The Minneapolis Fed has been calling for increased market discipline
and greater use of market data in the bank supervision process for about
20 years. I think the timing and content of this conference, and its predecessor
hosted by our co-sponsor, the Federal Reserve Bank of San Francisco, is
right on target because we need to go beyond general policy discussions
and good intentions, and address the practical advantages and problems
raised by increased use of market data in supervision.
I assume supervisors will take Chairman Greenspan's concluding
remarksnamely, that supervisory systems must increasingly
rely on market discipline and data to avoid obsolescenceto
heart. In my opinion, the chairman views favorably supervision
of LCBOs (large complex banking organizations), and communications
about such supervision, that include analysis of market data.
I would like to add some detail in support of the general
sentiments of the chairman. In particular, I will explain why
I think market data will be valuable to supervisors in their
assessment of banking organizations, and I will offer specific
suggestions for the use of such data. Finally, I will conclude
with some thoughts on the future agenda in this area.
It seems that a fair number of supervisors think economists come to discussions
of market data and banking supervisory issues with strong prior beliefs,
namely that regulatory data and assessments are useless and market data
tells us all we need to know. This straw man does not encompass my views.
I firmly believe that both market and regulatory data contain valuable
information, even though both are "noisy" and though assessments
based on such data inevitably are imperfect. That is, the available data
may not fully and accurately reflect the risk taking, and therefore potential
difficulties, of a banking organization at any point in time.
For example, price signals from fixed-income instrumentssubordinated
debt issues, for examplereflect the supply and demand
for particular bonds in addition to participants' views of risk.
More generally, interpretation of movements in market prices
and interest rate spreads is far from a trivial task. But on
the other hand, supervisors may not fully appreciate the risk
in a bank's derivative strategy or understand how poorly an
organization is executing a merger. In short, relying exclusively
on either supervisory or market data leaves some uncertainty
in our assessment of financial condition.
The amount of uncertainty associated with regulatory assessments
will vary over time and by institution. There is no reason to
be defensive about this, given the inherent complexity and opaque
nature of many banking activities as well as the probabilistic
nature of risk. There is not an easy way to determine the chances
of a "bad" outcome occurring in the myriad business
activities of a large, complex banking organization.
I feel comfortable raising the specter of uncertainty with
regard to Federal Reserve responsibilities because it occurs
frequently in my job. The setting of monetary policy is characterized
by doubt. No single set of data can inform monetary policymakers
as to the correct action at Federal Open Market Committee meetings.
In response, monetary policymakers review as much data as possible.
Chairman Greenspan, as many of you have probably read, reviews
a plethora of information on the economy to which he gives varying
weights depending on the circumstances. The point is, we do
not and should not throw information away, and this is particularly
true when acquisition and analytical costs are reasonably low.
The logic for this broad approach is fairly straightforward. Each additional
piece of relevant information we review may reduce the uncertainty that
policymakers face. Greater certainty allows us to act sooner and be more
precise in our response. I should stress one caveat: I do not advocate
collecting information for its own sake. Rather, we review information
that has a reasonable connection to our task, preferably a link validated
by experience and empirical research.
The approach policymakers take toward monetary policy seems
directly applicable to banking supervision. The more relevant
information a supervisor has, the smaller the variation around
the resulting risk assessment and the faster and more precise
the supervisory response can be. I would think that supervisors,
with their reliance on data and analysis, would be sympathetic
to improving their assessments through additional, and potentially
Some skeptics of using market data in bank supervision point
to the differences between the markets' and the supervisors'
assessments as a justification for not taking market data seriously.
But market data are well suited to complement regulatory assessments
because of its particular attributes:
- Market data are generated by a very large number of participants.
- Market participants have their funds at risk of loss. A
monetary incentive provides a perspective on risk taking that
is difficult to replicate in a supervisory context.
- Unlike accounting-based measures, market data are generated
on a nearly continuous basis and to a considerable extent
anticipates future performance and conditions.
- Raw market prices are nearly free to supervisors. This characteristic
seems particularly important given that supervisory resources
are limited and are diminishing in comparison to the complexity
of large banking organizations.
