View original document

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


DAVID W. M U L L I N S , J R .

R efbrm
of the

A Speech By:
Vice Chairman
Board of Governors of the Federal Reserve System

This speech was delivered on May 19, 1993 in Washington, DC at
a Catalyst Institute conference on the U.S Treasury securities
Catalyst Institute conducts and disseminates objective research
that improves public policy and business decisions affecting
financial markets and institutions and that improves the
performance of government. Catalyst Institute is an independent,
U.S. domiciled, nonprofit corporation.

Dean Witter, Discover & Co. provided funding for the printing
and dissemination of this speech.

Any opinions expressed are those of the speaker
and not necessarily those of Catalyst Institute.

Catalyst Institute 33 North LaSalle Street Suite 1920 Chicago, llllnola 60602
Telephone/ 312 541 5400 Facsimile/ 312 541 5401

I appreciate the opportunity to talk about the process of
reform in the Treasury markets. I shall talk primarily about
the progress of reform over the last couple of years and the
effort that went into the joint report 1 . In some sense this is
old news, but the relevance of it might be in discussing the
reform process -- to give a sense of the institutional and
political context within which these decisions are made;
which might be useful input to your discussions and further
research on this issue.
In August of 1991, John Gutfreund of Salomon Brothers
called the regulators and the press and began the revelations
of Salomon Brothers' abuses. To put it bluntly, all hell broke
loose in Washington. In a matter of weeks, we began the
first of what has been nine testimonies before Congress.
At the first hearing one congressman, denouncing the abuses
with characteristic understatement, called them financial
treason. This was the sort of attitude in the early days here.
It was alleged that Salomon Brothers had rigged the Treasury

The Department of the Treasury, the Federal Reserve, and the
Securities and Exchange Commission (collectively, the Agencies)
undertook a joint review of the government securiUes market in
September 1991. The joint report is the product of that review.

markets, submitted fraudulent bids, blatantly evaded
regulations and cornered the market in several issues,
creating a manipulative squeeze. This came against the
backdrop of a number of years of insider trading cases, of
financial institution fraud -- the Keating's S & L case and
BCCI, for example. It was argued that this time, fraud had
reached directly into the pockets of every American
taxpayer. It didn't matter that, very narrowly speaking, in
this incident money was deposited in the pocket of the
taxpayer, but the concern was the blatancy with which the
regulations had been violated. One concern was that these
sorts of episodes and crises tend to result, or at least
historically at times have resulted, in bad legislation and bad
There were two fundamentally different
approaches to addressing this episode.
The first approach was:
"Let's get tough on regulation.
Let's layer on every regulatory measure we can think of."
For example, there is a rule that no bidder can buy over 35%
of an auction. Some suggested lowering that to 25% and also
applying it not only to dealers but to the customers of
dealers. The effect would be to limit the participation of
people who would like to buy a large volume of Treasury
securities at a high price. More generally this first approach
was: "Let's impose regulation. Let's wire up this market
and these participants and monitor every move to snuff out
When one looks back to the political heat of the late summer,
there was considerable danger that a brute force regulatory
blanket approach might be adopted. However, during the
rush to judgment, the Agencies thought it would be a good

rush to judgment, the Agencies thought it would be a good
idea to take a time out, to step back and see if we could
think out a coordinated approach to the issue. From this
coordinated approach, I think a second alternative emerged.
This alternative was derived from the observation that the
U.S. Treasury market is the deepest, largest, most liquid
market in the world with trading volume ten or twelve times
the average of the New York Stock Exchange. Yet the
number of trades are only roughly l/50th of the New York
Stock Exchange, so it's a large sophisticated wholesale
market. There are relatively few issues. The issues are
homogenous and have none of the idiosyncrasies of stocks.
The question was, "How was it possible that a couple of
people can corner such a market. Why isn't the enormous
force of competitive pressure in such a market sufficient to
preclude any manipulative behavior?"
It seemed to us that the reason was that even though there
was a broad competitive force in the secondary market,
bottlenecks existed which impeded the flow of this
competitive force into the primary market. What were these
bottlenecks? They included impediments to bidding at
auctions by those other than primary dealers; the manual
nature of the auction as opposed to an automated auction;
and the five percent deposit requirement for non-banks or
non-primary dealers. There also was a lack of market
information in that only primary dealers, at that time, had
access to interdealer broker screens. All of this tended to
narrow direct participation. Moreover, the multiple-price
sealed-bid auction discouraged bidding because of the
winner's curse, and the primary dealership system had

