Stress Testing: A Discussion and Review (Morning)

Stress Testing: A Discussion and Review (Morning)


Hi, I’m Bill Bassett. I’m a senior associate
director in the division of financial stability at the Board of Governors and
I’m also a member of the model oversight group which is responsible for the
development and implementation of the models used in the photos of stress test
program I will be popping up on stage occasionally today to introduce our
speakers and our three sure to be lively panel discussions first I’ve been tasked
with reiterating the ground rules around today’s program and the related
interactions we’re on TV all proceedings are on the record
including all speakers and questions and comments that are recognized by the
session moderator on the other hand all the informal conversations and comments
during the conference including at receptions meals and coffee breaks are
off the record unless otherwise established as such if you wish for a
conversation to be on the record you should establish the ground rules with
the participants by attending this conference you consent to be
photographed video recorded and audio recorded by the Board of Governors of
the third Reserve System and the Federal Reserve Bank of Boston seems like
everybody’s staying to get things started I’d like to introduce Lisa rue
Lisa is a senior associate director in the division of banking supervision and
regulation where she oversees the cross portfolio supervision function that
includes the Federal Reserve Supervisory stress test program risk identification
and analysis shared national credit program and business technology risk
lisa chairs the model oversight group and is also a member of the allistic
operating committee please welcome Lisa to the stage thanks Bill
I’m hoping this picture will go away at some point and I think somebody pointed
out I’d look very relaxed and this picture is because it was taken before I
actually started working on stress testing and personal in the Fed and that
explains the real you know the my relaxed mode good morning I am as build
it just introduced me I’m Lisa to you I’m speaking today as a co-chair of the
conference organizing committee along with Bill Bassett who introduced me and
on behalf of the organizing committee I would like to welcome everyone to this
conference it is aptly titled stress testing a discussion and review I would
we own many thanks to president Rosen gran and staff at the Federal Reserve
Bank of Boston for hosting this conference so before I actually go into
the my prepared remarks which is always dangerous Joan zone when I started to go
off the script but I you know I was going over my prepared remarks last last
night just thinking about what I’m gonna talk about today and then kind of it
started to come my mind started wonder and started thinking about what led me
to to be involved in the stress testing among many other things I could have
been doing and then I concluded that perhaps me being a near a life time fan
of New York Mets may have something to do with it so so let me explain you know
what I’m trying to get to I know there are several people on the audience who
know me is wondering where’s she going with the statement and I I thought I
told her not to talk about sports and and and all those things but so let me
see let me see if I can bring it back stress testing is obviously hypothetical
exercise we can envision a potential stress scenario however is remote I see
how the banks remain resilient to this we Mets fans however no bad things will
happen regardless of you know whether you start the season with four great
starting pitchers one of whom happened to be a former Cy Young winner why not
something will always go wrong and we think throughout the season what else
can go wrong so I’m thinking that being a lifetime Mets fan prepares me well for
being a male perhaps close to a lifetime stress tester here at the Federal
Reserve so it has been 10 years since the Federal Reserve and other US banking
regulators first used stress testing as we know it today
to publicly gauge the resilience and health of the the largest and most
complex banks in the United States much has changed since then as we are all
aware the banking sector is much better capitalized than it was then stress
testing has also become a crucial part of the US banks supervision for large
bank and is now an established regulatory vernacular not only in the US
but also globally we do have a speaker in the first panel who is representing
Bank of England and their stress test and that is the testament of the how
stress testing has become an accepted norm in the regulatory community here in
the US we have just completed ninth annual stress testing it’s hard to
believe it has been nine years but this has been the ninth year and and and at
least we would like to think that has been another successful year of a stress
testing yet as some of the key architect of that first dress s who who are
attending this conference can attest some of the discussion topics today
echo many of the questions and discussion we had in 2009 so let me just
go over some of the we are still grappling with hopefully
which will foreshadow some of the discussions to come today those
discussions include is a stress testing an effective policy tool for all times
not only during a crisis but also during peace times can and should the stress
test be used for counter cyclical purposes and if so in what ways and if
not what other alternatives would they be what is an appropriate role for banks
internal models how transparent and dynamic should a stress testing be what
are the unintended consequences and how do we mitigate those unintended
consequences and finally what has been and would be the effect of a stress
testing I think these are the all questions we’ve been asking clearly it
is not easy question these are the difficult questions and which is a
precisely reason why the debate endorsed you today your answers will reflect a
particular lens through you through which you look at this questions so for
example the question of the optimal dynamism and transparency which is a
topic for the second panel of today will likely differ depending on whether your
primary focus is about the effectiveness of the stress testing for ensuring
resilience to all kind of dynamic changes externally or us your primary
focus is a battement of a certainty of the capital requirements to better
facilitate planning forward likewise the appropriate role of banks internal
models in capital regulation would depend on your views on on the
importance of an independent and consistent in your stick and whether how
idiosyncratic risk the banks face should be captured over the past few years the
Federal Reserve has incrementally increased the trance
pairen see of his stress testing in past few years we have we held a something
called a stress testing model symposium here in Boston it provided a forum where
the industry representatives and regulators could come together to
discuss best practices earlier this year we published the aluminum or details
about our stress tests but not full details and while many responded
positively about certain aspect of the disclosure some including my former
colleagues thought we went too far and we shouldn’t go as far as we did and
others however as for even more and a few even requesting the full recipe
clearly there’s a spectrum of responses the jury is still out and hopefully we
will be enlightened by the discussion will have during our panels to answer
some of those questions the Kondos conference is another step
forward towards great transparency in public policy making and seeks to
benefit from active discussions and discourse on this topic with that in
mind I’m particularly pleased to see such diverse groups attending this
conference so let me just go over eight groups of people who are attending
today’s conference their academics banking analyst their bankers the
representatives of the Community Development and public interest groups
congressional staffers somewhere here in the audience
regulators both domestic and foreign and finally doesn’t also area students
my not so modest hope today is that the student attending today’s conference
will be so inspired by today’s event and discussions that they will consider
public service for their future career one can always hope today’s three panels
consists of speakers with diverse experiences and
perspective that should enrich the discussions some of the panelists are my
former colleagues who had a hand in shaping the stress testing in the u.s.
in 2009 and afterwards some bring an academic rigor and analytical rigor the
honed in academia and others bring various empirical perspectives from the
banking industry the investment community and finally the public we are
particularly indebted to authors and co-authors of the three discussion
papers there nicely frame the three broad questions we posed to them I very
much look forward to open and active discussions among the distinguished
speakers but also with the audience following the footsteps of a long and
fruitful tradition and policymaking finally I cannot possibly close my
remark without giving some shoutouts to to numerous and dedicated colleagues
around the Federal Reserve System who have made the stress-testing it is today
and finally to Tim walkins who’s over there
without whom we could not have this conference today and thank you and I
look forward to discussions thank you Lisa for kicking off today’s
program and I guess you could say that four world series later bringing up the
Mets in the city is much less of a stress test our next speaker really
needs no introduction for this group Federal Reserve and FOMC chair Jay Powell
has served on the board of governors since May 2012 and began his current
four-year term as the chair in February 2018 prior to joining the board he was a
visiting scholar at the Bipartisan Policy Center a partner at the Carlyle
Group an assistant secretary and under secretary of the Treasury under
President George HW Bush unfortunately chair Paul could not be
here today but he has sent along these taped remarks good morning everyone and thank you to
president Rosen grande and the Boston Fed for hosting this conference the
Federal Reserve is strongly committed to stress testing as a cornerstone of our
bank supervisory and financial stability missions stress testing is perhaps the
most successful supervisory innovation of the post-crisis era but if stress
tests are to continue to serve their critical function they will need to
evolve in the years ahead to keep pace with the ever changing financial system
as they have since the first round of stress tests in 2009 before looking to
the future let me recall a bit of history a little more than 10 years ago
the United States and the world teetered on the brink of economic catastrophe
what was urgently needed was a way to restore confidence in the financial
system a daunting challenge neither the banks were the regulators nor the public
had a reliable sense of the strength and resilience of our major banks the
announcement of forward-looking stress testing results in May 2009 helped
restore confidence and stabilize banks by providing a credible and independent
picture of their finances these original stress tests evolved into
the comprehensive capital analysis and review or seek our program which has
served to institutionalize capital planning by firms and supervision by the
Fed as a forward-looking endeavor since 2009 large banks have added more than
800 billion dollars in common equity capital giving them a much thicker
cushion to deal with losses banks have gotten much better at assessing and
managing their risks effectively tracking commitments across their
organizations anticipating capital needs and planning for different scenarios the
stress tests of the future five and ten years from now will need to continue to
ensure that banks remain able even in a severe downturn to provide the credit
that households and businesses depend on as financial institutions in the
financial system of stress testing we’ll need to keep up
when the next episode of financial instability presents itself it may do so
in a messy and unexpected way banks will need to be ready not just for expected
risks but for unexpected ones thus the tests will need to vary from year to
year and to explore even quite unlikely scenarios if the stress tests do not
evolve they risk becoming a compliance exercise breeding complacency from both
supervisors and banks we might also inadvertently encourage the development
of a banking system where over time all banks would look much alike rather than
the banking system we want and need one with diverse institutions with different
business models we simply can’t let these things happen the purpose of
today’s gathering is to help us think about how to ensure the tests continue
to foster a dynamic banking system financial stability and a healthy and
growing economy we have invited a wide range of participants fellow regulators
bankers analysts academics and community groups with diverse perspectives because
we do not claim a monopoly on knowledge or wisdom we have invited many who have
disagreed with us in the past I strongly believe that diverse perspectives and
healthy debate will sharpen our thinking and lead us to better answers I’m sorry
I’m not able to be with you today and I particularly want to thank all of the
speakers discussants and other participants for helping us grapple with
the challenges of ensuring that tomorrow’s stress tests remain as
effective and vigorous as they are today thank you you program bike is also the author of many
articles on value at risk stress testing and credit derivatives and with that
I’ll turn it over to you Mike okay good morning we’re having our own audio BT
audio-visual stress test today it’s a plan to part of the program it’s not not
unexpected so well let me add my welcome I’m Mike Gibson I’m the director of
supervision and regulation at the Federal Reserve Board in Washington and
I’m pleased to be the moderator for the first panel really appreciate all the
analysts being willing to speak and share their thoughts on stress tests as
a policy tool let me just go over the background rules for the panel first
we’re gonna have the presentation of the paper by Andrew metric Andrews a
professor at Yale University and the paper is co-authored with Greg Feldberg
who’s the director of research at the Yale program of financial stability then
we’ve got three distinguished panelists and I’ll just introduce them now the
first panelist speaker will be Charlotte gherkin she’s the director of the
cross-sectoral and insurance policy at the Bank of England second will be
Dennis Kelleher he’s the president and chief executive officer of better
markets and our third panelist will be Brian Lee he’s the chief accounting
officer and controller at Goldman Sachs the panel is gonna the the paper
presentation will lead off then each of the discussants will make their remarks
and then we’ll have plenty of time at the end for questions I’ll ask some
questions in my role as moderator and then we’ll also
an opportunity to have questions from the audience so please think about your
questions now because we’ve built in plenty of time for discussion and we’re
really interested in having a great exchange of views today so with that I’m
gonna turn it over to Andrew metric to make the presentation up here oh there
it is all right all I have to do is click clicked for me I’m seeing okay good morning i’m andrew metric from
Yale and I have this paper with my Yale colleague Greg Feldberg who’s sitting
down there the opinions expressed in this paper are not necessarily the
opinions of Yale University or the Federal Reserve System that was joke
just kidding we don’t have to give a disclaimer when
we talk okay all right so what I’m going to do today is a little bit of an
introduction I mean it’s kind of why we’re here what it is we were trying to
do and then really we have a series of observations and implications of those
observations for stress testing and of the big picture of stress testing we are
none of the things that we have today that I’m going to talk about here today
are specific policy proposals where we think you should do exactly one thing or
another thing or models if you look at our paper you won’t find any models in
there at all there is no math whatsoever and we did that on purpose we really
were trying to take a 30,000 foot view of stress tests and and think about
where we are and where we’ve come so where does that begin we were asked to
think about stress tests as policy it begins really thinking this this
thinking about what the stress test was meant to do when the SCAP was first used
in 2009 and it was very much a wartime exercise then we were really quite
desperate and we people were concerned that the banks were completely insolvent
and not just under capitalized but insolvent and if you have insolvent
banks and the government’s not going to support them then we’re gonna run on
them and so they needed they were either going to need to convince people that
they were well capitalized or they were gonna have to go get capital right then
that’s the wartime exercise of a stress test but now we have peacetime stress
tests we’ve just did our 9th and question is how are we doing
are there any principles that we can use going forward that would help to inform
that and they’re really the first thing to recognize is that we really are
anchored a bit in the way we think about stress tests still in this time of
crisis so a lot of the conversations that I have with people about stress
tests are stressful so you talk to the people and they’re nervous that tomorrow
there’ll be a run on the banks and that is kind of a feeling that that we still
have and I think one thing that that we had to do when we were writing this
paper and thinking through these things was to really try to situate ourselves a
little bit differently imagine what if we had invented this these stress tests
purely as a peacetime exercise that didn’t have that anchor oh what one
thing I also just want to mention here that is quite important as an
introduction is we intend to say nothing about the average level of capital that
is needed at banks this is much more about the changes over time so the the
level of capital debate is one where we don’t have great cost-benefit models
we’re never gonna have a perfect answer it’s actually quite frustrating
great minds can disagree on both sides but that’s not actually what the stress
tests are about sometimes things about the level the average level of capital
over the course of the cycle creep into our debates about stress tests but
that’s really they shouldn’t we kind of separate that out it’s much more
variations around that average and how they happen over the cycle and across
banks that is the the right topic for for what’s true what we should be
thinking about when we think about stress tests okay so I have five
observations Greg and I at various times had nine and three and five we’re down
to five okay so we have five things that we thought when we started that they
were going to be pretty uh non-controversial and we each had five
observations that we were sure were non-controversial that were
non-overlapping so so it turns out it’s not bad
see to figure out things that you do think are really non-controversial but
but now this is this is as close as we’re gonna get at least for us okay so
the first is individual banks no more than supervisors do about the bank’s
idiosyncratic risks but supervisors no more than banks do about system-wide
risks so the point here is to start us along the path of thinking what are the
comparative advantages of running stress tests in a certain way and if we can
think about those comparative advantages what what direction might we want to
move in with any individual policy changes and this observation is just the
first part of it I think nobody will argue with which is that how whatever we
think of the risk management practices at banks at any point in time they’re
gonna have a better insight into their own idiosyncratic risks than anybody
else would outside of their business the second part of it sometimes you’ll get
people saying that they they understand the get banks not saying that they
understand the whole landscape better than the supervisors do but I don’t