I would think that market data would be of value whether or
not they confirm supervisory assessments. Imagine a case where
a supervisor believes an institution is becoming less risky
but still has some doubts. At the same time, the trends in market
data suggest that market participants believe the institution
is posing less risk as well. To the degree that a separate,
independent source confirmed the supervisory assessment, I would
assume that the supervisor would become at least a little more
confident of his view.
But what about the case where the two signals are at odds?
Even here, uncertainty eventually may be reduced. Consider,
for example, a case in which market data indicate a change in
the condition of a banking organizationa deterioration,
saywhile regulatory assessments reflect no such change.
Reliance on the supervisory assessment may lead to claims that
markets "overreacted." Such conclusions are simply
erroneous and are akin to judging a treacherous activity "safe"
because you survived improbable odds. It could very well be
that the supervisor underreacted. At the very least, market
information should encourage the supervisor to take a second,
There are several venues where supervisors can make effective
use of market data, including assessments across institutions.
The Federal Reserve has increased its efforts to allocate limited
supervisory resources across districts and institutions. Those
institutions posing the greatest risk would presumably receive
the greatest attention. Market signals are a natural complement
to regulatory data in making these relative comparisons. Because
relative comparisons eliminate general market noise and benefit
from forward-looking data, cross-institution comparison could
represent a potentially effective use of market data.
Descriptive statistics based on market data also offer supervisors
alternative methods for assessing the overall condition of the
banking industry. For example, supervisors can look at the mean
or median market-based measure of risk for the LCBO group to
determine the current condition of these institutions and how
it has changed over time.
Credit quality is another area where market data could enhance
supervision. Traditional asset quality ratios have well-known
failings, such as their backward-looking nature. Banks' internal
ratings can be more forward-looking, but comparing ratings across
institutions is a challenge. Supervisory ratings of asset quality
and asset classification offer a potentially good source of
information but also suffer from a lack of timeliness. Market
data, in contrast, is forward-looking, timely and could facilitate
comparison across institutions for at least part of the organizations'
portfolios (for example, exposures to publicly traded firms).
The bottom line of my talk is that supervisors should include
market data as input into assessments of regulated institutions.
In cases where market data confirm supervisors' original view,
I would expect faster supervisory response. As noted, market
data could also act as an excellent "check" on supervisors'
views and actions. And I think we know that actions of supervisors,
in particular, are subject to second-guessing, particularly
after the "forbearance" periods of previous banking
crises. Because of forbearance, some influential observers of
banking policy argue that market data, with its highly visible
and public nature, would make an ideal tool to accelerate, if
not force, regulatory action. Links between prompt corrective
action (PCA) and some market measures are advocated, for example,
in order to preclude forbearance.
One might be tempted to dismiss such PCA reforms as uninformed
and unlikely to become law, but the original PCA proposals often
provoked the same reaction. Rather than dismiss such proposals,
it seems prudent for supervisors to proactively monitor and,
where appropriate, respond to market signals, or at least be
in a position to explain why such reactions did not occur.
Market data could also be of value in discussions with bank
management, particularly when markets and supervisors take a
dimmer view of the institution than its leaders. The third-party,
objective nature of market assessments reduces the ability of
an institution to characterize supervisors' assessments as off
base or isolated. Moreover, banking organizations behave in
many ways that suggest market prices provide information. It
would be difficult for these institutions to later dismiss market
data as noise.
This conference will conclude by establishing an agenda for
future work in the market data arena, so let me add my thoughts.
We will leave here with a continuing need to determine what
data we should monitor, how we should obtain the data, how we
should analyze the data to distill its risk assessments and
how we should interpret changes in market signals.
Many declarations that call for greater use of market information
stop at requiring additional disclosure. Now is an opportune
time for policymakers to go further and articulate in detail
their expectations for use of market data by supervisors.
In that vein, I would expect progress toward clarity and direction from
the appropriate senior management throughout the Federal Reserve System.
Such direction could lead to incorporation of market data into formal
supervisory guidance and processes. The benefits of market data seem real
enough and the costs of increased use low enough to justify a higher priority
to the study and use of market data even with the data's flaws and with
our limited knowledge.