membership requirements which limited the number of
primary dealers to a very small number. The effect of all of
this was to take the powerful force of competition in the
secondary market and funnel it through the narrow, artificial
aperture of the primary market, through a few dozen primary
dealers. As a result, a competitive secondary market was
transformed into an oligopsony in the primary market.
The solution was to try to dismantle systematically these
impediments and open up the primary market. In some
sense, this competitive alternative was diametrically opposed
to the regulatory approach. The regulatory approach would
encumber the market by imposing tighter constraints. The
competitive approach did not involve imposing artificial
constraints on the market but rather tried to enhance
competition in this market. These were the two alternatives
we faced: smother this market with heavy-handed regulation
or open it up to the healing sunshine of competition. Of
course, it was not quite so clear-cut. There was ax¬Ľo a
legitimate need to strengthen surveillance and enforcement
It's interesting to ask the question, "Who was on the side of
competition and efficiency, and who favored regulatory
suffocation?" The answer is even more interesting. In the
end, there was no one who argued forcefully for encumbering
this market and trying to overregulate it. Indeed, in all the
Agencies there was remarkable unanimity on the philosophy
of the approach we adopted, even in Congress. Very early
on, a few people in each of the Agencies suggested proposals
such as lowering the bidding rule and the like. But, even in
Congress, there was a lot of force for improving the market,

from people like Congressmen Pickle, Neal and Markey, as
well as Senators Riegle and Dodd, and other members of the
Senate banking committee. There was interest in seizing this
opportunity to make some progress.
One of the interesting aspects of this report is that this effort
could have dissolved into a fundamental fight among
Agencies based upon different philosophies. Nonetheless,
there is a remarkable unanimity in the basic philosophy of
the joint report. With dozens and dozens of major and minor
issues in the report, the Agencies agreed on all but a handful.
One can count on one hand the issues on which we disagreed.
This is a remarkable result, and it differs from some other
areas in which the Agencies are set off against each other.
What explains this result is that we've been through a
number of crises in recent years, and the Agencies have
developed a level of professional expertise which is
impressive in this area. As a result, we did try to follow this
competitive approach in the report.
The basic premise of the recommendations is that the most
powerful and effective force in enhancing market efficiency
and in reducing the potential for abuse is the force of
competition. The report sets out a coordinated set of
recommendations which seek to open up the primary market
to broader participation and to dismantle systematically the
barriers which prevent the competitive force in the
secondary market from impacting the primary market.
I'll run through the basic recommendations. The first
recommendation was to open up bidding to non-primary
dealers and to automate the bidding process. The secondary



market is highly automated. It is fair to say that the primary
market was not highly automated. It is clear that automation
facilitates surveillance. If Salomon Brothers submitted false
customer bids, under an electronic system it would be easy
to check. The automated process is underway. It is true
that, so far, this has only cut a few minutes (about fifteen
minutes) off the delay from the time the bids are submitted
to the time the results are announced. Ultimately, as
everyone gets wired up and the software gets completed to
compile all the bids, the delay will be shortened. Treasury
intends not to require everyone to submit bids electronically.
But as we move forward, we might consider whether once the
automated process gets going we should require people who
want to submit bids manually to do it a bit early. Then the
potential exists to reduce this auction time lag very
Another major component of the report was to open up the
primary dealer system. Although this system has worked
well for over 30 years, it was designed for a period of a
developing market. The market has changed. It has
developed. Some people suggested abolishing this system.
We thought that would be a bit drastic. The primary dealer
system is enmeshed and imbedded in a whole set of
institutional arrangements. For example, many state and
local treasurers can only deal with primary dealers.
Treasury-only mutual funds can invest in Treasuries and
repurchase agreements (repos) only through primary dealers.
So, abolishing the system might have caused institutional
disruptions. Moreover, we noticed that when one looks
around the world virtually every issuer of securities has some
dedicated group which commits capital and expertise to