think that’s true the supervisors are seeing data from all over the place and
they’re really able and have a comparative advantage at knowing for
example if there are a whole lot of concentrated risks in this system at any
point in time so that’s our first observation second observation is in
bank supervision transparency is a double-edged sword and here actually I’m
not just talking about transparency of the models which is going to get a lot
of attention today I think the transparency of the inputs that goes in
but also the transparency of the outputs what we say about the banks and how well
they did on the stress tests publicly overall transparency is kind of like you
know apple pie and baseball not the Mets but other baseball mecan’t have this
just like a great thing it’s an all-american thing we want to be
transparent how could you be against transparency in any particular way
that’s a very securities market way of looking at things in in the world of
banking actually my colleague Gary Fortin has a has a whole research line
about banks as secret keepers that basically saying that banks can do what
they do because they are in in many ways opaque that in fact opacity is a feature
not a bug of the way we design and create and manufacture safe assets so
and there’s there’s a lot of research that that really supports the idea that
actually transparency when it comes to the safe debt side of the world is very
very different animal than transparency on the securities market side so in
terms of outputs one thing here is just to recognize starting with a tendency
and not with a specific policy proposal but a tendency to say that it’s just
having more transparency is really not not necessarily a good thing on that
side on the input side you know it’s an it’s a it’s an obvious kind of statement
of Vice Chair Quarles has said you know he understands you don’t get about the
answers to the test before the test or the questions so people understand that
but I think also some of the debate about transparency and I’ll get to this
in a few minutes is really a debate about something else that the
transparency ends up being a substitute for complaints really about what the
effects of the test are so I’m probably if use like half my time already how
long am I supposed to go for I have ten more minutes
all right so I better speed off cuz I got a couple more of these things I’ll
have one more slide after this I am well prepared okay so so next even with
perfect models accepted by everyone thanks and their regulators will
rationally disagree on the optimal speed of adjustment after a stress test so I
think a lot of times when we end up fighting about the models and the models
being good the models being bet it’s really just a fight about what are the
what are these models going to get used for and how fast am I going to have to
if I am a bank raised capital and it’s important to note and to have always in
the background of the argue of of the discussion as any mediator would do a
recognition that it is costly to have to change capital plans and capital
levels quickly that’s a costly thing even people who say that it’s not costly
to hold capital to have more capital in a bank I think would agree that changing
it quickly is costly there’s a long corporate finance literature on that so
banks won’t want to do that that’s privately costly to them on the other
hand the banks are never going to internalize the externalities of their
failure on all the other banks and so to the extent that raising your capital
levels quickly will lower your probability of failure the the
regulators are gonna want that to happen faster it’s always going to be the case
so that even if we could agree here exactly the right models and here are
exactly the right scenarios we’re going to have this tension and we sometimes
I’m almost certain when we’re arguing about models we’re really arguing about
our fast do I have to change after those models so the fourth observation is in
peacetime regular stress testing is critical to keep risk managers and
supervisors eyes on the ball so what is it not this is this is okay yeah we want
to keep our eyes on the ball what is it not it’s not necessarily in peacetime
about very very quickly going and raising capital so when I think about
what would be very very useful in a stress test in historical crisis
episodes if we had them a couple years before the crisis it would be to be very
informative about risk and about things that we were missing in a panic or in an
incipient panic you sometimes need to capitalise things very fast and moving
very fast will be necessary to restore the confidence that enables you to
function outside of a panic I can’t think of examples like that where it’s a
oh god if we just very very quickly gone and done that often that action by
itself is enough to destabilize things that actually the speed is not
necessarily such a wonderful thing in in peacetime okay and then fifth migration of
financial activity away from banks without appropriate oversight of the non
Bank financial sector will weaken financial stability say what does that
have to do with stress tests migration must be mentioned anytime you mention
anything about bank capital it is a really of its the reason we had the last
crisis was we had a tremendous amount of migration it wasn’t just the banks and
going forward whenever we think about what particular policies we’re going to
do for banks we need to be thinking about what those will have what effects
those will have on migration and that needs to be part of the planning process
so stress tests can be particularly good at this if they do their job well they
will be able to see what is going on in the non-bank part of the world faster
than we would in the absence of having stress tests it could be a very
important tool okay I think I have about five minutes or so and so I have three
implications so all all or three or four four implications I’ll do them fast the
first and these are very this Maps almost exactly lease it to the questions
that you asked earlier I was very pleased so you gave your questions maybe
because we knew them before I don’t know the first is macro-prudential verse
micro-prudential so again focusing on comparative advantages here what is it
that the stress tests can do since the since the supervisors are really able to
look across and go is going to have a comparative advantage at looking across
there’s a huge value to using the stress test as much as possible for macro
prudential things so the counter-cyclical buffer is a good
example what they do in england with with using the stress test to help
inform counter cyclical policy strikes us as being a great use of the
comparative advantage of stress tests second on speed of adjustment i’ve
already talked about this a lot i do think that a tendency here is that we’re
very focused on the level of capitalists necessary we might want to have higher
average level capital at thanks i do think that it is worth thinking about
whether or not if we were to make the speed of adjustment to the newly
required capital higher or lower after a stress test
slower if it weren’t an immediate thing that actually a lot of the arguments and
debates and anger about the models and how secret the models have to be would
go away and that actually we could make the stress tests you didn’t use this
word in the paper but I wanted to use it Nastia even worse scenarios even tougher
even even tougher scenarios even wilder ideas we could learn a lot more from
them and we could and we would be able to do that without the costs that are
currently seen as the most the worst costs by industry on transparency I
think Greg and I make almost everybody unhappy I there should we’re gonna let
mark who has a whole paper on it I do most of the work but for our short 30
seconds to say that overall here the tendency I think should go in the
direction of less transparency across the board less transparency of the
models less transparency of all of the inputs less transparency of the outputs finally coming back to this theme of
peacetime and wartime part of the the reason that perhaps some of the things
that when we said them to our friends when I say them to other people work on
stress tests they look at me funny is because we do continue to think about
the stress test as a wartime animal and that actually in wartime you want to do
stuff differently and we would need to then think about what is as many have
asked us the transition between peacetime and wartime bold ideas like
having some kind of DEFCON clock like they have a Defense Department turn out
not to work all that well since it probably wouldn’t be useful for
maintaining the calm in the economy for the Federal Reserve to say we are now at
DEFCON 2 in the banking system it seems that we actually have a reasonably good
balance right now and the balance that we have right now is a variety of system
risk type exceptions that exist in a variety of places in our regulatory
system that effectively act as the wartime triggers when we can do things
that we weren’t able to do in peacetime and that this balance seems to be a
pretty good one and that outside of that we should be thinking about what it is
it that the stress tests or any regulatory any kind of regulatory
measure can bring us in peacetime to help us to prevent the next crisis and
learn the huge amount of information gaps that we have that are usually the
things that precede these crises that’s it almost on time okay thank you very
much Andrew that was great I think you’ve definitely raised many
probably not all but many of the issues that we’ll be talking about today the
first discussant is Charlotte kirkin good morning
and thank you very much to the Federal Reserve for inviting me to this
conference at the Bank of England we’re currently reviewing our approach to
stress testing and will publish a revised approach document later this
year so while anytime is a good time to talk
about stress testing now is a particularly good time and I’m looking
forward to learning a lot today and ideas about advancing our approaches
you’re advancing yours Greg and Andrews paper has given us much to think about
and I’ve had to narrow down the number of topics I can kind of give reflections
on now and I’ll offer a few thoughts on the why
what and how aspects of stress testing and policy I’ll comment on the macro
versus micro-prudential policy objectives the use of stress testing for
quantitative capital requirements and more qualitative improvements in
business understanding planning and risk management
and also pick up on some of the points in the paper about model risk both banks
performing stress tests and for policy makers well around the Y the paper
reminds us why bank stress testing became such a useful tool for policy
makers during the crisis and it does remain a powerful tool to ensure that in
a financial or macroeconomic shock banks can absorb shocks rather than amplify
them the paper discusses the macro mic policy objectives in connection with
this aim the macro prudential objective in the paper the banks can continue to
lend to the real economy in stress times certainly guides our approach in the UK
the concurrent stress tests we perform under the annual cyclical scenario on
the largest seven UK banking groups have an explicit link to macro-prudential
policy through the counter-cyclical buffer that andrew mentioned the aim is
to adjust systematically the severity of the scenario in relation to the
financial policy committees risk assessment informed by risk indicators
and thereby make the changes in the scenario severity of the stress test
fairly predictable to reduce the volatility around requirements which
Greg and Andrew highlight as the design issue and Vice Chair Qualls referred to
in his speech last October the Bank of England stress tests include a severe
global shock as well as the UK shock that directly affects the
counter-cyclical buffer so that we can gain sufficient coverage for the various
different business models of the banks in the test to meet micro-prudential
objectives the Prudential Regulation Committee is also informed by other firm
specific analysis including the firm’s own individual capital assessment in
order to decide whether to take micro-prudential action through a firm
specific PRA buffer as far as what stress tests are trying to achieve a
stress tests have helped both banks management and regulators bring
confidence to their financial resilience measured through risk weighted and
leverage ratios but over time as we’ve learned and
developed our approach the stress tests have enabled banks and ourselves to
identify and understand vulnerabilities both from a macro micro perspective and
enabled stress testing to become more of a pre-emptive policy tool which enables
action to be taken before a stress rather than being used to make running
repairs during one in relation to observation one in the paper I think
there is a feedback loop between a firm’s understanding of its
idiosyncratic risks and the input provided by supervisors by way of an
example in 2017 our policy committees wanted a more rigorous assessment of
consumer credit in the annual cyclical scenario the PRA had found that lenders
were reducing interest margins and risk weights associated with consumer credit
while beginning to increase lending to high risk market segments we wanted to
explore whether the fallen defaults over the preceding five years reflected
factors that should be discounted when assessing how loans would perform under
stress whether banks were attributing too much of the improvement to
underlying improvement in credit quality and too little to a macroeconomic
environment of sustained employment growth and low interest rates as well as
the growth of interest-free credit card offers while the results certainly had
implications in terms of capital requirements the more the more enduring
outcome we were aiming for was a shift in bank’s assessment of the risks
they’re running noting the balance of stress testing between quantitative
capital requirements and broader vulnerability assessment or risk
management one type of stress tests can achieve
only so much we introduced another a concurrent stress test that speaks more
to the objective of improving banks and our understanding of vulnerabilities and
strengthening risk management we ran what we call the biennial exploratory
scenario for the first time in 2017 being exploratory in nature and each
each one is designed differently according to the risks the
explored in 2017 it was designed to have a longer scenario than the five years we
use for the annual cyclical stress test it looked at the scenario at a scenario
with some long-term trends that could affect bank sustainability and ability
to generate capital organically the trends in scenario included long-term
low interest rates and the implications of competition particularly from FinTech
the firm’s cost income ratios but perhaps the FinTech aspect didn’t need a
long-term scenario after all we learned quite a bit ourselves about running this
different sort of stress test and about the relationship between banks business
functions here the strategy function as well as frontline business units finance
and risk the latter Orsi being more involved in the annual cyclical scenario
test we’ve already flagged our intent for 2020 ones exploratory scenario to
conduct a climate stress test for financial institutions to help bring
into the mainstream climate risk management I said I would touch on model
risk which Greg and Andrew discussed in the context of transparency and the risk
of banks focusing their efforts on gaming the feds models I agree with
their assessment that a model monoculture could lead everyone to
missing a tail risk because they observe happened and they run up to the
financial crisis and model risk management was an area we focused on in
our qualitative review in 2018 we noted that all the banks participating in the
stress test demonstration increased awareness of the need to implement
effective model risk management and some banks had made good progress against the
PRA s expectations but other banks needed to make substantial improvement
to raise the management of Mullah risk to a standard required for stress
testing and the majority of the banks are needed to increase their board and
senior management understanding of limitations in their key stress testing
models where material adjustments are applied banks management needed to
consider whether the judgments well supported with the relevant
accounting standard changing last year for us to our first nine which also
relies heavily on models decision-making by firms let alone policy makers is even
more exposed to model risks whether from input data assumptions or methodologies
at the Bank of England we’re trying to do as we do as we say or do as we tell
others to do in developing our own models for stress testing one area we’re
working on is modeling systemic risk amplification aspects for example though
there are major challenges in the availability and robustness of data as
well as how we currently scope our stress testing framework which restricts
how banks respond to the stress for example they need to maintain a lending
market share during the stress period we’re at a very early stage in using our
models for policy and rather than presenting a single specific scenario we
take a range of plausible scenarios and outcomes using what ifs to communicate
potential systemic risks to the committee’s in increasing the
effectiveness of our stress tests for policy we’ve been looking at three
themes to take forward one is taking a more holistic approach to scenario
design leveraging supervisory expertise more and conducting sensitivity analysis
exploring variants of the scenario second we’re reviewing our data
requirements not dating our modeling strategy and third we’re working on our
external disclosure strategy both to guide policy makers decisions and to
provide enhanced feedback to the banks but with the expectation of the
productive discussions today I I fear and expect or perhaps hope that the
number of themes could well get longer thank you Thank You Charlotte
the next speaker is Dennis Kelleher thank you good morning
first I’d like to thank chairman Powell and vice chairman Quarles for holding
the conference and for inviting me to participate and importantly for
understanding and acting on the need not only for transparency with the public
but for going outside the usual suspects and including alternate views and even
dissenting voices as chairman Powell recently said disagreement is not only
healthy but important before turning to the paper for those of you who have
asked and many of you have who’s better markets better markets was founded just
after the dodd-frank law in 2010 it’s a non-profit independent DC based
organization that promotes the public interest throughout the economic and
financial policy making process in Washington we participated in more than
200 rulemakings many of the related lawsuits testified in Congress numerous
times issued lots of reports we lots of information on our website if you really
want to know more turning to the topic of the panel I’m
gonna offer some quick observations on policy related to the issues raised by
Andrew and Greg and their thoughtful and thought-provoking paper that I encourage
you to read because there’s too much to say in too little time if you want to
know more about better market views on the details of stress test in the Fed
specific proposals again they’re on our website at www.americanoutback.net is
really what we’re talking about when we’re talking about stress
the last crash is going to cost the United States more than 20 trillion
dollars in lost GDP which better markets did a study on called the cost of the
crisis in the cost of the crash those dollars don’t reflect the human
suffering all across this country as tens of millions of Americans lost their
jobs homes health care and so much more if stress test fails fail and banks
don’t have sufficient loss-absorbing capital then the public is again going
to suffer the consequences and get the bill for that failure and they should
have a seat at the table and their interest should be the center of any
discussion which I will return to at the end of my remarks a second observation
would be don’t snatch defeat from the jaws of victory u.s. stress tests are
the gold standard stress tests have work exceptionally well you don’t need to be
