an
active research area.
Multiple-Unit Auctions
Auctions in which multiple units are sold simultaneously are more complex than
single-unit auctions. Computing optimal bidding strategies in multiple-unit auctions
may be complex and difficult. Designing multiple-unit auctions so that government
revenue is maximized and so that scarce resources are likely to be assigned to those
who value them the most can be challenging. The federal government, however, has
successfully used complex multi-unit auctions to allocate electromagnetic spectrum
for wireless communication and related uses.
Manipulation and Demand-Reduction Strategies. Bidders in some
situations can benefit by strategically withdrawing bids on some items in order to
lower prices on other items.39 The logic of this strategy is analogous to standard
monopoly or oligopoly pricing models. A monopolist or a firm with some market
power can raise profits by reducing output below the level that would prevail in a
competitive market. Just as entry by new competitors can reduce the market power
of existing firms, auction designs that encourage many bidders to participate can limit
the effect of demand-reduction strategies.
Complementarities. Auctions that sell complementary goods can be
extraordinarily complex.40 Items are considered complements when groups of items
are more valuable than the sum of individual items. For example, take-off rights
from a specific airport that a government might auction off will be more valuable if
the airline can obtain landing rights at a different airport. Federal Communications
Commission auctions of electromagnetic spectrum involve complementarities
because a license in one geographic area may be more valuable to a
telecommunications firm that holds a license in an adjacent area.
Complementarities among troubled assets may complicate current reverse
auctions implemented as part of TARP. For instance, some assets backed by sub-
prime loans were often linked to collateralized default obligations(CDOs), which
provided something akin to insurance to investors holding those assets. In that case,
the asset backed by a pool of sub-prime loans and the associated CDO would share
a strong complementarity, which could affect behavior in a reverse auction.
Dutch Auctions with Multiple Units. The U.S. Treasury, as noted above,
uses a multi-unit Dutch auction mechanism to sell government securities to primary


38 David Lucking-Reiley, “Using Field Experiments to Test Equivalence Between Auction
Formats: Magic on the Internet,” American Economic Review, vol. 89, no. 5, pp. 1063-1080.
39 Lawrence M. Ausubel and Peter Cramton, “Demand Reduction and Inefficiency in
Multi-Unit Auctions,” Univ. of Maryland working paper, July 27, 2002, available at
[ h t t p : / / www.cr amt on.umd.edu/ paper s 1995-1999/ 98wp-dema nd-r e duct i on.pdf ] .
40 Paul Milgrom, Putting Auction Theory to Work, (Cambridge, 2004), ch. 5.

dealers. The federal government and some corporations have used reverse Dutch
auctions for some procurements. Treasury auction mechanisms can, under certain
circumstances, be vulnerable to manipulations related to the demand-reduction
strategies discussed above.41 The U.S. Treasury and many entities that use auctions,
however, have developed methods designed to detect or mitigate manipulation
strategies.
Uniform-Price vs. Individual Price Auctions. In some multiple-unit
auctions, such as Dutch auctions, all successful bidders pay the same price. Such
auctions are called uniform-price auctions. In other auctions, such as first-price
auctions, in which successful bidders pay their bids, different bidders pay different
prices for identical items. Such auctions are often called discriminatory or multiple-
price auctions. A 2002 International Monetary Fund report found that 10 of 18
advanced industrial countries surveyed used uniform-price auctions, and 15 of 1842
used multiple-price auctions. The U.S. Treasury claims that a uniform-price
auction raises slightly more revenue than a multiple-price auction because it reduces43


winner’s curse risks.
41 Yongdong Shi and Xianfeng Jiang, “Manipulation in the Treasury Auction Market,”
working paper, July 26, 2004, available at [http://ssrn.com/abstract=575441].
42 International Monetary Fund, “Guidelines for Public Debt Management,” Aug. 4, 2003,
Table 1.2, available at [http://www.imf.org/external/pubs/ft/pdm/eng/guide/080403.htm].
43 Paul F. Malvey and Christine M. Archibald, Uniform-Price Auctions: Update of the
Treasury Experience, U.S. Treasury Study, Oct. 1998.