distributing and making markets in that issuer's securities whether it's called an underwriting group, a selling group, or
a syndicate. So, perhaps it's a natural phenomenon.
What we did instead of abolishing the system was to
eliminate the rigid barriers to membership. We kept some
sense of responsibility to make markets and bid in auctions.
We eliminated the membership requirements and replaced
them with clear-cut objective capital requirements. Anyone
who meets the capital standards is allowed in. We also
eliminated the one percent market share. One was required
to have at least one percent of all customer trades in the
secondary market in order to be a primary dealer. This
limited the number of dealers to 100 if they were equally
distributed. Since they're not, it limited it to effectively a
small number. It also produced some strange behavior people trading back and forth artificially to meet the one
percent barrier. The logic here is, "Let's remove the
artificial constraints and let the market determine the
appropriate structure of the primary dealer system." It will
take a while to evolve. We've seen very little change so far.
The fourth major component of the report was to explore
new auction techniques to replace the multiple-price sealedbid auction.
Here, not only did the report suggest
experimenting with a uniform single-price auction, but also
an ascending price, descending yield, iterative open auction.
We were originally going to call that an "open outcry"
auction, but we were fearful the agricultural committee
might claim jurisdiction. So, we called it an "open iterative
real time" auction system. Obviously, the single-price aspect
combats winner's curse, which discourages less sophisticated

investors from bidding, and also encourages bidders to shave
their bids. The open iterative real time auction alternative
was only put in to stimulate discussion. But it seemed to
many of us, whenever it's physically possible to get all
bidders in the same room, the technology of choice seems to
be an iterative bidding process, not sealed-bids. One sees it
in auctions of art and used cars, specialists opening stocks on
the New York Stock Exchange, and in the pits in Chicago.
Computerized auction pricing allows the electronic
equivalent of having everyone in the same room. This open
iterative process would have the advantage of combatting
surprises and collusion. It may be even cheaper to collude
under a single-price sealed-bid auction than it is under a
multiple-price auction. That was the motivation for putting
an alternative on the table for discussion. The object of all
these auction proposals was to open up the auction process,
reduce the cost of the Treasury's financings, and reduce the
chances of manipulative abuses.
Conceptually what we were trying to do with the auction
process was to design an approach that reduces three sources
of uncertainty or dead weight cost. The first source of
uncertainty was the time delay between the time when one
puts in a bid and finds out the results. This is just pure
uncertainty. You don't know how much you own. You don't
know how much to hedge. You're vulnerable to whatever
information comes in during this time delay. Automation is
the key to reducing this uncertainty. When I came to the
Treasury, the delay was two hours. Then, we reduced it to
roughly one hour, and now it's a little less than an hour.
Bidders raise their bids to be compensated for the risk in this


period of uncertainty. The shorter the delay, the more one
can reduce this risk.
The second source of uncertainty was the wiiuier's curse
which the single-price auction was designed to address.
The third source of uncertainty was the incentive to collude
and surprise a market in a sealed-bid auction. The iterative
open auction process was designed to deal with this.
In recent years, if one compares the yield Treasury pays in
the auction to the yield in the when-issued market for the
same issue prevailing at the time of the auction, Treasury has
been paying about 3/4 of a basis point higher than the yield
in the when-issued market at the time of the auction. That
3/4 of a basis point reflects the cost of the risk of dealing and
bidding in the auction.
The biggest payoff should be the simplest to implement - to
automate the auction and reduce the delay from an hour or
two hours to a matter of ten or fifteen minutes. By reducing
that period of uncertainty, one wonders whether we won't be
able to save on the order of 1/4 to 1/2 a basis point. Of
course, 1/2 a basis point certainly isn't much unless you
auction over two trillion a year, in which case, it ends up
saving well over a hundred million dollars. Next, the
question is, "Are there incremental benefits in moving on to
the single-price auction or the iterative auction?"
Conceptually that's the way we viewed it. That extra 3/4 of
a basis point really benefits no one. The bidders don't
benefit since it just compensates them for their risk. Our