told that they have been used at one of the most perilous times in our country
since the Great Depression of the 1930s that real-time live stress test of
stress test restored the confidence of the financial system and the public and
then remember the opposite happened in Europe they – at that time employed
stress tests virtually all the banks passed but the tests weren’t considered
rigorous lack transparency and left substantial capital holes and thus
Europe stress tests lack credibility that’s the risk here losing credibility
and that leads to my third observation I’d suggest that stressed has to be
thought of primarily as credibility tests for the Fed that credibility has
already taken some hits some are already referring to them as well on the way to
being no stress stress test others have observed that stress tests have gone
from confidence builders to mere capital ejection mechanisms some have noted that
what was originally thought of as a process to create a capitol floor has
created a capital ceiling we should ask if a test where 100% of the test takers
pass every time with flying colors is a valid credible test such results are
usually a red flag we need to ask if it is wise that a bank failing the stress
test should be avoided at all costs entire new words and processes are being
developed and deployed to avoid even saying the word fail
when Goldman Sachs and Morgan Stanley failed the test by falling under capital
the capital requirements last year the Fed didn’t fail them rather it invented
a euphemism quote conditional non objection close quote we need to think
about how these actions changes in mindsets are fundamentally challenging
the credibility of the tests and yes the credibility of the Fed I’ll turn to my
paper with my fourth observation as you’ve heard Andrew and Greg have
suggested an analytic framework organized around the concepts of
peacetime and wartime regulation while theoretically interesting the likelihood
of correctly knowing when it is peacetime is impossible and being wrong
even by a little bit will be dangerous if not catastrophic we don’t have to
speculate about that we have definitive proof let’s take a minute to speak
honestly and openly about the years before the 2008 crash we had failures of
judgment of historic proportion by policy makers elected officials and by
regulators including prominently the Federal Reserve Board its staff and its
many affiliates like the New York Fed they were mostly wrong dead wrong with
grave consequences a tragic groupthink blinded most to
growing risks and the coming catastrophe we’re not talking about ancient history
like the 1990s think back to the celebration of the Greenspan years at
the 2005 Jackson Hole conference even the mildest indirect academic dissent
was not just greeted with disagreement but disdain ridicule and name-calling
peacetime reigned no need to even question the prevailing dare I say
wisdom it was peacetime everybody agrees markets know best the least regulation
was best the biggest financial institutions in the world could
self-regulate in self-police none of them would ever make outside take
outside risk that might endanger the viability of their firms or reputations
for brilliant sophistication and risk management now none of that is to cast
aspersions those are just the facts they are facts that no one saw coming even
the brightest of the bright the unavoidable conclusion is that we all
must be deeply humble about our ability to determine when a crisis is coming
or frankly even when we’re in the middle of one that means that artificial
constructs of peacetime and wartime and the false comfort the former instills
should be rejected as unworkable and should not be a guide for policy and
certainly not policy deregulation because it’s supposedly peacetime for
example in the years before the 2008 crash exactly when did we move from
peacetime to wartime when subprime loans a synthetic CDOs reached a certain level
when the Bear Stearns hedge funds collapsed in 2007 when they shot their
books earlier in 2007 when Goldman flipped to the big short when Northern
Rock failed to ask these questions as to reveal that they aren’t answerable even
in hindsight and certainly not in the middle of peacetime going to war time
maybe war time however I would suggest that when it comes to financial
regulation you are always at war which is my next observation policymakers in
general and regulators in particular should accept that conflict and
financial regulation is inevitable healthy and indeed a sign of success the
paper says that stress test exercise quote remains a more or less contentious
process close quote which the authors attribute to quote bankers bristling at
the perceived loss of control of a basic capital planning decisions close quote
and they note that it is important that there will always be conflict here quote
but I’m sorry but largely limit that to quote the speed of capital adjustment
after a stress test close quote that limitation is too limited
regulators and bankers are always an inevitably at war even if they don’t
always feel it that’s because it’s not a hot war but a steady-state cold war you
are financial institutions pressing the limits at all times as they profit if
not bonus maximize we have to stop pretending that we agree on the
financial reform goals and mostly on the ways to achieve them ending too big to
fail and workable resolution plans are prime examples effective stress tests
and sufficient loss-absorbing capital are others regular regulators and
bankers bring and should bring different perspectives to these issues
which lead to different views and inevitably disagreement and that’s why
the authors noted that the quote banks have said thank you
close quote for greater transparency provided by the Fed but the banks have
nonetheless quote asked for even more close quote of course they did that’s
their job it is indirect but it isn’t direct
conflict with the job of the regulator’s that will that will inevitably result in
tension and disagreement and frankly that’s a good sign
I want to make a related observation it is most important to understand that
evil evil actors in or evil motives by the private sector are not required for
any of these observations or concerns it is the nature of markets and financial
firms individually and ultimately collectively it is the siren songs of
profit maximization and competitive pressures as Upton Sinclair said so well
quote it’s difficult to get a man to understand something when a salary
depends upon is not understanding it close quote
that’s why banking regulators and supervisors as well as oversight
regulation and enforcement generally are so critically important will always have
conflict a concluding observation the concluding section of the paper entitled
quote balancing the cost and benefits for banks and regulators close quote
raises some very serious concerns it begins with a troubling question that
frames the discussion of the entire section quote have we struck the right
balance between the needs of banks and their regulators parenthetically and
taxpayers in the stress test policy close quote taxpayers are only mentioned
in the parenthetical never to appear again bankers needs and wants and
contrast are very prominent to the point that the conclusion is quote a banks
board should have more power over capital planning in peacetime while
regulators should be able to intervene as war approaches close quote in
addition to rejecting the peacetime wartime framework I would suggest that
focusing on balancing the so-called needs of banks and regulators turns the
world upside down taxpayers and their needs should be the center of the
question and now and answer I would propose asking the
question do stress test serve their purpose in protecting the public
taxpayers the financial system and our economy from undercapitalized over
leveraged too-big-to-fail banks that pocket profits and bonuses in peacetime
and shift losses to taxpayers in Main Street in wartime and let’s look at the
evidence banks had to be bailed out in 2008 by taxpayers because they didn’t
have the capital to absorb their loan losses why was that because those banks
spent down their capital on stock buybacks and dividends in the years
before the 2008 crash and indeed think about this up to and after the collapse
of Lehman on September 15th that’s right after the collapse of Lehman on
September 15th they were still ejecting capital the banks intentionally in
needlessly reduced their capital cushions even as the crisis was upon
them even as they were taking big losses starting in 2007 even as Bear Stearns
collapsed even as Lehman crashed the regulators said nothing and did nothing
all the clearly ominous events and warning signs of 2007 didn’t slow the
capital injections in closing as I said I would at the start taxpayers must be
central to the entire discussion and above the quote needs of banks close
quotes and I reiterate the question that should frame policy making do stress
tests serve their purpose in protecting the public taxpayers the financial
system and our economy from under undercapitalized over leverage too big
to fail banks that pocket profits and bonuses and at all times and shift
losses to taxpayers in Main Street in a crisis that is what is at stake and
stress tests and losses of absorbing capital thank you Thank You Dennis our third panelist is
Brian Lee good morning so I’ve already learned something this
morning eyes I’ve often wondered as to how my life has become consumed by
stress testing and Lisa gave me the answer except I have the double whammy
you’ll be in a mess and Jets fan so it’s quite the burden that I have also
punished my children with but they’ll have to carry that one so thank you Mike
thank you for having me here I’m honored to be amongst the panelists today I’ll
also say I’m speaking on behalf of myself based off my experience running
the seek our process and managing capital at a large financial institution
I really enjoy Greg and Andrews paper which is balanced and well researched
then addresses some subjects that I have been focused on for some time
today I’ll focus my remarks on three main points first stress testing should
indeed involve in peacetime second bank models can be powerful micro-prudential
tools third coherence is critical and stress testing within stress testing in
the broader capital framework I’ll be quick on my first point I wholeheartedly
agree with Greg and Andrew that stress testing is one of the most powerful
policy and risk management tools we have but with common equity for large US
banks more than doubling since the crisis and with significant advances in
sophistication of stress testing practices it is time to to recalibrate
the stress tests in peacetime I think that goes for both how the process is
conducted and the assumptions within the stress test which I’ll address in my
next two points my views are also reflected by my belief that US stress
testing has evolved from a capital raising to a capital allocation tool
even if that wasn’t the intention the market implications are real and should
be taken into account in adjusting design in calibration on my second point
that bank models can be used as micro-prudential tools I also completely
agree with Greg and Andrew that the role of bank run stress tests an determining
their own capital needs could be in chanced that just uh I had a few
thoughts my own practical experience to start by Framing why this matters
there was a thought-provoking risk dotnet article recently about how
capital rules have overwhelmed Bank strategy the gist was that regulatory
capital requirements are having a significant influence on strategic
decision making we certainly feel this we measure the returns of our business
based on their regulatory capital usage amongst other things if a business isn’t
covering its cost of capital we have to ask ourselves whether to continue
allocating as much capital to it it would be helpful of regulatory capital
including stress capital we’re aligned with how we view our risks and if not
we’d at least like to understand the difference between our and the
regulatory view but due to limited transparency we can’t do that today to
me a better alternative would be for bank models to be used as
micro-prudential tools and supervisory models used as macro-prudential tools i
agree that banks have greater understanding of how to model their
idiosyncratic risks and supervisors might and i agree that if bank models
were used to inform micro potential bank capital requirements i would feel less
urgent about fed model transparency but also at the heart of the assertion is
that I think the scenarios are the secret sauce of stress testing while the
models are just a calculator if the Fed is concerned about counter sec ality or
bank’s ability of whether an unexpected change in interest rates or a shock to a
particular asset class they can design a scenario to test this scenario design is
the primary tool to dynamically and on a forward-looking basis test the strengths
of each Bank and the banking system as a whole the models used to calculate
stress losses on the other hand are just the formulas that estimate what will
happen to a banks exposures given the stress scenario so I could envision a
world in which micro-prudential capital requirements are determined based on the
results of supervisory scenarios as calculated using Bank models I agree we
would need a process to manage the volatility of scenarios used for this
person for this purpose and yes this would mean different models are used to
set each bank’s capital requirements but to me this is
solution to the model model culture concern that we have the models would
have to be subject to continued rigorous supervision including a collaborative
dialogue about any differences from fed results and as suggested the Fed could
continue to use their models as a macro-prudential tool to monitor
system-wide linkages concentrations and other risks I don’t think decreased
reliance on supervisory models would diminish the Federal Reserve’s ability
to influence the behavior of supervise banks use of bank models would simply
result in a stronger connection between strategic capital allocation decisions
and risks and I wouldn’t be making this recommendation if not for the tremendous
efforts of the Federal Reserve post-crisis to raise the modeling and
validation standards throughout the industry now I’ll switch gears to my
third point on coherence on which I’ll spend a little bit more time given this
wasn’t a focus of Greg and Andrews paper to reiterate regulatory capital
requirements including stress testing or influencing bank capital allocation
decisions as long as that is the case coherence is critical in the context of
today panel today’s panel I’m thinking of coherence on two levels
first the components of the stress test should be consistent with each other
second the components of the regulatory framework should add up to an aggregate
capital requirement per business that is commensurate with the risk of the
activity digging into the question of coherence within the components of the
stress test I don’t think many would disagree that the seek our global market
shock large counterparty default and the nine quarter macroeconomic scenario are
often intentionally inconsistent with each other the macro scenario is
actually in the realm of what we might expect in the severe recession with
double-digit unemployment rates 50% declines in equity prices and severe
declines in real estate and international markets
by contrast the global market shock which simulates the severe industry-wide
fire sale is calibrated at a completely different level of severity to the
macroeconomic scenario Shipman and Blackrock have attempted to quantify the
severity of the global market shock taking the corporate bond market as
example they estimate that the shock applied to a single a rated corporate
bond as a probability of one in 100,000 that’s severe but when it’s other low
profit ability market shocks that are assumed
to occur instantaneously across all asset classes in all regions that the
global market shock becomes unrecognizable relative to anything
we’ve seen historically estimates may vary but it should be obvious that
there’s a large discrepancy between the far tale of the tale calibration of the
global market shock and other components of the seeker exercise as for my second
point about capital requirements being commensurate with the level of risk
I fear that we’re not thinking about Agri capitalization rates at the
business level when capital requirements are designed and the cumulative
calibration across the stress test and the broader capital framework is leading
to systemic risk increasing market changes some of which Greg and Andrew
have highlighted continuing with the example of the corporate bond market
it’s clear that capital requirements and other requirements have driven
intermediation activities from regulated banks to the shadow banking sector
although some of this make migration may be intended research is suggesting that
shallow market depth is likely resulting in increased borrowing costs for small
corporate issuers and increased transaction costs for asset managers my
bigger concern is that we don’t know what will happen in stress but it’s hard
to believe that reduce liquidity and greater fragmentation are helpful
this is particularly concerning considering that US companies largely
financed operations through US capital markets as opposed to direct lending I
am of the opinion that the stress test would be extremely conservative with
recalibrated assumptions based on a thoughtful review of the coherence of
the scenario components in the aggregate loss calibration for business coherence
is also critical when we think about how stress testing fits into the broader
capital framework we are supportive of incorporating stress test results into
daily capital requirements to distress capital framework Greg and Andrew have
highlighted an interesting and important point though if FCB SCB quantifies the
capital we need to withstand losses and remain viable than by adding SCB on top
of the g-sib surcharge we have materially increased the g-sibs post
stress minimum but that’s not how the g-sib surcharge was codified in Basel 3
and we have not seen any analysis support why the g-sib surcharge is the
right measure for how much more post stress
in equity than four-and-a-half percent a g-sib would need to remain viable these
are complex fundamental changes that require the process I understand that
the initial reaction may be that we’re better off to err on the side of
conservatism when it comes to bank capital requirements all I know is that
when you add up all of the pieces of our capital requirements from the bottoms up
there’s double county of risk to see more reflective of rules being written
in silos than deliberate calibration that takes into account all the
post-crisis reforms and that’s leading to real market implications in the form
of increased costs and potentially increase systemic risks I don’t have
time to get into the detail but I’m even more concerned about coherence when I
look ahead a couple of years and look at things like the fundamental review of
the trading book Basel three revisions Cecil and potentially the
counter-cyclical buffer on the horizon when each component of the capital
framework is being separately modified without thinking about the overall
coherence of the capital framework it is unlikely the capitalization rates at the
business level will be appropriate with that I’ll wrap up where I started now in
peacetime design choices that were made a decade ago to compensate for lower
capital levels and less robust modeling practices are due for revisiting the use
of bank models and more coherent assumptions and calibrations to set
individual bank capital requirements will improve market outcomes without
weakening the stress testing process or safety and soundness conversations like
the ones we’re having today make me hopeful for the future of stress testing
and thank you again for letting me take part in the dialogue thanks Brian so we’ve now got half an
hour of questions and discussion and as taking my privilege as moderator I’m
gonna ask the first question so please members of the audience be
thinking of your questions because we’ll certainly have time for those in a
minute so one of the themes that I picked up on