objective was to try to squeeze this dead weight cost out of
the system.
A final and important component of the report was the set of
recommendations, agreed upon by all the Agencies, designed
to enhance surveillance and deal swiftly and effectively with
abuses. The basic approach was to rely on low-cost marketbased mechanisms, rather than some broad-based regulatory
mechanism. The specific approach was based on the
proposition that episodes of manipulative behavior can be
detected through close monitoring of market data; prices,
repo rates and the like. These statistics should be sufficient
without resorting to a broad-based reporting apparatus, to
report all trades or positions. Once anomalies are identified
by the New York Fed, the SEC can launch an investigation.
If manipulative squeezes are acute and protracted, Treasury
can reopen the issue and engage in supply management. The
threat of Treasury reopening should be sufficient to deter
abuses. Of course, we fully recognize that the authority to
reopen involves a cost which is priced into the market. But
this cost should be offset by the benefits from reducing the
chances of manipulative squeezes. The decision to engage in
Treasury reopenings needs to be understood in the context of
the political setting and the alternatives.
The other
components of the recommendations would, through time,
result in broader participation and reduce the chances of
abuse. There was a strong feeling that we needed to come up
with a credible effective approach to combat abuses then, in
the political heat of the time.


disruptive and damaging to the market. That's why everyone
signed on to the reopenings. We all agreed on these
components. Taken together they constituted a pretty
thorough modernization of this market.
There were a few areas in the recommendations where the
Agencies did not agree. Some of the Agencies preferred
large position reporting and audit trails, and authority for the
government to impose transparency2 requirements. The
Federal Reserve Board favored none of the above. The fact
that there were differences on these issues reflects I think
different institutional responsibilities.
It is wholly
understandable that the Agencies on the front line of
surveillance and enforcement will support measures to
minimize the cost to the Agencies of surveillance and
enforcement. We think about maximizing global efficiency
from the perspective of the system as a whole. It's natural
that agencies directly involved in surveillance on the front
line prefer more data to less. They argue that it is better to
have a cop on every street corner in order to save tne
incremental costs of going out and collecting evidence when
there are abuses. We believe it's cheaper to have a fast
response call to 911, since there are very few episodes of
abuse and will be even fewer with these changes. The
question is, "Why impose costs on the entire market?" We
weighed the costs versus the benefits and thought the
imposition of broad based reporting requirements was not a
good bargain for the taxpayers.

Transparency is the availability of Umely, accurate price and volume
information to market participants.


weighed the costs versus the benefits and thought the
imposition of broad based reporting requirements was not a
good bargain for the taxpayers.
We were concerned that large position reporting would
increase the direct cost on every large trader. It might have
an indirect cost, in that large investors value their privacy
and confidentiality in financial dealings and are not
interested in revealing their finances or trading strategies.
They might withdraw from the market.
The same is true with the audit trail and transparency
requirements. Transparency always sounds very good. How
can one argue against transparency? How can one argue for
opaqueness, to put it the other way? Of course, one
traditionally looks at the central bank to argue for
opaqueness, and we did not disappoint. It is not that we
dislike transparency. We think that it should be developed
by market participants, not imposed by the government. The
concern is the authority behind transparency. This authority
potentially involves the ability to require the full panoply of
reporting requirements and could fundamentally redesign a
market. It might change the basic character of the market
from a wholesale market into more of an exchange market.
The Treasury Market has evolved as a wholesale market -over-the-counter trading, with razor-thin bid-ask spreads for
large sophisticated investors. It's very efficient. If people
are worried about fraud, it's hard to understand why dealers
will stand there and trade either way at a tiny spread. It's
very efficient, in part, because it is an over-the-counter
market. The market is not encumbered by some of the

institutional arrangements and regulatory requirements
found in other markets. It is similar to markets such as the
foreign exchange market.
The concern was that if the government started imposing
these restrictions it could interfere with the efficiency of the
market. The stakes are quite high. If enough people
withdrew from this market because of reporting
requirements and the like, resulting in an increase in the
average cost of Treasury's borrowing by 1 basis point, this
translates into well over two hundred million a year in
increased Treasury financing costs. Absent a comprehensive
reporting apparatus, it may be more expensive to go out and
gather the data to find abuses. But this incremental cost
must be weighted against the potential efficiency cost. The
evidence did not convince us that this was a good bargain for
the taxpayer or the markets. So we did not support some of
these additional requirements. I would again stress that
these areas of disagreement encompass only a small number
of the issues that we dealt with. We agreed on virtually all
the issues as well as the basic philosophy. In the legislative
process disagreements are often magnified. Nonetheless, the
report was generally well received in Congress, and the
recommendations have been substantially implemented.
I would like to talk for a moment or two about topics for
future research. The Catalyst report is a very useful addition
on a variety of fronts. There are a few areas in which more
research would be useful.
The first is the area of reopening policy and strategy. The
idea is very simple: Treasury should reopen an offering only