across I think each of the presenters you talked about the difference between
peacetime and wartime or times of stress and I want to link that to the point
that Andrea and Greg’s paper made a brown the speed of adjustment to the
stress test which I thought was a nice point to make and not one that’s often
focused on and there was definitely a diversity of views across the panelists
on this topic as well as many others as you heard so I’m really interested in
everyone’s reaction or anyone who wants to speak so my question is the speed of
adjustment Andrew your your comment was that it could be slower in peacetime and
that’s a way of you know smoothing out some of the bumps around stress testing
Denis you made the point that we don’t necessarily know when we’re in peacetime
and when we’re in wartime and Charlotte you made the point that one of the
values of stress tests is that it’s forward-looking and it hopefully gives
you the capacity to take actions before the stress has really happened so my
question is really about the combination of those they there’s there needs to be
a reconciliation because the the viewpoints are so different can we
really be relaxed in peacetime or should we take the view that the stress is just
around the corner so maybe I’ll ask Andrew to start first since it’s
directly coming out of your paper and then offer anyone else who wants to
speak the option okay thanks anyone else wanna well I
think we’re due back to first principles when we think about this the premise is
that at that point in time the institutions level of capital is too low
as first principle so you’re too low in peacetime assuming you know it’s
peacetime and as I won’t repeat it but I think that the the strength of the view
of peacetime wartime we’re moving from one to the other is dubious at best and
and and you can think about let’s assume for a minute where we are we’re
objectively in peacetime we know it right now and in 20 years from now we
look back we confirms we’re in peacetime the question I have is you should be
building buffers in peacetime not thinking about you know how to make
things easier in fact counter-cyclical buffer should be put in place now when
the when you have you know great break record-breaking profits in revenue and
things are going great and record-breaking bonuses we should you
know the suggestion is we should be saying well they’re short on capital now
in peace let’s let’s think about how over a
period of time they can get the way they should be today and I and I would say
where they should be today is where they should be
they shouldn’t have got themselves deficient today they know better they
and if they don’t know better than you have qualitative issues not quantitative
issues so I think that too many issues are getting alighted by the discussion
of speed or saying it relates to the level it’s not you have to go back to
first principles and they should be there now they’re not
let’s get them there fast so to put my perspective on it and again I think if
some of this does come back to the level but if you were taking a perspective
that we are at approximately the right level and that’ll be I’m sure another
part of the whole QA but I I think the important thing here is really a
question which gets back to my secret sauce comment it’s all about the
scenario design and the volatility if it’s actually if we believe we’re in the
aggregate at roughly the right level and we’re seeing volatility coming from
scenario design then that in particular should be weighed in the cost benefit
with regard to the speed that we’re making those adjustments we can look at
numerous different scenarios that will give us different answers that doesn’t
mean any one of those scenarios is the absolute you know perfect answer we know
we’re never going to forecast whenever the next crisis will be that we have
forecasted the perfect scenario but the important thing is understanding that it
needs to be a severe scenario and how we thinking about capitalizing the overall
system which I think we certainly have have achieved which also led to some of
my comments not about changing any of the overall process but much more just
about recalibrating and rethinking after a decade and then on counter cyclical
the only thing I would highlight is you know we certainly see and feel through
scenario design again the counter cyclical nature of how the scenario has
been designed and if you even look at this last jurist scenario I could
certainly argue that it’s been the second most severe certainly continuing
to take in a counter cyclical approach yet building on
verón your point that’s I thought it was thinking about the speed of adjustment
in sort of two ways the speed with which we as regulators need to adjust the
severity of the scenario and in looking at our counter-cyclical buffer as we’ve
done research on how you know rapidly would we have sure should we have
changed accounts that click a buffer before the crisis given what the
indicators the risk indicators are telling us before but we try to reduce
the volatility of the requirement so the speed of adjustment for firms by giving
them 12 months you to move to an increased and cyclical buffer but we can
really reduce it immediately the other way I was thinking about the speed of
adjustment was if if a firm doesn’t pass the stress test then what actions it
needs to take for us with a that’s a longer scenario we can afford to look at
where in that in that stress period is the low point of the stress and so
whether the management actions will address the deficit in time or whether a
firm would actually have to act sooner and we comment on the the banks of the
banks own capital plans as to whether they will be sufficient or whether
actually something speedier would need to be done okay thanks
let me look to the audience and please just raise your hand if you want to ask
the panelists a question over there Sean yeah they’re bringing one so thanks my
name is Sean Campbell Ann I just want to thank the panelists I really enjoyed the
discussion I think I learned a lot this morning as I as I’m sure everybody else
did I just want to return to something that
was sort of stated quickly in the observations that I’d like to get maybe
I’ll provide a comment and then ask for the reactions of the panelists starting
with Professor metrics since it was his observation I think one of the
observations that was listed on the slide
as sort of a dichotomy I’d like to push back against on and I’d like to sort of
get some reaction from the panelists which is this notion
that the banks understand their risks better than the regulator but the
regulator has a better assessment of system wide wrists in the bank so that’s
how I read it and I’m sure it’s more nuanced in the paper but that’s sort of
the four point version and I’d like I’d like to push back strongly against the
second part of that bullet point so I would surely submit and it’s truly the
case of the Fed and regulators have information about exposures across the
banking system I would grant that at the outset that simply does not imply all by
itself that the regulator has a better handle on existing wide risks than other
entities in the economy in particular large banks so if you ask yourself the
question what is a large Bank and what does it do is in the risk assessment
business across a variety of sectors of the economy and when you think about
what drives the stress tests what determines whether a bank is going to
have a good or a bad year in the stress test is what’s happening to the
corporate default experience what’s happening to consumer loans what’s
happening to auto loans what’s happening to mortgage if you
think about those system-wide risks and who’s in a position to understand what’s
on the horizon respecting those risks I think it’s the banking sector and as you
pointed out in the context of your discussion what’s really important is
the increasing size of the shadow banking sector when you think about
large banks and their commercial interactions with the shadow banking
sector I also think they have a relatively strong handle on what’s going
on in that part of the sector and I’m a little bit worried about conferring this
immediate advantage to the regulator and in particular the Federal Reserve is
having some all powerful knowledge about what’s going on the rest of the system
and banks don’t have and like a I’d like your reaction to my to my reaction to
your bullet point and I’d like your perspective on how you think information
outside of the Federal Reserve may be information sitting inside the banking
sector should be used in the process of risk assessment and scenario provide just a another comment I guess
from my perspective it also just provides an opportunity to think a
little bit more about again how to leverage scenario design and maybe go
into some of Charlotte’s comments about how Bank of England does it and one of
my fear is getting back to scenario volatility is we’re almost experimenting
with a sort of a loaded gun on an annual basis versus using the macro potential
sort of focus to be able to run more scenarios look for concentrations look
for other kinds of systemic issues through running unique and separate
scenarios but not necessarily making them the annual exercise and so I think
that’s where there could be a lot more value at it to be more thoughtful and to
make sure that we don’t become stale over time just only a very quick comment
which is you began by saying the observations that banks understand their
risk better than regulators and because regulators understand the systemic or
industry-wide risks better and you disagreed with the second part I
disagree with the first I mean you know everybody here you know I believe is you
know older than 15 years old to start with the premise that banks understand
their risks should should engender healthy healthy skepticism you think
about two thousand four five six seven even eight nine ten there is we should
all be as I said before incredibly humble about drawing conclusions about
what we know and the confidence with which we know it or we attribute to
others there’s no question they have massive amounts of data massive amounts
of talent they’re in the middle of the flow of how many markets so they’re
seeing both their individual firm issues and they’re seeing systemic issues and
they’re uniquely positioned with unique knowledge in many ways and we should not
jump from there to the conclusion that they understand the risks and Dennis I
thought you were going to see I thought you’re going to refer back to your room
about incentives and whether thanks management are incentivized to perhaps
look to the downside and I think there that there could be agreement about what
what is happening in the in the economy or the markets but I think the the range
of uncertainty of expectations about what might happen a regulator will be
perhaps moving to a perhaps a more pessimistic view than the firm’s and a
firm’s management might I gave an example in my comments of where we had
re identified where we thought your firm’s
models were working well in good times but were they really set up to think
about a downside stress and where actually then the questions we were
asking and I think that’s really the important thing about that the stress
testing is that asking those questions as to what have we really thought about
how a portfolio may perform in conditions other than those in which
model is working perfectly fine Mike maybe just a follow-up almost taking the
question to another level and following up a little bit on risk management you
know I think one of the things anything it ties into this conference and the
overall you know efforts over the years is the investments that have been made
by the regulators as well as by firms particularly from a stress testing
perspective some of the conversations were having today I don’t know if I
would have taken a strong stance five years ago there has been a huge
investment that doesn’t by any means meaning that regulators or individual
firms have perfect vision but I think it is important to recognize the you know
the great effort that has done interpreting it to us much better place
then we certainly were requested okay thanks
other questions yeah there’s one over here on the aisle thank you mm-hmm I’m
Michael Greenberger from University of Maryland School of Law with regard to
the proposition though the economy structure of the
economy as well as regulators do another point I think needs to be we need to be
reminded about the London whale when one rogue trader JPMorgan Chase I believe in
2010 or 2011 lost six billion dollars at JPMorgan Chase
when it first happened Jamie Dimon said it was a tempest in a teapot the first
assessment was that Bank and lost two billion CFO was fired because she could
not get on top but the losses were in the end the losses were six billion
dollars now JPMorgan Chase had the cat in that case it had the capital to
sustain that loss but a lot of banks would have been had problems with that
kind of loss and also it was one rogue trader if there had been many a loss
could have been much greater also about banks understanding what’s going on in
the economy in terms of credit risk I’m reminded of the long-term capital
management failure in 1999 where 13 banks thought they were the only bank
that was a counterparty and a lender to long-term capital they were called to
the New York Fed 13 or 14 of them sat in a conference room and found out that
each of them were counterparties and creditors so long-term capital
management and I do not think in the run-up to to 2008 even in September the
middle of September 2008 the banks as a whole understood what was happening in
the economy the regulators are better positioned today because one of the
there’s a lot of research that a major destabilization in 2008 was
private nature of naked credit default swaps that there was not repository
dodd-frank creates a swap repository which federal regulators have access to
and they can look in that case about who is exposed to whom and I do not believe
because of privacy concerns confidentiality that banks have access
to who the counterparties are on every credit default swap that’s entered into
in the United States but more important I have a question for Dennis because
there is a major theme running through the discussion capital accumulation as I
understand it are so severe that it’s favored shadow banking and it’s driven
of these risks out of them off the radar screen and therefore we’re in worse
shape today I think Andrew and Gregg bring up a good point which we all need
to keep in mind which is risk migration and no question that pre-crisis
regulatory arbitrage resulted in massive risk migration from a more regulated
banking system to a less regulated to no regulated shadow banking system
including prominently the derivatives markets and we’re seeing that again
today I don’t think and and there was a it was
almost a suggestion in the paper it may not have been that banking regulators
should be mindful about how tough they make regulation on banks so you’re gonna
push some stuff out of the bank the regulated banking sector and we should
in the banking regulators should worry about that I would say no banking
regulators need to do what they need to do on federally insured Fed backed open
window access banks and if that causes migration of activities to less
regulated areas then you should be regulating those
areas more you should not be in any way decreasing the regulation in the banking
sector and it’s unfortunate but you know whether it’s a you know I worked the
United States Senate at the time when dodd-frank was passed and I was involved
in much of this and there was a lot of discussion and agreement by the industry
and others that we needed a regulator or a regulatory entity that oversaw we’d
got out of the silos and saw everything or had the responsibility duty to look
at everything an eff sock was created no matter now no matter how perfect the
imperfect F sock might be the whole purpose of F sock was to address the
regulatory arbitrage in the two-tier regulatory system between banking and
non banking and and hit head-on the shadow banking system now we can all
talk about whether or not it was done well or right or whether or not the Fed
should have had the responsibility for regulating that designated not as
systemically significant non banks we can argue about all that but that was
the point of it and today for all intents and purposes and I know some
people will disagree but I think it’s factual F sock has shut down the shadow
banking system is being revived regulatory arbitrage today is literally
out of control and not regulated and so there is a huge risk that Andrew and
Greg have pointed out here it should not it but under any circumstances cause
banking regulators to pull back to avoid that migration it should cause everybody
else to increase the regulation on systemically significant non banks to
stop the regulatory arbitrage okay yeah right here another question morning I’m more expulsed ein retired
from the Peterson Institute I am Andrew two points he made in his presentation
one was that his paper was independent of the question of the long run
bright capital ratio for banks the second one was that a lot of
difficulties there might be resolved if we thought about longer adjustment
periods so suppose one made the judgment that
the right long term operation let’s call it the tier 1 leverage ratio for the
eight g-sibs was in the neighborhood and say 15% rather than the 8% they have now
but let’s think about that happening over a period of a decade I mean you
could think of you know long-run adjustments in other areas like the fuel
economy standards developed over decades in which the industry at their inception
said all of this will be a disaster impossible yet over long to do that yes
you have to deal with the migration create but why not think in that way and
why not think of the stress test as an instrument that would help move toward
that a few years ago I wrote a book on stress testing in Bain Capital reform
and I originally started out of singulars this sort of two separate
things and then I came to the conclusion not too long that that was really not
possible you couldn’t really judge the stress test think what they useful if
you didn’t make a judgement about the current capital staff we have a minimum
stretched threshold or the SLR 3% to one leverage of 4% do you think that makes
sense well then when you see everybody past
you have one judgment if you think that’s willfully low and that the the
peacetime non-stressed in is willful oh you come to another so it’s an
instrument for that and if we learned over the next five years you know
running all these stress scenarios helps us to see that the world is a much
scarier place where we fought before the overall level should be higher that
would make sense that’s not any consistent with with the way is what we
were trying to pull away from a little bit here is the the general to be so which is that you know this is the way
we’re going to the level of this is the thing that currently bites stress test
sort of a thing this is the way to the level and I just
wanted to try to separate from an intellectual debate that actually if
we’re gonna talk about stress testing methodology it should be in a world
where we have somehow compartmentalize the question of the level it could be
the knowledge about that would be doing these for nine years which book would
say it reaches the issue of the benchmark in which way you’re looking
and Brian said this quote common equity more than doubled since the crisis well
that’s the wrong benchmark during the crisis the levels of everything were the
lowest since the crash of 1929 the benchmark should not be over so much
better than we were when it was a disaster in everything failed that’s not
the benchmark Martin wolf said back in 2014 when people were talking about how
much more capital they had he said look three times zero is still zero so the
question isn’t how we do we have more than we had at the worst point in the
last hundred years are things better than they were than they were at the
worst point in 100 years well gosh darn I hope so the question is where should
we be now that I know we can all disagree some people can disagree about
that but that’s the benchmark we need to get away from talking about how we’ve
got double this and double that and so much better than we were in 2008 when we
were when you were staring a second Great Depression in the face that isn’t
the benchmark we should banish the entire comparison and start thinking
about the right benchmark which is where we need to be and how we get there how
we stay there and who we’re doing it for so my only follow-up would be because
obviously some statistics can have different angles but when we look at
where we are today and again all the investments
have been made in the severity of the stress test and the overall
capitalization of the system you know that is really what at the end of it
when I go through all of my comments and ask and talk about recalibration its
recalibrating that process because we’ve implemented and done an enormous number
of great things in the last decade from a regulatory reform perspective but we
do need to take a step back and look at how they all interact together to ensure
we’re getting to that right level okay we have time for one question if it’s a
quick one and they’re bringing a microphone I can’t Hurley online lending
Institute in Boston University I’ve done a lot of talk about wartime
versus peacetime stress testing community that there’s some difference
between wartime and peacetime is the stress tests in 2009 were accompanied by
a pledge by the government that it would fill the hole if the banks could not
that was the essential difference between the European stress tests which
failed and ours which succeeded like my my question is that perhaps we are
trivializing stress tests such that in the next financial crisis can we go back
and say oh now we’re gonna do stress tests that really matter when we’ve done
the stress stress tests in peacetime for successive years and he thought comments
on that a larger issue of our peacetime planning
for water which is effectively non-existent in the from a legislative
perspective I would say that we have it really what we would need to do is what
we just described and we need to set aside
I can’t imagine such a thing would happen again really thought it through
and he’s fine what that more time stressed us we need to look like is
going to be a problem so I’m agreeing with your concern
yeah meet you okay so we’ve reached the end of the timeslot before we break up
I’d just like to remind everyone that we’re gonna come back in here at 11
o’clock to the next session and please join me in thanking the panelists you but we’re gonna just keep moving moving
forward with the the rest of the agenda our second session today focuses on
dynamism and transparency and stress testing our moderator for this panel
Beverly hurdle is executive vice president head of the research
statistics group and director of research at the Federal Reserve Bank of
New York and that role she oversees a team that supports monetary policy bank
supervision payment systems and financial markets Bev is a veteran of
the stress testing program having served as Deputy Chair of the Federal Reserve’s
model oversight group and has published numerous articles on stress testing and
other banking topics and with that I’ll turn it over to thank you and a special
thank you to Lisa and and Bill and their colleagues for putting together such a
tremendous program today the title of this panel is transparency and dynamism
of stress testing and to start us off we have a paper by Mark Flannery mark is
the Bank of America eminent scholar in finance at the University of Florida and
then following Mark’s presentation we will have a panel discussion we have
three terrific panelists in alphabetical order we have Tim Clark who’s the former
deputy director in the supervision and regulation division at the Board of
Governors I’m sure many of you know Tim then we have Randy Gwynn who’s the head
of the financial institutions group at Davis Polk and Wardell and finally Andy
chriskiss who is the chief risk officer at State Street so with those
introductions I will turn it over to mark thank you very much Bevis it’s really
nice to be here partly because I struggled with writing this paper and
now it’s done so that’s a large part of why it’s nice to be here for me
I got to go back think okay so this is about transparency and model evolution
another word for dynamism and what has gone on with the the seek our process is
a whole game changing way to compute minimum capital ratios so before the C
car process the idea was you look at your balance sheet you multiply by Basel
risk weights or your internal estimates of risk weights and you get your capital
requirement and if you’re a little bit short there there appeared from the
outside I can’t say from the inside there appeared from the outside to be a
somewhat leisurely process of converging toward the the required capital or the
required capital cushion and and they talked about that a lot in the first
session now for D fast and C car the first thing that gets an a-plus-plus is
that they’re forward-looking risk weights they’re not static risk weights
they’re forward-looking risk weights that depend on the anticipated economic
performance in the in the in the economy and in the loan portfolio however the
models are unknown which adds noise to the process of managing a bank holding
companies capital structure capital planning and on top of that and I think
I agree with with Andrew this is very important on Coppa on top of that you’re
going to rectify deficiencies immediately not immediately the week
between D fast and C car you’ve got a rectified deficiencies so there is
uncertainty about the the model your capital requirement and you’ve got to
rectify it quickly and those two things together have generated a lot of
pressure to know more about the model now the way I think about this paper in
the process is that we’re balancing transparency against dynamism that these
two are in conflict that the transparency banks would like to know
their positions their required capital positions in advance dynamism the stress
tests have to change all the time I mean we’ve had a deep depression for nine
years and okay I get it the large banks aren’t
going to suffer huge losses from deep depression so let’s move on if we think
there are other risks to be considered and one of the big things I worry about
is the APA I was at the SEC for two and a half years and I became deeply
impressed with how much of a force for stasis is the APA and is the the comment
and disclosure and comment process and so that’ll come implicitly in a lot of
what I’ve got about to say now before I start let me make a background thought
about costs before C car there was an allocation of risks and there was an
allocation of financial services across the various kinds of financial
institutions and then C car came along and raised minimum capital requirements
C car and some other things that raised the cost for C car bank’s higher higher
cost of funds relative to smaller banks and relative to non banks and so it’s
inevitable that we’re going to get risks and services redeployed across the
relevant institutions and so one of the things that that I want to do and it’ll
be important when I when I talk about the cost of uncertainty for the bank
holding companies is to distinguish between the private and the social costs
of regulatory changes so if the C car banks are making less C&I loans
that’s not socially costly and less than until we establish that there’s no
supply of alternative C and I loan services coming from elsewhere in the
economy and sometimes when when I read Basel documents I read FSB documents and
and I read lobbying documents sometimes people seem to not to distinguish
between those two but I think that’s an important background for what I’ll talk
about in terms of the costs of transparency or the cost of opacity so
I’m going to talk about three kinds of transparency today and then I’m going to
make some suggestions about how to incorporate new sorts of risks into
model simulations not because I have any risks in mind but as I thought about
alternatives to the deep recession stress
it became clear to me that this was going to be require a qualitative change
in the way we specify the risks so what are the bank holding companies like
about model transparency it improves their ability to predict minimum capital
that means that the total required capital required capital cushion they’ve
got a hold is bigger on account of the uncertainty that the feds DFAS model
overlays it but it’s important here to point out that what the the uncertainty
is about is the total amount of capital so if I can predict as a bank holding
company the total amount of capital I don’t really care what the feds opinion
of this loan versus that loan versus another learn what the risk weights are
on the various components so the the need to hold extra capital is driven
largely but not exclusively by the uncertainty about how much capital in
total is going to be required now how large is that cost if you look
at the stress test results recently it turns out that the bank holding
companies are pretty good at squeaking by just above the minimum requirements
in the most recent year that just ended two out of the 18 tested bank holding
companies had to go back for a second bite of the Apple had to cure their
deficits one was six tenths of one percent of risk weighted assets the
other was one tenth of one percent so they got the second bite of the Apple
later on in the year there is an opportunity I don’t know very much about
the the mechanics of this and how real this opportunity is but later on there’s
an opportunity if you wind up holding too much capital you go to the Fed and
you say well things have changed I’d like to to disperse some more of my
capital may I do it so there’s apparently an informal process for doing
that and and as I say the the banks are doing a pretty good job of hitting their
minimums so it looks to me as if the costs overall that come from this
capital planning uncertainty are pretty minimal now some of you are saying well
that’s easy for you to say it’s not your capital and I would be perfectly happy
to be convinced otherwise but but my my view of the situation is that those
costs are not very big what about the regulator’s well the
regulator’s want to have accountability they want to have a certain amount of
credibility with the public and with the banks the banks are quick to point out
that all of the models and all of the scenarios would be better if the Fed
would only accept some intellectual input from the regulated firms but my
guess is the Fed gets a lot of intellectual input from the regulated
firms so you know the the notion of making the models transparent or the
stress scenarios transparent because we want more intellectual consistency
strikes me as as probably a second-order effect but the important stories here
there are two of them one is the Ohio story so if you don’t know they’ll
failure story let me tell it to you there was an act in 1992 congressional
act whose name is very long it created ou FeO the office of Federal Housing
Enterprise oversight Fannie and Freddie and the stress test was specified in the
statute and the stress test was further it was further specified in the statute
that everything about the stress tests would be available in the Federal
Register now Scott frame and a couple of co-authors wrote a paper looking at the
the stress test that started in 2002 and ended of course with the conservatorship
in 2008 and they make a couple of interesting and observations one the
data that went into the models for the stress test that were used was estimated
or their models were estimated over 1979 to 1997 no changes there of course
between 1997 and 2002 but it was estimated over that model and it never
changed afterwards and and frame and all placed the blame for never changing on
the burdensomeness of complying with the APA sorts of comment and disclosure
policies the other part of the so the first part of the Ohio stress test was
lots of defaults the other part was a 600 basis point parle L shift in the
term structure that’s roughly what it was
Dwight Chaffee has a paper where he he looks at the the term structure chef
and he said well what the agencies what the GSE seemed to have done is immunized
themselves against the parallel shift in the term structure and left all of the
other potential shifts unhinged so they were there was lots of room to
take risk that the model wasn’t going to reveal so the bottom line is that that
when we specified or over specified the transparency of this o fail model we got
bad models that didn’t change and gave us bad results
so you could look at that model and you wouldn’t really have learned what you
need to learn about the capital adequacy of the GSEs now the other point to make
is the model monoculture and and first thing a banker says about the mod amount
monoculture is my model is better than theirs I’m not going to use theirs I’m
not going to change my portfolio however simple statistic says if you’re trying
to predict something and you have a model and you have another model the
optimal prediction is a combination of the two models in inverse relation to
how what standard error is the forecasts are so it’s perfectly rational that to
some extent all of the banks and the stress test if they had a transparent
view they would use the the fed’s model to make the predictions and there’d be a
little bit of the DFAS and everybody’s credit allocations and that’s the sense
in which i think of the monoculture that it’s going to reduce the diversification
that we have across models in the private sector and of course all models
are wrong so when we add in a new model we might get a better estimate but we’re
not sure that we’re getting to the right place now in March of 2019 there was a
dramatic change in the information released it was the first quantitative
information provided in the form of credit card and CNI loan portfolios
sample portfolios and and loss rates now I looked at that and I said first of all
it’s the first quantitative information the Fed had been pretty cagey before
that about giving qualitative information it seems
to me to be supplying hints about the parameters without actually telling us
what the parameters are and as I thought about that and I thought about the way
it’s likely to go the two words that occurred to me were slippery slope if
you say to me do I think this has gone too far
my answer is maybe yes if you ask me to do what I like to go further in the same
direction my answer is probably not because I’m very concerned about the
parameters getting out for both of the FAO and the mono model amount of
cultural things so that’s transparency about the models
what about the results here I think the story is really positive I really like
the amount of detail that’s given out I wouldn’t mind more detail the D fast and
the seek are announcements continue to move stock prices significantly as
recently as 2018 I don’t have the 219 results and they move stock prices not
only for the tested banks but the non tested banks so they’re telling us
something about something fundamental of the banking industry but I think the
really important thing about being transparent about the results is that
the day is going to come when there’s bad news to give so far we’ve had good
news everybody passed or came close but someday there’s going to be a large firm
that doesn’t pass or doesn’t come close and because of the the tradition of
being open about this the feds not going to be able to hide that and so following
the example the s cap that one of the speakers this morning and the
questioners referred to following the tradition to the s cap the Fed is gonna
have to come to the market with a problem and a solution or they’re gonna
have more pressure to make the solution happen before the C car tests and they
have to go to the market so I view the the transparency here is very important
in part because it puts some pressure on the supervisors to act quickly and not
to forbear for so long which we have all sorts of evidence is not helpful so so I
view that as sort of the the second edge to the sword of of transparency the
scenarios I’m actually not going to say very much about the scenarios I don’t
care if the banks know them after the after the as update I don’t think
they’re worth much you know it’s a deep recession two new kinds of stress shocks
now we’re in the dynamism world two new kinds of shocks that I’ve been able to
figure out have to do with first of all that it is not a deep recession the deep
recession is getting scaled scale so if I’m gonna have a broader set of shocks
to consider and I’m not saying we need any particular one but if I’m gonna have
a broader set of shocks or stresses to continue I’m gonna have to change the
parameters and models I’ll explain why in a minute and I also think we can make
better use of the massive trading data that’s collected so why are we gonna
change parameters well the PPN are pre provisioned net revenue is a rich mine
too to try to explore there are twenty-four sub components this is all
public information of course 24 sub components 8 loan interest income ratios
to total assets or risk weighted assets 9 deposit expenses six components of
non-interest income and three components of non interest expense including op
risk so when I think about that I think about the richness of that specification
one of the first things I realized was that the importance of a stress shock to
losses in the loan portfolio depends very importantly on how fast you
increase your loan pricing as you enter into the stress scenario so if I if I
have we have two banks and I start to price up quicker than you do then I’m
gonna have more capital coming in in the form of provision that revenue and I’m
gonna have a lower marginal contribution to capital from that’s required from
other places same thing with deposit rates same thing
with changes in business models so all of these things are going to require
changes to the parameters of the model not to the external variables like the
interest rate and and the real sector shocks
and that gets me right back to the APA if I’ve got to go through a proposal and
comment period for all of these it’s going to be very awkward and very
unwieldy loan losses have the same thing we talk a lot about people frequently
say we ought to stress the banks to see if they’re resistant to a kind of risk
that hasn’t shown up yet so today the poster child for that would
be leveraged C&I loans the trouble of course is that you can’t go to the data
for information about those losses because the reason the sector is so big
it’s because there haven’t been big problems yet so what you’re going to
have to do to assess the sensitivity of of particulars to this particular risk
is you’re gonna have to shock the loss equation you have to move the intercept
or change one of the speeds of adjustment or something like that and
again I don’t want the APA involved the other way I think we could improve the
stress test is to take the trading information the massive amount of
trading information tens of thousands of Greek values Greeks provided by the the
large banks to the Fed the Fed crunches the numbers and calculates total losses
for each individual bank that part seems to ignore completely the macro
prudential implications of the banks having exposures so I would really like
to know what asset change has the biggest cumulative effect across all the
banks okay is there some kind of exposure that the banks are all along
are all short because among other things if the banks are long a risk the rest of
the economy is short the risk so I think I think doing more with that would be
very useful I know it’s a lot of data it doesn’t have to be done on the same
timeframe as the rest of the the C car but I think that’s very much worth
looking into then here’s my one suggestion for for making the the test
better and some of you who know me won’t be surprised to hear this the defense
takes an initial loss-absorbing capacity measured by capital and tracks changes
in appear time if the initial level is off all of their levels are wrong and we
know examples where the market value of
equity has been way different from book value of equity and so I’d like to
suggest that the process start out with sort of a mini 8qr
asset quality review but instead of looking at the individual assets I’ve
got a formula here that says why don’t we just do something for example like
the starting capital is the minimum of the book capital or a weighted average