when there's an abusive manipulative squeeze. The question
is, "What is that and how could one detect it?" In our view,
many squeezes are natural phenomena. People guess wrong.
A big financing need develops in the corporate market, and
market professionals need to short certain Treasury issues to
help the distribution process. One can't just look at special
repo rates. There are natural patterns which occur. In fact,
if you line up repo rates in event time with a zero date as the
date of the announcement of new issues, when one issue goes
from "on the run" to "off the run," you can see recurring
patterns. One guess is that there is probably no free lunch
in those patterns. The patterns in the financing rates simply
reflect and offset anticipated price changes associated with
the roll of "on the run" securities to "off the run" securities.
Perhaps the Treasury should look for other evidence -- for
example, market failure before considering reopening. This
is a difficult area worthy of research.
The work in the Catalyst volume provides useful insight.
Much has been learned recently about financing rates and
financing patterns in the theory and practice of the Treasury
market. It is important; not that we distrust the current
Treasury, or the past Treasury, or even the next one, or the
next after that, but because having this authority to reopen
is something which has the potential to be priced into the
market. As such, it is worthy of continued research.
Moving to other research topics, it would be interesting if
someone took a look at the primary dealer system as well as
the whole issue of transparency, a concept that has a wide
and deep following. Everyone, including us, 6xtols the
benefits of full dissemination of information. There may be

some confusion with international issues and access to
Japanese markets. Transparency sounds great. However,
economists raise interesting questions about transparency.
"Why should those who do their homework and get the
information be forced to give the results of that work for
free? We don't require car dealers to disclose to customers
the prices of all recent sales. We expect people to shop
around. Why shouldn't the market rule here? Why should
the government mandate transparency? Would transparency
requirements reduce the incentive to invest in information
gathering? Should we eliminate patent protection and
require all new drugs to be available by all producers? Some
would argue that this would make the world a better place by
reducing the costs of new drugs and making them more
widely available. Of course, we would have fewer new
drugs." These are economists' questions. You shouldn't
misunderstand me. The lack of transparency in this market
was a significant impediment. I think we've made a lot of
progress, and the industry voluntarily has decided to move in
the direction of increased transparency.
We fully support these developments. However, it is a
broader topic because it's going to be a major issue in capital
markets around the world. It would benefit fiom some
thought as to what conceptual guide we might use so that
when we discuss developing transparency requirements, we
can think of doing so in ways which fit different markets
instead of just imposing these restrictions. I would argue
that this is one topic which could use some rigorous analysis,
because implicit in the decision to mandate transparency
requirements is an enormous potential authority to
reorganize and redesign markets in ways which could

adversely affect market efficiency. It might also be
interesting to explore whether there is any way to get an
empirical estimate of the indirect cost of imposing reporting
or transparency requirements, assuming someone can find
some clean experimental setting in which to make the
The final issue, the big issue, to which academic research has
made a major contribution, is auction technology. It seemed
to me that the dealers in the industry responded favorably to
virtually all of the aspects of the report, including the
changes in the primary dealer system that promised to
virtually eliminate any franchise value associated with
membership. But, it seemed to me, they didn't like the
auction proposals - either the single-priced or the open
iterative auction techniques. Why? They said, it's confusing.
Bureaucrats and Congress had little problem in
understanding it. I suspect market participants should be
able to grasp it.
What one suspects from the dealers' responses is that the
current auction technique is very important to the system
and to them, more important than the structure of the
primary dealer system or other aspects, the manual aspect,
of the auction. One suspects that the price discriminatory,
multiple-price, sealed-bid auction is a major source of the
benefits they receive in the current system. I might add that
the benefits are in some sense competed away. So don't jump
to the conclusion that just because there are benefits,
somehow they're carrying away windfall gains. But it seems
to me that some of the benefits in the system must be