of book and market so if market is really low then it drags down the
starting capital but I do think we got to get some sort of market information
into that initial condition so that gave me the highlight I sign summary I think
sharing the model the parameters is gonna help the banks but I don’t see the
cost that the banks are experiencing as being that large and the sharing of the
model is gonna threaten the informativeness of the tests and perhaps
invite sector-wide exposure to common risks I really like the commitment to
continue full revelation that’s going to be important particularly when there’s
bad news the stress is so far have gotten pretty repetitive after nine
times how much you need to know about a depression or a deep recession and new
stresses are going to require changing model parameters and that’s a
qualitative difference not to an economist perhaps but it may be legally
a qualitative difference and that would be I hope a bad thing because I think
there’s a trade-off there and then the final thing I’ll say and I left a blank
space in between to make it clear to you I would really like to see some market
assessments in the stress tests and I don’t know exactly how to do it I’ve
talked to people for a long time but I think it’s something that’s worth
keeping Tom fine thank you very much thank thank you mark our first
discussion is Tim Clark good morning hi everybody good to see you many thanks to
the Federal Reserve for putting this on it’s a bit of a homecoming for me and
there’s a million people we have dozens of people here that I worked with three
years and I just want to make sure they say hello and thank you to all of you
continue to appreciate all the incredible hard work you do and I also
want to reiterate Dennis Kelleher’s point about the openness to hearing
different perspectives and I hope that that will continue to translate into
considerations of the policy objectives and where we go from here because I
think you know I’ve read all the papers I think they’re excellent papers and I
think if you my perspective you if you read all of those papers and you kind of
roll it all together when I think about some of the proposals that are out there
right now both the formal proposals on those that have been discussed as maybe
pending they feel both to me I almost as unwise as they are unnecessary so I
guess you know where I’m coming from right from the start the you know I’ll
get I’ll get into more specifics as we go I like to start also with the
obviously one other things very good to come after the earlier panel because
Charlotte made some great points that I now can scratch on my talking points
Dennis made a number of points that I have already scratched from my talking
points so hopefully I can get through this pretty quickly but I do want to go
back to the objective of this which as we all know and as Charlotte pointed out
is you know to do all that can be done to reduce the likelihood that the banks
contribute to cause or exacerbate a downturn and turn it into something
approaching the kind of horrible crisis we experienced and that’s of course you
know often expressed is continuing to lend through throughout a downturn if
that’s critical but we should not forget that actually problems can come out of
the banking system that can impact and then cause a downturn and so constantly
probing and thinking about the issues that the banks is critical in addition
to a very important aspect of just making sure they can they can see
by the severe stress and continue to function why is this so important well
Dennis I think made exactly the right point about the incredible cost of the
crisis I will not repeat them other than to say I agree with what he said they
were immense they were life-changing in many many
ways for many many people and not for the better
but there’s a couple of other things that I think are important since the
crisis the toolkit that authorities have available to them has been weakened to
some extent there are actions that were taken by the Federal Reserve the FDIC in
the US Treasury that would likely not be able to be taken again or we very
difficult to get them and and I think that that’s important and I’m gonna
circle back to that when I talk about scenario severity and the difficulty of
capturing what things would have really been like if the government hadn’t
stepped in but I’ll come to that in a minute so I guess you know I should let
me acknowledge that there have been advances in resolution and recovery
preparedness certainly Ola and a lot of work done by the agencies and the firm’s
my view is that still completely a work in progress the uncertainty about the
impact of actually pulling the trigger on a resolution even in good times for a
single bank would likely make it I think a hard decision to make and in bad times
for a number a number of banks may be under pressure I’d really think it will
just have to be deferred so having said all that reducing the probability of
default of the largest banks should remain the core goal in support of the
objective for where I started the good news is from what we’re hearing we’ve
heard from a lot of people and 30 today and thank you Brian for your kind words
good to see you the stress is pretty much the program that the totality
program has been pretty broadly recognized as being useful to the end of
reducing the probability of default of the firm’s and in a couple of ways
making the firm’s think long and hard about their internal processes and be
better at managing and their risks and understanding that the issues that they
face as well as having the post rest capital requirements so that’s that’s
the good news and I guess what we’re here for is what
kind of changes can the stress testing program
stand and continue to be effective you know the rationale that for the proposed
changes that I most often have seen and I acknowledge that I’m sure there are
other things I haven’t seen is usually couched in terms of making it more
efficient now efficiency is a good thing of course
we all would like to see things be more efficient but efficiency in attempting
to achieve efficiency by taking actions that may actually undermine the
effectiveness of what has been a very effective tool to date I think are of
concern and I and I would I would actually I think effectiveness should be
the goal not efficiencies efficiency is second order though not unimportant so
the proposed changes that have been I mentioned that have been made public
formally or are being considered soon to fall basically in three camps reducing
the volatility and uncertainty around the stress test results that’s basically
what largely this panel is about lowering the capital requirements for
the banks two things that’s bring to mind there’s discussions of eliminating
the pre-funding of dividends which I think could be a huge mistake and I’ll
get to that later and removing the post-race leverage ratio as a
constraining element of this just assessing program which I actually think
would also not be a good idea and then their third element the third piece
which is already done which is has been to reduce the pressure on the banks to
maintain and continue to improve their risk management internal controls and
governance and support of their capital planning I hope that through the
supervision the supervisors are great at their jobs the Fed does have tools in
the form of public enforcement actions and other things
ratings etc well I really not been about this if anyone’s interested they could
go back and read it but the basic point is I think they eat the Fed has weakened
its hand and the ability to hold the firms accountable for not making this
practice is stronger and one of the papers for this conference actually has
a little section in it that talks about the qualitative and seems to have come
to the result from talking to a number of people that in fact as predicted
predictably the the banks will probably reduce the emphasis they put on this and
stop trying as hard as they were before and I think that’s very unfortunately
let’s hope it’s not true okay so so now so those are basically they were the way
the proposals lay out the review of post-crisis innovations I think is
incredibly important I’m really glad that it’s continued it started in 2015
we met with a number of banks public interest groups academics etc to review
this just to see program in 2015 got a lot of very interesting and
valuable input from those groups which continues to float around you know for
me one of the things back then that I was struck by and perhaps others were
but I won’t i won’t even others might be was actually that the input we got from
the banks hadn’t really changed from the same things they were telling us we’re
bad about the program when we first put the program in place which is to say
that they they didn’t like the concept of it I didn’t like the idea of it and
once they got over the shock that the supervisors and their bosses the Board
of Governors were actually going to take this seriously they they were had a few
particulars things they were concerned about and and those are the things that
were talking about today so so so quickly on that I just you know I’m also
struck now that those same issues continue to be as such a big issue and
such a big topic of discussion largely about being surprised and volatility of
capital needs and etc and and so that that I just feel like we haven’t evolved
evolved as far as maybe we could and that’s in that sense and and I and I’d
like to applaud mark in his paper and many things that she talks about at the
end there ways in which we could we could make the test more dynamic rather
than less I guess I would argue again with the the banks I think could be
described as having benefited from this program
I think the banks are actually in pretty good shape and and point of the program
was not to benefit the banks but but I I’m surprised that they continue to push
back on the same issues to finish my thought the wind in fact we’ve now had
nine years of it and it seems to be going pretty well is pretty effective
you’re looking at me so I have how many one one minute to go goes really this
goes really fast so the key question can’t effective must be maintained or
enhanced by making the test less dynamic and more predictable
I think the Bark’s paper makes a great case that there are many reasons to be
skeptical I worry about what’s commonly referred to or maybe somewhat lazily
referred to as gaming the test changing positions moving positions the the
herd-like behavior that may result if all the firms try to have just managed
to the fed stress test okay I’ve got about I’m only about halfway done so I
will actually skip straight to my conclusions well I have to say one other
thing sorry but I’ll try to do this really quickly on the scenario design
topic so Brian you brought up the issue helpfully there lots of scenarios are
good the Bank of England has this I think a really good program and in the
scenario that they use on the off years to test things out personally I could
just question why you don’t then make the bank’s capitalize against it but
that’s just me and that’s probably not going to surprise you that I said that
the you know in a perfect world I think there would be to keep it dynamic there
would be a lots of shifting scenarios and there’d be a lot of type of
different scenarios changing from year to year but it’s probably not really
practical and I know that Lisa staring at me thinking you know please don’t
recommend that we do lots of different scenarios and so I want to point out
real quickly because this Fed makes is clear but it’s not often discussed the
Fed isn’t trying to predict the the future with their scenario it’s not
trying to predict the crisis they’re trying to calibrate capital needs so the
banks will be able to withstand us a range of crises I think that’s the right
approach given some of the limitations I still think they could continue to use
the salient risk feature that I Frank rule to probe for new
things and I hope that they will continue to do so but one thing about
the severity that I want to come back to I mentioned at the beginning and I guess
I will end on this the there’s this often critique that the Fed scenario was
too is more severe than the global financial crisis and as I mentioned
earlier the global financial crisis you know there was tremendous input from the
authority some of which can’t be done again but even if it could be done again
the banks should be required to capitalize should be able to internalize
we could internalize their externalities as the economists would say and so I
would hope that the Fed could think about severity scenario a scenario
severity in a way I don’t know how you do this that would try to strip out the
benefits that the banks gained from all of the taxpayer-funded government
government actions that were takes in the crisis so I’d actually don’t think
that it’s right that this scenario is anywhere too severe and I think we could
probe boys to make it more so so focus on increasing effectiveness don’t forget
the importance of the objective and last but not least again to Mark’s point on
on transparency I think the forbearance point is an excellent one the more that
that can be done in public of the better of self disciplining mechanism it is
personally I when we talk about transparency especially about giving
information to the banks I’d like to see more information being given to the
public I think the public deserves to have a better understanding of both the
feds views on and actions towards strengthening the banking system as well
as on the banks themselves and I guess with that I will leave it and clearly
I’ve left out about 40% of what I was going to say so hopefully we’ll get some
good questions thank you thank you Tim brandy so you’re now going
to hear the other view about the value of transparency my thesis is that full
model scenario and results transparency is essential to the public
accountability political legitimacy and even the into continued independence of
the Federal Reserve transparency is essential to the proper functioning of
any democratic system of government the public can’t hold their elected
officials accountable without full information about what they’re doing and
how they’re doing it that’s why the lawmaking process in Congress is open to
the public and why laws enacted by Congress are published that’s also why
the Administrative Procedure Act requires agencies to subject their
rulemaking to public notice and comment and why regulations are published in the
Federal Register there are of course legitimate expect exceptions to the
strong public interest in transparency but those exceptions are generally very
narrow for example Congress has created exceptions from the general presumption
in favor of transparency for matters that would compromise national security
if the public believes that Congress or the president are hiding too much
information from public scrutiny however the public can replace those
representatives in the next election with individuals who are more committed
to transparency the public interest in transparency is even higher with respect
to the actions of unelected officials such as the Federal Reserve principals
and staff who are not directly subject to the public election process secrecy
may be justified at the Federal Reserve but in general secrecy is only justified
if it’s necessary to prevent a serious identifiable and immediate public harm
such as a financial panic or to encourage regulated firms to voluntarily
share proprietary information with the Federal Reserve and those just efficient
justifications are only persuasive if there’s not a more narrowly tailored
means other than secrecy to prevent the public or firm specific harm while the
US banking regulators have long treated certain information developed in the
supervised we process as exempt from the ordinary
transparency requirements my law partner Meg tyre recently argued persuasively in
my view and congressional testimony that the modern explosion in what has been
treated as the secrets of the temple or confidential supervisory information has
made the historic balance btran between transparency and count and secrecy
untenable most of the secrecy is no longer justified by any legitimate need
to protect the public against any serious identifiable or immediate harm
instead its main effect seems to be to insulate the Supervisory process from
public accountability the lack of transparency undermines the public’s
confidence in the feds supervisory process and if left unchecked will
eventually undermine the political legitimacy and independence of the
Federal Reserve as applied to the stress testing process these principles mean
that the Federal Reserve should provide full transparency into all aspects of
its stress-testing operations unless secrecy is needed to prevent a specific
serious and immediate public harm such as a financial panic or to encourage
firms to voluntarily share information to its credit the Federal Reserve has
recently provides substantially more transparency about its supervisory
models scenario design and stress testing results but the Fed has not
provided full transparency out of fear that that might undermine the
effectiveness of its stress tests in their papers professors Feldberg metric
and Flannery all seem to be against any further transparency and questioned the
wisdom of some of the transparency already provided Feldberg and metric
believe that further inputs inputs transparency would enable banks to game
the system an argument that we’ve heard over and over again at least I have but
in my view Gaming is a loaded word it’s too vague to justify an exception to the
strong presumption in favor of transparency much of what is labeled as
gaming is indistinguishable from compliance for example when the posted
speed limit is 75 and you observe cars driving at seventy four point nine or
even seventy nine point nine miles per hour is that gaming or compliance you
say tomato I say tomahto the feds narrower justification for
secrecy based on a specific form of gaming is more persuasive but it’s not
persuasive enough to overcome the strong presumption in favor of transparency
that’s because there’s a more narrowly tailored means that can effectively
deter banks from engaging in such manipulative behavior the Fed can issue
a regulation against specific forms of gaming and impose m-ras enforcement
actions or even fines on a bank of a detect such gaming like professors
Feldberg and metric professor Flannery is generally against any further
transparency and even describes the feds recent disclosures about loss rates as a
slippery slope toward the disclosure of equations and parameters this first
argument is that the benefits of further model transparency would not be
significant to the banks because they’re already pretty good at predicting
required capital but that description of the public benefits of further model
transparency is far too narrow rather than focusing on the benefits to the
banks the focus should be on the benefits to the broader public including
the bank’s academic experts members of Congress and the Federal Reserve
principals and staff themselves such a broader view would show that the
benefits include first making the feds stress testing process more accountable
to the public thereby preserving and promoting the feds political legitimacy
and independence giving the public including academic experts sufficient
information to perform an effective evaluation of the strengths and
weaknesses of the feds supervisory models giving the Fed the benefit of a
more informed credible challenge from the public which should help it identify
and correct weaknesses in both its Supervisory models and the firm’s
company models increasing public confidence in the stress test and
helping the bank’s better understand the capital implications of their business
decisions before they make them professor planning Flannery’s second
argument is that the greater greater model transparency would substantially
reduce the regulatory value of stress tests and impede their dynamism he also
implicitly argues that subjecting the Fed supervisory scenarios to public
notice and comment would produce the same
adverse effects as evidence for these assertions he cites or fails ineffective
stress tests of Fannie and Freddie on the eve of the 2008 financial crisis