generated and distributed to dealers through this auction
The current auction encourages people to bid through the
primary dealer system, either directly or indirectly through
the when-issued market. Not only does this funnel business
to dealers, but they extract valuable information on demand
in the process. This information is useful in their own
dealings. Indeed, it is reported that dealers often do not
charge large customers foi submitting their bids at auctions,
at least dealers do not charge customers directly. So while
changes in the auction methodology may in theory be
beneficial, such changes could disrupt and require alterations
in the way costs and benefits are distributed among dealers
and customers. Nonetheless, even though many dealers were
critical at the beginning of the process, my sense is that the
experiments have gone reasonably smoothly, and dealer
criticism seems to have diminished.
The important decisions on auction technique lie ahead. Let
me mention a couple of immediate issues with respect to the
auction. The first one is how should Treasury decide what to
do at the end of this experiment? They've outlined some
criteria. It would be useful to think of a rigorous research
methodology along with decision criteria well before the end
of the experiment. It might be useful to get some group of
professionals involved, practitioners and academics. It would
be interesting to see if one could actually estimate and
observe a shift in the demand curves in the auction data as
predicted in the literature on single-price auctions.
Fundamentally, Treasury is going to have to decide what to
do - whether to implement the single-price auction or

whether to expand the experiment to other issues or whether
to abandon it.
A second point has to do with automation and its interaction
with the Dutch auction. An automated, single-price auction
has, in my view, the potential to increase direct auction
participation by large customers. In the old system, a large
customer could not place an anonymous bid easily. There
was a manual auction system. Customers had to worry about
winner's curse, so they tried to spread their bid among many
dealers in fear of showing their hand. Automation will
facilitate large customers placing direct bids anonymously, at
least, anonymously with respect to other market participants.
With a single-price auction, they can avoid winner's curse.
I think this may change and improve the competitive aspects
of the auction. To achieve these benefits requires not just a
single-price auction but an automated single-price auction,
and it may take a while to develop and encourage
participation by large customers. Thus, in evaluating the
Treasury experiment, it's going to be important to recognize
that there are really two experiments. The first is the
manual single-price auction, and the second is the automated
single-price auction.
Where do we go from here? Is it really worthwhile to
experiment with some sort of open iterative auction process,
a transparent auction process, which compared with the
sealed-bid process may be less vulnerable to collusion? I
think the biggest potential benefit is the reduction of the
time delay. We should get a significant percentage of the
return from that alone. But we should evaluate the other
proposals, as well.

To sum up, it's ironic that the most serious abuses in the
history of the Treasury market, the Salomon Brothers
episode, served as the catalyst to engage in a ground up
refurbishing of the Treasury market. It has been an exciting
process; it is a continuing process. So far the process seems
to have been enormously successful even though there may
be much left to be done. Often, crises lead to bad legislation
and bad regulation. That didn't happen this time, in my
view. It's a credit to the professional expertise of those
involved here - primarily in the front line Agencies, the
Treasury, the SEC, and the New York Fed - that this crisis
was transformed into very substantial progress.
The work to date is only a beginning. I think the Catalyst
studies add to our knowledge. We would invite your
continued input into the important process of designing and
implementing improvements that both enhance efficiency
and reduce the chances of abuse in this important market.



David W. Mullins, Jr. is Vice Chairman of the Board
of Governors of the Federal Reserve System and has
been a Member of the Board since 1990. Dr. Mullins
received his Ph.D. in finance and economics and
masters degree in finance from the Massachusetts
Institute of Technology, and his undergraduate
degree from Yale University. Prior to becoming a
Member of the Board, Dr. Mullins served as Assistant
Secretary for Domestic Finance at the Department of
the Treasury. Prior to the Treasury, he was a
Professor of Business Administration at Harvard
University, G r a d u a t e School of Business
At the Department of the Treasury, he was
responsible for Federal finance, financial institutions
policy, securities market regulations, corporate
financial policy, government sponsored enterprises
policy, and synthetic fuels projects. He played a
major role in the development and implementation of
the Bush Administration's legislation on the S & L
crisis. Upon leaving the Treasury, he received its
highest honor, the Alexander Hamilton Award.
He has been a consultant to a wide variety of firms
and governmental agencies and served as Associate
Director of the 1987 Presidential Task Force on
Market Mechanisms (known as the Brady
Commission). He has published articles in leading
economic journals on a wide variety of topics in
financial economics including cash management and
banking as well as corporate finance and capital
market topics such as cost of capital, dividend policy,
mergers, stock issues and bond issues.