according to one study those tests were ineffective because a fail was required
to provide or in part because they’re required to provide full model and
scenario transparency through a formal public notice and comment process that
process was slow so slow and costly that it deterred oh fail from updating its
Supervisory models or data in a timely manner thereby reducing the dynamism and
regulatory value of its stress tests but this example is insufficient to overcome
the strong presumption in favor of transparency the Ohio example doesn’t
provide a persuasive reason for assuming that the Fed would also fail to update
its models and data prop promptly if it provided full model or scenario
transparency let’s be honest o fail was widely considered to be a weak regulator
and was therefore replaced by the FHFA on the eve of the financial crisis the
most logical implication or fifth most logical explanation for o feyo’s failure
to update a supervisory models or data was its weakness as a regulator not the
model or scenario transparency requirements to which it was subject
similarly there’s no reason to assume the Fed couldn’t limit the public notice
and comment period or its supervisor scenarios to 30 days and promptly
proceed with its scenarios as promptly as as proposed or quickly adjust them
moreover the Fed is likely to benefit from comments that his proposed
scenarios will have a disproportionate impact on some set of covered banks the
third argument is that increased model transparency would result in a mono
model mono model model monoculture that would lead to an increased correlation
of assets but again this reason is insufficient to overcome the strong
presumption in favor of transparency first since the fed supervisory model
already determines each bank’s minimum capital requirements are already ISM
model monoculture but contrary to conventional wisdom that bias is a
one-way bias in favor in in favor of low-risk assets rather than high
risk assets as long as the CRO zuv the bank genuinely believed that their
proprietary models are more accurate than the feds supervisory models for
example suppose that the Fed supervisory model says a particular category of
assets is more risky than the ban then a banks proprietary model and therefore
results in a higher capital charge than the bank would otherwise allocate to
those assets based on its proprietary model the bank will respond to the
higher capital charge by reducing its exposure to the category of assets now
suppose that the Fed supervisory model says that a particular category of
assets is less risky than a banks proprietary model and therefore results
in the lower capital charge and would otherwise be allocated well the bank
respond to the lower capital charge by increasing its exposure this category of
assets I don’t think so as long as it’s CRO genuinely believes
that its own model is more accurate than the Fed supervisory model and I will
skip the for any interest of time I’ll skip sort of addressing the final
argument about the as of date because I don’t actually think that is you know
disclosure of the global market shot before the as of date would provide any
benefit and I’m not sure that’s realistic given that the purpose of the
global market shock is to actually be you know to see how resilient the bank
is in response to a surprising shot thanks thank you and finally Andy well I
guess I’m batting cleanup today but hopefully not unleashes Mets so I’d like
to thank the Federal Reserve for organizing the conference inviting me to
speak in this panel as a risk practitioner I’m a long-standing
supporter of C car and regarded as the most powerful policy innovation to come
out of the crisis but in my experience lack of
transparency has been C car’s Achilles heel and efforts to strengthen seek are
very much in sync with vice-chair quarrels as proposals for C cars the
next chapter should focus on improving transparency as I’ll explain keeping the
bank’s guessing about C card scenarios and models is not a public virtue it
leads to insert or what I’ve referred to in other
context as the indeterminacy of capital where banks don’t know the regulatory
costs of risk I say this in full knowledge that others see more costs
than benefits to transparency and would prefer keeping banks in the public
somewhat in the dark about the design of C car scenarios and importantly the
models the Fed uses to translate scenarios into quantitative outputs for
assessing banks in particular I see a different weighting of costs and
benefits than mark Flannery does in his paper and I’m grateful to have more
contributed to this debate not only because he’s an economist for whom I
have deep respect but because his paper has helped sharpen my arguments about
transparency in making the case for transparency I’ll try to respond to some
of the excellent points mark and others have made in mark his paper and others
in the contributions this morning just to set the record straight like others I
fully endorsed the view that stress testing was instrumental in
recapitalizing US banks and restoring confidence in the financial system post
crisis I’m also a strong supporter of C car as a capital assessment tool the
question is not whether C car is a good idea but how we can make it better
and despite C cars many advantages its major shortcoming has been
unpredictability associated with the lack of transparency as vice-chair
Quarles has noted there are two sources of unpredictability first is the
calibration of the C car scenarios which determines the nature and severity of
the C car stresses and second are the models used by the Fed in their
quantitative test which for a given set of stresses determine the actual losses
for an individual Bank I’m gonna start by addressing the lack of transparency
in the Fed models first because this is where I differ most fundamentally from
mark and several others who spoke this morning through much of C car the models
the Fed has used to determine Bank losses have been in effect black boxes
there’s been limited disclosure of the general methodology behind some of the
Fed models and various white papers but not enough disclosure for banks to
know with any confidence what the output of the Fed models will be for a given C
car scenario instead banks rely on ranges of estimates in a fair degree of
guesswork to predict the outcome of the Fed’s quantitative test in fact there’s
something of a cottage industry within banks of trying to build models to
predict the feds model but the reality is that these predictive models often
dump don’t come very close this is particularly true for the non credit
parts of C car such as PP n r o CI balance sheet growth and the trading and
counterparty losses associated with the global market shock the limitations of
prediction can be seen in instances of Mulligan use and conditional non
objections for the largest banks the eight US g-sibs by my count going back
to 2012 there have been 16 instances of the g-sibs falling short in these
categories and coming up short than in their in their seeker estimates that’s
25 percent of the time which to me is an impressive miss rate
the result is that lack of disclosure of fed models creates uncertainty without
transparency banks don’t know the regulatory cost of risk and since risk
is a key factor of production not knowing the regulatory cost of risk
undermines balance sheet optimization business planning and sound capital
management it’s a bit like asking companies to make investment decisions
without knowing the effective tax rate because the IRS won’t disclosed the
model it uses for calculating depreciation the cost of uncertainty
isn’t limited to the banks that have had to use the Mulligan to scale back their
capital requests it’s embedded in the capital cushion that all banks maintain
to clear seeker minimums and a kick and it cascades down from there to the
inability of banks to know the cost of risk for individual transactions hedges
asset classes investments and business strategies and while this may be the
bank’s problem under biu conditions uncertainty could become a systemic
problem in a crisis or an Andrew metric or
to his in times of war suppose a bank needs to de-risk its balance sheet
because the economy is turned and it needs to conserve capital without model
transparency the bank won’t know which assets to sell to increase its capital
buffer for a given level of stress the bank will have to wait for its
de-risking actions to be run through the feds calculator to know what its
regulatory capital position will be importantly the Fed won’t be able to
anticipate how banks will respond to the need to de-risk either because the banks
won’t know how their actions will be treated under C car now there are four
main claims against transparency with Fed models which I’ll address in turn
claim one maintains that keeping the banks guessing is a good thing because
it will force them to practice defensive balance sheet management and run with
higher capital buffers to me if we want banks to hold higher capital buffers we
should raise them explicitly clarity is a virtue
I’d rather operate in a system with a known higher set of capital charges than
an unpredictable set of capital rules that gets thereby steals claim 2 is
about motto monal coach I can’t say it either models monoculture the notion
that publishing the feds models would lead firms to abandon their internal
models and adopt the fed’s model instead this claim ignores the requirement that
banks use their internal models for their secret submissions and public
disclosures of their results banks see car models have built up over a
nine-year period now and they’ve been subjected to rigorous internal model
validation standards and intense supervisory review the notion that banks
are going to give up these internal models and just mimic the Fed overlooks
these supervisory requirements and of course that overlooks the bank’s own
self-interest banks do have an interest in adopting the best models for their
own risk management purposes take for example PPN are no regulatory model
could do a better job of forecasting revenues and expenses than a banks own
models developed for it from for its budget purposes it would be nonsensical
for a firm to adopt the feds published PPN our model for its earnings forecast
than models that have been developed internally for strategic planning and I
think to a certain extent the same can be said for interest rate credit risk
and market risk models all of which are used for other purposes outside of seek
our claim 3 is that withholding details of the Fed models prevents banks from
managing to the model or as we’ve heard before gaming the system I’m not
entirely sure what we mean by gaming here if the Fed thinks mortgage loss
rates are going to spike in the next downturn shouldn’t we want banks to take
the Fed models into account in their assessment of risk worried seems to be
that banks can exploit discontinuities in the feds models to lower their
capital and since no framework a model is perfect every set of rules or models
creates boundary conditions but even where the Fed models may have
discontinuities it doesn’t follow that it’s in the bank’s interest to exploit
them to take on unwanted risk claim for is that not disclosing the feds models
helps in short dynamism of asik our process dynamism is a good thing in C
car but can it but it can be achieved by the Fed varying the scenario inputs
within defined boundaries for secret calibration rather than an adjusting the
models that translate the scenarios into quantitative outputs and then not
telling the bank’s to the public about it there’s no reason why a justments
can’t be made to Fed models as they are too firm models every year and disclose
to the public against these four claims the lack of transparency and seek
remodels is an exception the general preference for known rules throughout
the regulatory system if uncertainty is good for C car why not extend it to
other parts of the regulatory framework would we be better off if the Fed
applied its own liquidity models to calculate the LCR and didn’t tell banks
what deposit runoff rates it was using or how it was measuring credit exposure
under the SCC L or if the IRS didn’t publish the effective tax rate these
rhetorical questions suggest a high burden for C card special treatment
let me turn next briefly to the secret scenarios themselves on scenario design
and calibration I agree with vice-chair quarrel
that we should preserve the dynamism and salience of the stress tests while
avoiding unnecessary volatility and seek our surprises we need to strike the
balance between transparency and disclosure in a way that gives the Fed
room to flex scenarios from year to year in my view we should aim for full
transparency of the underlying philosophy of scenario design and
calibration that sets the perimeter for C car severity this would define the box
for calibration and then each year scenarios could be constructed within
the box to test different vulnerability vulnerabilities but firms would know
what the out of bounds are and this can inform their capital planning processes
putting the scenarios out for public comment each year presumably after the C
car as of date as vice-chair Quarles has suggested will help inform the feds
thinking on scenario construction within the buck within the box and avoid
unintuitive outcomes let me wrap up in closing secrecy and unpredictability
should not should not be the principles on which we base our regulatory
framework as a risk practitioner I see enough uncertainty in the world as it is
adding an artificial source of regulatory uncertainty is not the way to
make the financial system more resilient when it comes to C car as Vice Chair
Quarles has advocated we can do better thank you mark I just want to give you
the opportunity briefly before we turn it to the QA portion to respond to any
of the comments specifically about your paper that you’d like to respond to and
then we’ll have some discussion okay thank you I did not use the word gaming
because I agree that I don’t know what that means I I think the the one comment
that I’d like to respond to that is not directly about the paper but um
Randi was talking about accountability and as I read at one point the
dodd-frank act I read it as Congress saying repeatedly to expert supervisors
we don’t trust you to exercise your expert judgment and when there’s an
expertise like that involved I agree that people need to be accountable go
testify Frank before Congress but but I think that it’s very hard for me to
reconcile the concept of accountability with the notion that there is expertise
that is specialized and hard for a large number of other people to understand so
so I that makes me raise questions and as I say there was a lot of stuff in
dodd-frank that that said that that said we’re not gonna we’re not gonna let you
guys in the agencies do what you think is best we want you to do it our way and
I don’t think that always worked out very well so let me take the moderators
prerogative and ask a question to the panel and then we’ll open up for Q&A
from the audience in March the Fed released an 80-page document describing
the Supervisory stress test models and providing projected loss rates on
hypothetical corporate loan and credit card portfolios several of our speakers
referenced the document I think mark I went into the spectrum call it’s a
slippery slope Randy if the other end may be said it
was not nearly enough so turning to to each of you if you think about the
different constituencies the different audiences for that document the banks
participating in the stress test analysts who want to understand the
stress tests the general public anybody who wants to think about financial
stability questions in the US academics how well did that document do at
providing information that’s helpful where could it do more where did it do
too much I’m not sure if you’re pointing at me bed but I’m happy to start off ok
you were ok great so I’ll start off and I I’m gonna guess that you can guess
where I am on the Marxist disclosures I think the March disclosures were a
step in the right direction but ultimately not sufficient and why were
they not sufficient because the acid test of disclosure to me is whether the
banks can use the disclosure to build models that are gonna be able to predict
the feds outcomes within a very narrow range that’s called plus or minus a few
percentage points because I don’t think that there were to be a mystery in the
process when it comes to the models that translate the scenarios into the
quantitative outputs and I think the March disclosures still left enough
mystery there so that the banks can’t do that they may they might be able to do a
reasonably good job of that now in some of the credit portfolios for a credit
asset classes for which model portfolio results were shown that may get pretty
close but there’s a lot of seek are outside of the loan books we talked
about PPN are there’s again you know the treatment about other comprehensive
income deferred tax assets there’s operational losses all of these
components really aren’t addressed in the kind of detail you’d need to be able
to build models that are gonna come quite close very close to the feds
models so to me the acid test is can the banks do that and I think the answer is
still no at least not for many of the core components of C car so so let me
I’d like to actually put the March report in the context of a whole bunch
of other initiatives that the Fed has done in the last year and a half or two
years that I think are quite remarkable that that
been the direction of further disclosure so the the 3 February transparency
notice documents the March the both the nineteen the 2018-2019 results of the
seekers and he fast dress testing frankly the financial stability reports
in the supervision of regulation reports have all kinds of information that has
been secrets at the temple for too long I think this is all really good and so I
don’t want to sound ungrateful is I think these are all really great things
in the and the March report included a lot of useful information but I still
think I don’t see the justification for continuing to hide the ball I’m not
saying that there aren’t justifications maybe someone can come up with but I
haven’t heard one other than these big things about gaming or some of the other
ones we talked about today that I just don’t find persuasive so I’m open to
hearing a persuasive one but I don’t think it is unless there is one that
someone comes up with I think that full transparency is what the Fed should do
so my view is slightly different than the other two but I guess I would at
this so I think it’s a it’s a it’s an important question it’s a difficult
question frankly this question of how much to disclose and I understand I’m
certainly not going to get into a Administrative Procedure Act debate with
the lawyer very fine lawyer sitting next to me at this point I think actually
where the Fed needs to focus on for the continuing dynamism is frankly in the
scenarios but I do that but I do because I think I think that I think it’s passed
the slippery slope personally and II keep trying to convince me that they
haven’t reversed engineered models we had this earlier on another conversation
III think that the that some of that mystery is lost I think that’s
unfortunate but because I think dynamism is very important so I guess I was again
say I would focus on the scenarios and having as much dynamism the snares as
possible I’m not sure they’re holding out that much left on the models didn’t
matter that much at this point but let me add to say too that if anything about
disclosing the models slows down the process of the Fed upgrading updating
and evolving those models when they discover that they may be
missing something I think we have a real problem and I think that’s where some of
the concerns about this come in and this gets back to the GSE issue where you can
portrayed it in the paper but but the Fed should have the capacity to identify
ways in which it can Hance its measures and it should be able to use those how
much rapidly as possible because there’s no margin for error with these firms
I’ll get back to the objective of making sure that they don’t create another
crisis so yeah I agree entirely that that the issue is do you want the firm’s
to have the models or not and I think that what happened in March was sort of
a halfway step and just bound not to make everybody happy the notion that
that we could have a comment a notice and comment period with say a 30-day
comment period you know Randy’s firm writes a lot of those comments and there
are people inside the regulatory agencies who read them and then try to
figure out what to do about them so the delay even if the comment period is 30
days the delay is substantially longer and I think that the the notion that we
give away too much of the model information creates the potential for a
huge number of arguments about whether the capital ratios are too high or too
low on individual securities which I think is ultimately not part of the
stress test but I think it would it would bog down to perhaps bog down the
stress test dynamism just one follow up on that I just want to make it clear I
am NOT arguing for you know subjecting the models to the necessary to the
Administrative Procedure Act it’s more the spirit of the Administrative
Procedure Act of public notice and comment it can be shorter if there’s an
emergency that means the Fed should move quickly you know there could be an
exception to that period it could be that the notice come period goes on
beyond a particular cycle so that it applies to future cycles but I think the
real thing I don’t think it’s the I don’t think it’s a public notice and
comment process if implemented correctly you know that’s gonna eliminate dynamism
in the fence models that’s a question of the feds backbone and I am completely
convinced that the Fed has backbone I’m not convinced it’ll fail had backbone
but as long as the regular actually has backbone and believes in the models
believes in those stress testing process which I think the Fed clearly does and
should then I don’t see why an appropriate input from the public has to
be some I don’t see that it necessarily you know interferes with dynamism
quick quick quick response than that so all actual points I guess one of the
things that I I met one of the 40% of the pieces of my 40 percent I didn’t get
to up there and maybe that’s even maybe was more than 40 percent is that you
know I don’t know if it’s a backbone issue per se it seems when I was talking
about the firms and what they’ve been wanting for all these years it and and I
think it’s safe to say in so far nodding I think it really comes down to actually
stability there’s lack of dynamism they want to turn the feds in your mouth
there is some desire I think to see the stress test turned into a very
complicated type of fixed risk weighted assets maybe a better risk weighted
assets not change very often and I worry about that and I think that since that
is part of what’s propelling some of the changes whether the Fed has the backbone
or not I believe the Fed has a lot of backbone I think you might get into a
cycle where they would they would find it very difficult to actually make
changes that led to significant volatility in the post rest capital
requirements and that could lead to again a desire not to move quickly to to
make changes to the processes or the scenarios that’s really one of my true
concerns about all of this I think they need to move quickly
at all times so I’m just going to ask Tim or others can you not get the
dynamism that you’re looking for from the scenarios and varying varying the
shocks and the scenarios again within a defined perimeter it seems to me that
that’s the place to look for dynamism and responsiveness and if you want to
test new vulnerabilities let’s do it there but why do we need the uncertainty
in the models that convert those stress inputs to outputs that I’m not you know
I’d like to get your perspectives on why you need the uncertainty there as
opposed to putting the dynamism in the first part of the equation yeah I I you
know when at one of the arguments that I tried to make which which maybe it
didn’t come through is that the the range of stresses that can be
implemented through the current stress scenario is very limited why because
because the models inside the the black box all look at interest rates and
unemployment rate and various components of real output but it’s a very small
number of variables and we know that historically they’ve all moved together
so if you’re going to confine yourself to those sorts of external stresses I
don’t know I don’t know I’m sure people can be more imaginative but I don’t know
how dynamic we can get if we confine ourselves to those sorts of stresses so
I I think that changing some of the model parameters is going to be an
important part of exploring beyond the kind of deep recession shock that we’ve
we’ve administered repeatedly all right why don’t we turn it open to the
audience I assume there’s microphones and if you have a question please
identify yourself and your affiliation so there’s one right down here in the
front hi I’m Marcus Stanley from Americans
financial reform and I guess it’s it’s always very interesting to hear people
who who want the stress tests to be more stable or frankly more static talk about
dynamism or frankly give lip service to dynamism I mean I mean the stress tests
are already fairly static over the last four years if you look over the last
four years the total loan losses or the the estimated loan portfolio loan losses
have been right around six percent every year the estimated trading losses from
the counterparty and in trading shock have been ninety to a hundred and ten
billion dollars every year and that’s pretty much four four four years in a
row so we’re already looking at a fairly stodgy stress test I guess and I just
wanted also to underline what Mark said about that you you can’t get dynamism
from the scenarios alone because the question is are these
scenarios going to be translated into losses in a way that we have not seen
before because that’s that’s what happens in a financial crisis is that
macro that macro the macro situation is translated into losses in a way that
people have not previously predicted and as Mark said that that’s got to come
from the the parameters and the model so I guess just to turn this into a
question to Andrew I was gonna ask this question but Tim kind of then stated it
you said very clearly several times that banks need to know in advance the
capital waits for their assets if the banks do know in advance the capital
waits for their assets then isn’t this just another form of risk weighted
capital haven’t we just turn it into risk weighted capital yes so I don’t
think that knowing the the weights that come out of the models makes this an rwa
exercise because they’re really they’re they’re two components here the
is the scenarios themselves and then there’s the translation and the scenario
a despite you know your your review of the record which says that it’s
relatively static stereos actually do their we quite a lot a year-on-year and
they test different vulnerabilities that’s certainly true in the global
market shocks where you’ve seen tremendous variation from year to year
and even if you look at things like the shape of the yield curve in 2018 versus
2017 and 2019 substantial differences and when I’ve looked at the last three
years the seek our losses as a percent of CT
one common equity tier 1 capital in 2017 I think averaged 28 percent on the jump
to 36 percent in sorry there’s in 2017 jump to 3 6 percent in 2018 and back
down to 27 percent in 2019 that’s a pretty big degree of variation and that
all came presume that came mostly from scenario design not from changes in the
models so I think you can get big differences on the scenario side so I
don’t think it’s an rwa in disguise Thank You Dennis Keller from better
markets Randy I have to tell you I tear it up at your advocacy of full
transparency and public interest in fact you know we’ve got openings at better
markets a couple of comments one is the apparent use of the public input as a
synonym for the industry should be thought about and the APA yes there’s
supposed to be public input but on stress tests like derivatives and others
it is 99.9% industry input and we saw how the industry used the APA process to
slow down kill rule after rule after rule and then when they didn’t get what
they want then litigate it so to say that the eight you could use an APA or
an APA like process in a relatively quick way when you’re opposing the
industry is just factually incorrect and the record is replete with example
of that um but in taking the prioritization of transparency to the
extent it seems like some of you are saying I assume that means two things
one is all the banks agree will agree to disclose their proprietary models and
inputs after all the public needs a holistic picture of what’s going on and
if the public is going to hold the fed and elected and regulators accountable
shouldn’t they also have the ability to hold the banks accountable for what
they’re doing it after all that’s where the taxpayer money is going to go if
this process fails so seems to me that there has to be complete and total
transparencies the logical extension of your argument and secondly in terms of
transparency it’s interesting that the banks fought tooth and nail on any
transparency for the bailouts in o8 no.9 lawsuits had to be filed we had to put a
provision in dodd-frank mandating the Fed disclose all the backdoor bailouts
the front-door bailouts the rescues and the use of the rescue programs and so it
seems to me to say when we’re in the middle of the crash and taxpayer dollars
are being funneled to the industry and the Fed is putting trillions of dollars
into the industry don’t tell the public a thing and then fight for years after
that to make sure the public doesn’t know but to then say in peacetime or
whatever you want to call it in 2019 by golly we want a one-way street here on
transparency it seems to me that it’s a double standard that now the argument
for transparency benefits the industry and that’s why you want it but when it
doesn’t benefit the industry when they get into bailouts no transparency
anybody see a problem with that well it’s always easy to create a bunch
of strongmen and attribute arguments to people that people have never made but
know I do I do think that transparency should be the presumption for the Fed
and for all all organs of government unless there’s a justification you know
based on public harm that you know such as such as trying to prevent fostering a
financial panic or something like that and so I do think there needs to be poet
transparency I think it’s not the same when you’re talking about private you
know businesses versus the public that’s not part of the government on the other
hand I think there’s probably a halfway point in terms of transparency you’re
talking about because it does strike me as reasonable for the Fed and they’ve
done this in other areas and send reports to disclose aggregate
information about what has come up from the the firm models or even to extent
they can do it without it being you know if they can do it on an anonymized basis
even ranges and averages and things like that if that can be done in a way that’s
consistent with I think the you know private competitive interests of the
banks involved yeah it strikes me that that might be useful information okay I
make a ripken I make a really quick point we move on just to the very first
part of it I think Dennis that’s raised a really important point least important
to me the the engagement with the public in the notice of public comment is it
just a challenge in this particular area and in many areas because the
complexities of the issues involved I want to actually thank Dennis and Marcus
and you know AFR and better markets for the work they do to try to make to get
people to understand and put that in there and I guess I would encourage the
Fed to find ways to get more engagement with the public to get more input from
them in this process I don’t know exactly what the answer to this is but I
think it’s very important and I’m glad Marcus raised it because frankly the
public in this in this business usually is the banks and a few scattered groups
trying to take the other side and the academics
anyway so anyway III would like to I would just like to see the Fed and
hopefully you know greater public engagement on these kind of topics and
this was a great start that again for mostly bankers and specialists here Francisco Kovas from the Ming Policy
Institute and this question is for mark when you mention about the social costs
of that so meaning there’s a huge literature right now documenting
migration of riskier lending out of banks to non banks and I think there’s
still the jury out there in terms of whether the quantities are being
impacted but clearly the pricing is being
impacted and also clearly you know the trends have been significant and easy
solutions are unregulated and we really don’t understand the risks they’re
taking for many of those cases and this is both on the corporate as well as on
the household side so when you say it’s there’s no social costs meaning I think
we ought to think deeply about these these trends and and you know what is
the impact of stressed I’m not saying everything is driven by
the stress test was clearing the stress test have a very important in fact this
is the question for you I say another remark is and I think we’ve seen looking
at the last six or seven years of stress testing you know the results have been
the Snider’s continue to be extremely severe and the peak to trough decline in
the in losses has been lower and of course you know we we a bank analyst
meaning we spend a lot of time trying to think how much is driven by the severity
of the scenarios changing as well as the Fed models changing as well as the
portfolio’s of the bank’s changing I mean I think it would be a tremendous
help for the industry and everyone to understand better you know as we’ve
looked over the last six seven years meaning how much of the stress test made
the bank’s portfolio so pristine and how much has migrated out of the banking
sector so another way just decomposing the changes every year they call it the
SCB the peak to trough decline in capital ratios how much is driven by
this trick be components don’t be surprised that Bank models play a huge
role and I think that’s really a key issue because you know that just insert
inserts a lot of volatility in banks capital requirements and as was
mentioned today meaning we know from economics that volatility leads to under
investment and so I think that’s a very important
issue well thank you thank you for those thoughts you know I think I have been
saying for a long time that one of the reasons I don’t like
higher capital ratios in banks is because it will drive business elsewhere
and I don’t know where the business and the risk is gonna need to go I think we
have very bad information about where the business goes on the other hand if
if what we find is that that the business that stays inside the banks
after the stress test rearranges risk allocation I believe it ought to be the
stuff that’s riskier I think the safe stuff is easier to securitize and push
somewhere else and so I think what you’re describing in a sense is is a
symptom of the rearrangement of risk bearing in response to changing
financing costs so but I agree with you entirely that that the the pushing the
business out of the banking system where we know it sort of works into someplace
else where we’re not sure it works has risks involved and I’ve always counted
that as one of the costs of higher capital ratios and to that a good again
so I would coming back to Dennis because actually death had anticipated number of
tremendous points in his earlier comments the fact that it’s not
regulated well outside of the banking sector does not mean it shouldn’t be
regulated well inside the banking sector right so I think I I understand the
concern about systemic risk propagation I think the answer to that is actually
greater oversight and and you know work understanding where it’s going and and
so you know it’s hard to when you see administration with recommendations a
couple weakening the ofr and the F sock with with these kinds of things I thought Andy your comment was
interesting rather operate in a system with higher capital requirements that
are predictable than have the kind of unpredictability that you talked about
with regard to the stress tests on just be very interested in hearing a sense of
what the magnitude of that trade-off is both from you and I’d also want to hear
from from Tim and from mark as you think about what that would look like in terms
of safety soundness and financial stability you know what would be that
kind of ire through the cycle capital well here I’m going to retreat to Mark’s
comment that maybe is also a metric that’s devilish devilishly hard to come
up with a cost-benefit analysis of how much capital is enough so I don’t have
the answer to your magnitude question but in all seriousness I do think that
if the reason we’re keeping uncertainty in this system is because we want an
uncertainty buffer and we really want thanks to just hold more capital then we
should just do that explicitly and we should be clear about it and have a
system of known rules and allow banks to operate then within that system so if we
think the current minimums are too low let’s raise them if you have 4.5% is too
low and we think 5% is better let’s have that debate and let’s raise them and
let’s move on but I think to keep it in this state where you don’t know its
volatility its uncertainty and you don’t know how much the buffer how to size
that buffer throughout the annual capital planning cycle I think really
leads to all the all the X and efficiency I was talking
about beforehand so I would much prefer a system where we had that debate open
and explicitly think we have time for one more question well I think I think
that carefully looking at the past nine years of stress tests we could get an
idea of how much excess capital the banks are holding in order to protect
themselves against this uncertainty and I think on average if Andy were average
which of course he’s not he’s much better than average but but if Andy were
average he’d take that as an estimate of how much extra across-the-board capital
he’d be willing to take so I think we could get an answer to that question by
being careful and subtle in evaluating the the misses in the in the past
capital start and I’d suspect that’s true we could get an answer to it I
think that if we did get an answer to it it would probably you know have to
include a lot of type of analysis currently carry that in the stress
testing program to figure out what the right number is and if possible maybe
increase the confidence we take away from the post Jess minimum right now I’m
going to say the post stress minimum should be higher factor that all into a
large calculation with the through the cycle number I think I don’t think the
industry would be particularly happy with it to be honest but we could give
it a try I wanted to just respect real quickly on the uncertainty question I
struggle with this one because you know I know you don’t want the feds read
capital requirements to become a new source of uncertainty but the world do
you operate in is tremendously uncertain it’s almost as if you’re asking the Fed
to create a artificial uncertainty by saying this is the number it has to be
and it shouldn’t vary or maybe I’m wrong maybe maybe I’m conflating a little bit
but but I want to be I want to understand a little bit better I turned that
exactly around him so I think we do live in uncertain world as a risk
practitioner experience that every day and my my please let’s not add an
artificial source of uncertainty which is uncertainty about the feds models
that’s an entirely invented manufactured sources of uncertainty why do we have to
deal with that in our capital planning exercise we all know the world as in
certain there’s nothing certainty out there that’s my point
right but you we could go back we could go back and forth for a while adjusting
the models would appropriate you would agree with there right if the Fed
identifies a problem with the model in it okay so it’s not static models it’s
just more transparency on the model well I think looking at the clock since
the next the next agenda is is lunch we will call it there thank you very much
to mark in to our panelists for a wonderful you

Leave a Reply

Your email address will not be published. Required fields are marked *